Adobe Firefly for Content Creation: Master AI Illustration, Digital Art & Graphic Design

1. Welcome to AI Art Content Creation with Adobe Firefly!: Hello, and welcome to the course on Adobe Far fly a generative AI application for effortlessly creating and justin images and design elements through simple text instructions or prompts. My name is ana, and I'll be instructor and mentor for the course. I was the director of product at a series BAI start up based in Singapore, where I relocated six years back to do an MBA program. I've always been curious about the field of AI and how machines learn in ways similar to us humans. So I kept a close eye on the development and innovation in the field. Firefly was the tool that immediately caught my attention, thanks to its straightforward accessible and simple to navigate interface. By joining the course, you will get access to over 5 hours of H Dvideo content, step by step tutorials and activities. A handbook with over 100 examples of how firefly style effects and your prime description could influence the final output and much, much more. We will begin by covering the key modules of firefly, including creating a for the realistic image or illustration from scratch, And with the reference image, we will also cover advanced techniques for designing and improving your prompts with tools such as CHA GPT, Microsoft copilot and Adobes stock. And if you are curious about going deeper into AI technologies, there will be optional lectures where you will learn in detail how AI creates art. This course assumes absolutely no prior experience in AI, I generation or design. So if you are just starting your creative journey with AI, I will be there to support you all the way through the process. So let's begin the course by going through an overview of f fly and some of its common use cases. I'll see you in the next video. 2. Introducing Firefly and its main use cases: Hello, everyone, and welcome to the first lecture of the course. Before we dive into the detailed tutorials on Firefly, let's do a quick run through of its core features, as well as some examples of how it can be used to enhance your work process. Let's get started. Firefly is a generative AI application from Adobe. For effortlessly creating and adjusting images and design elements through simple text instructions or prompts. I'm sure that many of you have probably heard of the term generative AI before. But let's formally define what it means. Generative AI is an umbrella term that includes various techniques focused on creating new original content that never existed before, like images, text music that mimic or inspired by real world example. Firefly is a standalone web application accessible at firefly dot.com and is available in over 100 languages. The core functionality of firefly web app includes the following modules, text to image, and generative feel. Let's briefly cover the capabilities of each of the modules. Text to image, one of the most advanced firefly functions allows you to create foreliistic images or illustrations from scratch or with the reference image of your own or the one available in the firefly gallery generative feel. With this module, you can modify images by adding removing or replacing elements or backgrounds using text proms. You can also stylize your existing image with digital effects as well as do for the restoration. These are features already available in firefly, and we will go through each of them in much more detail in the upcoming lectures of the course. However, the AB team is also working on introducing and exploring new functionality such as three D to image module that will create a three D scene and use a text prompt to generate an image. According to the firefly website, this feature is coming soon. Other two features are in exploration at the moment, including sketch to image that will allow turning simple drawings into full color images and personalized results to generate images based on your own object or style. Firefly doesn't provide a timeline when these new features will become live. But I'm monitoring the product updates, and I'll include new sections in the course as soon as functionality is available. In addition to the Firefly web app, Adobe also has the broader family of creative generative AI models, along with features powered by Firefly in Adobe's flagship applications, including Adobe Express, a tool for fast and easy content creation. Photo shop poder editing software, Illustrator, a vector graphics software and Adobe stock, a service that offers a vast collection of royalty free assets like images and videos. We will speak more about how to use generative AI capabilities in some of these products in the upcoming sections of the course. All right, o. Now that we've covered the key modules of firefly and what product features are around the corner, Let's talk about some of the examples of how Firefly can be used content creation. You can create for the realistic images, illustrations or videos for articles, blogs, social media posts, marketing materials, and so on. That is how I mostly use firefly for my work. For example, I used firefly to create illustrations of complex concepts and ideas in the course lectures about AI. To design the longer for this course and to craft course resource files. Another use case for firefly is graphic design. For instance, you can generate design concepts and assets such as icons, buttons, colorful texture patterns, mops, for your website redesign, product mops, and prototypes. Much more advertising campaign visuals. Use Adobe Firefly to produce creative visuals for advertising campaigns, allowing marketing teams to test different concepts and themes quickly and without the need for extensive photo shoots or graphic design work. You can also use Firefly for brainstorming, exploration, and inspiration. For example, let's say your team is brainstorming concepts for a new marketing campaign themed around sustainability by including keywords related to sustainability into the prompts, they can generate a variety of images and spark new ideas and discussions on how to approach the campaign creatively. Another premis used for firefly is in filmmaking. It can serve filmmakers by elevating the production values of the projects, making films appear as though they were crafted with a substantial budget, even when resources are limited, and of course, firefly can be used for numerous creative personal projects like designing the color and book, creating unique wallpaper designs, crafting home decor or creating one of a kind T shirts. Of course, these are just some of the use cases for firefly to give you an idea of what the product is capable of. If you already have some ideas on how you can work with firefly as your copilot and how it can help you with your work, please share them in the QNA section of the video. And that's it for the lecture. Let's briefly recap what we've just learned. Adobe Firefly is a generative AI application for creating and adjusting images and design elements through text prompts. Generative AI includes various techniques focused on creating new original content that mimics or inspired by real world examples. The core functionalities of Firefly include text to image, which creates images from text froms, generative field, which modifies images by adding, removing, or replacing elements. Firefly is part of the broader firefly family of creative generative AI models. With features integrated into AdobEpress Photoshop, Illustrator and Adobtok. Examples of firefly applications include content creation for articles, blogs or social media, graphic design for generating design concepts and assets, brainstorming and inspiration for marketing campaigns or creative projects, filmmaking, as well as personal projects. Okay, cool, I'll meet you in the next video. 3. Account creation and setup. Generative credits. Commercial use: Everyone, welcome back. First Ts first. Let's make sure you can access Firefly. Apart from getting your account radio, we will also talk about terms such as generative credits and why you need to keep track of them to use the Firefly web app. And the generative AI capabilities available in other Adobe products. Last but not least, we will also discuss whether images generated in firefly are safe for commercial use. This lecture is important. Please pay attention. First, let's cover what generative credits are. Adobe generative credits are a system introduced by Adobe to manage and limit the use of its AI powered features, such as text to image, generative fiel, text effects, and others. So each time you click Generate or refresh to create an image from a text front, it costs you a credit. This method is part of a wider trend where companies control how much you can use AI features to ensure they don't use too much computer power. Generative credits allow Adobe to offer everyone access to these advanced AI features in a way that matches their subscription. Ensure the technology is used wisely, and keeps working well for everyone. The information on how many credits are available in each subscription plan is subject to change. I recommend you check Adobe's generative credits FAQ page for the latest updates on how many credits come with every plan. Please note that this allocation of generative credits is designed to reset monsm I leave a link to the FAQ page in the resources section of the video. Now, you may have a very logical question. What will happen if you spend all your generative credits in the given months? Well, that depends on your subscription plan. For some of them, such as creative cloud. You can keep taking generative AI actions, but the use of the generative AI features may be slower. Other plans may require you to purchase additional credits to get uninterrupted access to generative AI capabilities. More up to date details on this topic are also available on the generative credits FAQ page. Don't forget to check it out. All right, cool. Now that you understand what generative credits are. Let's talk about how to get your Adobe count dim. For this course, you will need access to the far fly web app. And if you plan to file along with the course projects, you will need access to Adobe Express and Photoshop, as well as premier if you don't have access to any other video editing software. The best option to get access to all of these products is by signing up for the Creative Cloud all apps plan that gives you access to Firefly, Express, photo shop, premiere, and more than 20 other apps. If you already have access to this plan, then you are all set for this course. If you don't have access to Creative Cloud all apps, you can start by signing up for the free plan that gives you a certain number of generative credits. Later on, you can upgrade to a paid plan to continue creating assets with features powered by Firefly or wait until credits reset the next month. Adobe also has a subscription plan that gives you access to a single app Firefly. You can also choose to sign up for the trial version first, and then upgrade to premium. I don't recommend this option, since when you've reached the limit of your generative monthly credits, you will be able to do just two generative AI actions per day to create images or vector graphics until your credits reset the next month. The creative cloud all app subscription gives you more flexibility as even when you've spent all your generative credits this month, You can still keep working with generative EI features, but at a slower pace. However, the final decision on what subscription plan to choose is yours. I leave the links to the sign up forms for both plans in the resources section of the video. Once you've signed up for an account, you can always check your generative credit account. For this, go to your account details page available at account dot.com. And click your profile icon. On the next screen, there will be a counter that shows the number of monthly generative credits allocated to your account and the number of generative credits you have used in this cycle. All right. Cool. We have one remaining topic before finishing this lecture. Let's figure out if images that you create with Firefly are safe for commercial use. So the current firefly generative a model is trained on a dataset from Adobe stock. As mentioned earlier, this is a service that provides designers and businesses with access to millions of high quality curated and royalty free pots, videos, illustrations, graphics, three D assets, and templates for all of their projects. Apart from adobo stock database, the firefly model is also trained on openly licensed content and public domain content where the copyright has expired. All the ensures that images generated through firefly are safe for commercial use, meaning that they can be used for such business purposes as creating marketing materials, product designs, selling digital content, or any profit oriented business applications. That's it for the lecture. We just covered everything you need to know and to do before creating your first image with firefly. Let's briefly recap the main points generative credits are a system introduced by a Dobm to manage and limit the use of its AI powered features. Credits are consumed with each generate or refresh action and are designed to reset monthly. What you can do after you spend all the generative credits depends on your subscription plan. Some plans may allow continued usage at a slower pace or require purchasing additional credits to access the far fly web app, and other products we'll be using in the course. Consider subscribing to a trial version of the Creative Cloud all apps plan and upgrading it to a paid plan later on. Finally, we said that the images generated through firefly are safe for commercial use as the model is trained on the Adobe stock dataset, openly licensed content, and public domain content. Please don't forget to set up your account before joining me in the next lecture where we are going to create your first artwork. I'll see you there. 4. Introducing Firefly Text-to-Image functionality: Everybody, and welcome to the new section of the course. You know what they say. A picture is worth 1,000 words. But in this section, you will learn how to create 1,000 pictures with just a few words, which is now possible with the firefly text to image functionality. We'll kick off this section with creating your first piece of art. I will guide you through the process and explain how to navigate through firefly toolbar and improve your image in seconds. We'll also explore how to apply style effects, as well as reference images to your text descriptions to influence the outcome. Whether you're aiming for a realistic photo or a digital illustration. Understanding how to instruct fireflies AI is crucial to ensure your creative output matches your original vision. In each of the tutorials, I'll use several examples to illustrate each step of the process. So you'll see firsthand how we define your descriptions and styles can prides use vastly different results. Towards the end of the section, we will speak about current I limitations in the image creation space, and most importantly, how you can help the technology to get better. Of course, I'll share my recommendations on how to come up with the most effective prompts or instructions that convey an idea of what you want to see, but leave plenty of room for AI creativity and imagination. So we have a lot to cover, and that's enough for the intro. Let's begin. I'll see you in the next video. 5. Step-by-step guide to creating your first artwork: Everyone, and welcome back. So let's go and generate our first picture with Adobe fireflies, text to image functionality. As the name suggests, this tool allows you to transform your ideas into visual art simply by describing them in words. Before we start the dema, let me warn you that the firefly interface you are about to see may change very fast as the product is very new and the firefly team is actively working on improving and expanding it. Of course, I'll do my best to keep the tutorials up to date. But nevertheless, don't be surprised or frustrated if you come across some discrepancies between the videos and the current interface. All right, let's generate our first picture. Okay. So as the first step, open your web browser and navigate to Adobes fireflies website at firefly.com at the firefly website. You first sign into your account. And then click on text to image to begin image generation. You will see here a gallery of images you can browse through and use as inspiration for your own image. You can click on the image that you like and see the prompt that was used to generate it. You can also make modifications and generate a new image based on the new prompt. Alternatively, you can start generating an image from scratch by typing in your prompt into the prompt field. Let's do this for this lesson. I'll start with cute little robot and press generate. Here are the first results that we've got, which are pretty decent. Click on the image to expand it. From here, you can browse through all generated images to choose the one you like the best. Let's explore some settings that were applied to generate the images. First of all, we see which model has been used to generate the image. The latest model is selected by default. Next, we see that we can choose the size of the image. Square size is selected by default, and you can choose from three other variations. For this work, I'd like to set the image size to white screen. You can always crop the image later if you need to. The next setting is content type. Here you can choose if you'd like your image to closely resemble a photo. In that case, you select a photo. Alternatively, choose art. In case you'd like to create an illustration. If do mode is switched on, firefly will make the choice for you based on the prompt. Switch it off. If you'd like to make the selection manually. Let's change the content type to photo and see what results we'll get. Very nice. For this particular work, I'd like to create an illustration, so I'll switch back to art. Let's add more details to our prompt. The great thing about prompting in Firefly is that you can get prompt suggestions that you can take as inspiration for your image. You can click refresh to get new suggestions. In case you don't want to see any suggestions from firefly, you can always switch them off here. I'll add wearing vintage sunglasses and click Generate. Nice pictures. This one is pretty cute. Let's add some additional details to the prompt, standing in front of a futuristic building. Okay. Here is what we've got. In case you like a specific image, you can ask Firefly to generate images similar to it. To do so, click on edit on the top left side of the screen and select show similar. Firefly keeps the sample image and generates several more images, similar to the original one. Personally, I find that the original image is always better than the additional images generated by firefly. But of course, feel free to experiment with the show similar feature yourself. You can continue including more details into your prompt to get the final result. An additional option for how you can tweak your images is by moving the visual intensity slider. For content type set as art, like in my example, you can transform them from digital art to something more illustrative by moving the slider to the right hand side. You see that these images look more stylized than those we've generated with the visual intensity set in the middle. For content type set as photo, you can transform them from being more realistic to surrealistic. Let's see how it works. I'll change the content type to ph and generate. I'll save a link to this image and open it in a separate tab. Now let's move the slider to the left, reducing the visual intensity to a minimum. Let's open this image in a separate tab as well. Now, let's increase the intensity to its maximum. Now let's compare our results. I feel like I prefer the images with medium visual intensity. But of course, that's my personal preference. Please share in the Q&A section of the video, which of the three outputs you prefer. As you can see, I've started with the simple prompt describing the subject or what I want to see in the image and then expanded it by adding extra details about the character and its location. I use simple language as I would describe the picture to a friend, without including technical terms and abbreviations. As we've seen from the demo, you can always rely on prompt suggestions from firefly in case you're out of ideas and need inspiration. Finally, I changed the visual intensity of the image to see how it would influence the final output. We will discuss more on how to craft effective proms in the upcoming videos. For now, let's come back to our demo. If you're happy with an image, click on the hard icon to save it to your favorites. You can access the saved image later from the Firefly home page by clicking on the favorite stop. At the moment when I record this tutorial, Firefly doesn't have a history of generated images that you can browse and come back to for further editing. So if you like an image and don't want to lose your work, save it to your favorites to revisit it later. We will talk about other options for saving your image in the upcoming video. Okay, that's how you can generate your first image with Firefly. Now it's your turn to create your first artwork. Please also share the links to your images with other students in the Q&A section for the video and LCA in the next tutorials, where we will continue exploring Firefly text to image feature and see how we can influence our image generation even more with firefly style effects. See you in the next video. 6. 31 How to save your image: Hello, everyone. Before we start discussing how we can further influence our image generation by applying different style effects, let's make a quick pause and discuss how to save your image. This topic is important as unlike other art generation software, images are not saved automatically in firefly. You'll have to take action to save them. Otherwise, you will lose your work. So there are several ways to do this. The first option that we've already learned in the previous lecture is to save the image to your favorites. This method allows you to go back to your work and see the generated images together with the prompts and styles that created favorites are saved in your browser cache, which means that they are only available in that browser on that computer and can be lost if you delete local browser data. To save an image to your favorites. Click on the Star icon located at the top right side of your image. You can view any images saved to favorites through the main firefly page. Let's explore other options for saving your image. The download option allows you to save the image to your local computer. Here is the warning we receive when downloading the image. It states that content credentials will be applied to the image. Content credentials consists of non personally identifiable information. Attached to the image generated with firefly, accompanying the content wherever it goes in the context of generative AI tools like firefly content credentials include indications that generative AI was used in the creative process. This is crucial for ensuring transparency around the use of AI in content creation. It is part of the broader content authenticity initiative, which aims to establish a global standard for sharing the origins and details of digital content across platforms, extending beyond Adobe products to verify the authenticity of content through the content authenticity initiative. You can click on G to verify. This service enables users to inspect their content credentials associated with the piece of content. By uploading or referencing digital content on the verify platform. You can view the content credentials if present and assess the contents history and authenticity. Now, let's return to Firefly and continue exploring options for saving the image. As we were discussing how to download the image, I click on Continue here. Firefly includes your prompt description in the file's name. Allowing you to always refer back to the prompt used to generate the image. Let's explore additional options for saving and sharing our images. I'll click on share. From here, you can copy a URL link to an image if you'd like to share your project with someone else. For my workflow, I also like to paste these links into a separate file. For example, in Google Dogs so that I can access my projects anytime without relying on files saved in my browser cache. Like in case when I save an image to my favorites folder, I strongly recommend this approach to ensure that you don't lose your work. Save to library allows you to save a copy of the image to your creative cloud library and access it in other Adobe products when required. Submit to community. Lets you share the image on the community gallery. You can access the community tab from the far fly home page. While we are here in share, let's explore the remaining option open in Adobe express. By selecting this option, you can dit the image in express a tool for fast and easy creation of content such as loggers, banners, resumes, and more. We will explore how to edit an image generated with firefly in express later in the course. All right, we are done with the quick overview of different methods to save your work in firefly, and I'll see you in the next video. 7. Applying styles to influence image generation (part 1): Everyone, welcome back. Now that you know the basics of how to generate an image based on a text prompt, let's get into more details of the prompt to image functionality and explore how you can influence the image even further by applying different style effects. Style effects in Adobe firefly are not just filters or overlays that modify an already generated image. These are settings that influence I algorithms, guiding them to produce images that reflect the chosen style and modify the aesthetics of your images. This can range from mimic and traditional art styles like impressionism or surrealism to apply modern digital effects. And the best part is that you don't need to be an art gurau or have a degree in fine arts. With the user friendly, easy to use interface of firefly, You can preview different styles and try them out for your image. In my view, fireflies attempt to simplify the process of selecting different style effects, themes, techniques, and poder compositions is something that sets them apart from the competition and is one of their core value propositions. As a non artist, I was intimidated by the fact that I have to know so many nuances about artistic styles, colors, or composition to craft a good prompt. Frankly, Firefly was the first tool I stuck to for a long time and now use actively for my projects since it did such a good job, simplifying the whole process of creating artwork. But enough of the intro. Let's jump straight to Firefly to see how style effects work. To apply style effects in Firefly, you can start by creating an image from scratch, or selecting one you like from the Firefly gallery. Let's create an image from scratch. I'll write futuristic living room with cozy interior, large windows, and armchair and click Generate. And here are my results. Next, let's navigate to the effects panel where you can choose from a variety of styles. Here you can select some of the popular effects or navigate through different types that include artistic movements, themes, techniques, effects, materials, and concepts. Let's start with movements and select Baroque, then press generate. The generated image will be a combination of the prompt we entered, and the added style settings. As you can see, the baroque setting drastically change the image from the minimalist type interior that we got from the first generation. Apart from selecting a style effect, you can also modify the intensity of this effect by drasting the style strength slider. Let's move the slider to the left, reducing the baroque style effect to a minimum and click Generate. This style looks more modern and futuristic to me, but we still see some broque style elements like the golden decor on a chair and walls. Let's continue our experiments and move the slider all the way to the right to maximize the styles influence on the image. The latest image has more baroque golden style elements and accents, especially seen in this image. We can continue experimenting and moving the strength slider to find the image that represents our idea best. We can also combine different styles together. For example, let's explore themes and from here, select interior design to see if this setting will enhance the image. Here are the results. This one looks quite realistic and interesting. I'll keep it in favorites for future reference. Apart from the style effect settings, we can also set such things as color and tone, lighting, as well as composition. Let me choose golden hour from the lighting top and click on generate. Now, the image is lead with warm sunlight, add in a more cozy, relaxed atmosphere to our image. I'm happy with this picture. I'll download it so that I can use it in my work. In case you are not happy with the results you see, you can start applying stiles to your prompt from scratch. For this, click on clear styes adjust the prompt if you like, and go ahead with selecting the styles. You want to try this time. Let's try different prompt for a change. I'll write a pair of fashionable sneakers and headphones next to an iPad. Decent results, but not super great. Here is what you can do to quickly improve the image. Go to themes and select product so and click Generate. Now we've got results that are crisper. The colors seem brighter and overall, this shot looks more professional than the first one. In case we want to experiment with the brightness of the image and boost it further, we can choose vibrant colors from the color and tone section. Looks nice. To give the image an even more polished professional look. Let's select studio lighting from the lighting section and click Generate. You see that our shot becomes more and more live and realistic. Notice the shadows and reflections from shoes in the iPad. Great work of the firefly model. Let's also just the composition of the shot. As we have a product ph, we can choose a close up or shot from above. I'll select close up. I like this image, so I'll save it to my gallery so that I can download it or continue editing it later on. Another thing you can do is to change camera control settings. This option is available only if you've selected photo as your content type. Aperture allows you to control the background blow in your image by setting a specific F stop. Shutter speed allows you to control the sharpeners of your image. Field of view allows you to use a specific camera lens in your generation to adjust the viewable area of your image. If photography is not your hobby profession, these settings may not be something you want to experiment with. So select order for firefly to predict the right for a settings that will guarantee great results. As you see, you can drastically alter the statics of your image by including style effects, changing color and ton, lighting, and composition of your image, as well as asting the strength of applied effects. After exploring and experimenting with firefly for some time, you will find your favorite style effects and the combinations that match your preferences and tastes. Until then, I recommend you play with the settings to really get hands on with them. Begin by applying one effect, see the result and either change the style or add another effect to see how the combination works. Don't try to add multiple effects in one goal until you understand how each of them influence the image. Also remember that you can adjust the strength of the styles to find the best results. To speed up your research and exploration of style effects, I run through all the style effects myself and tested how they work. I used the same prompt and then changed styles to see how it impacts the image. You can browse through the image gallery, study the images that I got for each effect, and decide which one you want to try for yourself. You can also click on every image to see the prompt and settings inside firefly and make modifications as you see fit. I have collected all of this in the course handbook that I will introduce in the next lessons of the course. That's it for this tutorial. We have a second part where I'll show you more examples of style effects applications. I'll see you there. 8. Applying styles to influence image generation (part 2): One. Welcome back to the second part of the tutorial where we explore how to modify your image by applying style effects. Now, I want to go even further with the demonstration of how powerful these techniques are and how drastically they can change your image. So I'll take a really short prompt and then apply several styles to it to see the differences between the new and original image. Let's return to Firefly to do this. Let's start with a very short prompt. Let's say Flamingo. Firefly tells me that my prompt is too short. Let's add a few more words, Flamingo and feathers and click Generate. We've got pretty decent results for the first generation. This image looks nice. Let me save it two favorites so that I can compare every new image to my original one. I'll also copy its link and open it in a separate tap for future reference. Let's change the content type to photos and apply the first style called synth wave. All right, we already got quite different results compared to our original image. I'll save this image to favorites. Let me also a link to the image and open it in a third type. I like this pink color, but I want to add more brightness and contrast to it. Let's try to add on. I always recommend adding one style after another and not all at once so that you see how each new style affects the image. Let's see what the results look like for these two styles. Very nice. We see that colors became very bright, creating an almost luminescent effect. Since the background is dark, it allows the neon colors to stand out even more, increasing the glowing effect of the birds and their surroundings. Let's save the picture to favorites and pilink to the image to open it in a separate top. Okay, since we're experimenting with styles, let's add another one, surrealism. By the way, in case you want to search for a specific effect by its name. Unfortunately, you cannot do so with the current version of fireflies interface. However, you can search for text on your web page by using the keyword shortcut Command F on a Mac or Control F on Windows. Before starting the search, don't forget to click on all under effects so that the search is done across all the styles available. I'll click Generate and here are the results. The style has changed, but not as much as when we applied the neon colors to it. The last image seems to have a slightly broader palette of colors with more intense contrasts between the colors of the flamingos and their background, and the flamingo in the foreground has a more pronounced orange hue on its wing. Amazing transformation. I like this image we've got after applying the Non style, what's your favorite? You can share your answer in the Q&A section for the video. Of course, we can continue playing with the picture and add more sties to it. But let's stop for now. As you just saw from the demo, you don't necessarily need to create a super detailed prompt to achieve stunning results. Instead, you can use a very simple prompt and then tweak the image generation with one or several styles. Now it's your turn to apply style effects to your images and see what results you get. Remember that you can always refer to the course handbook with the collection of images created with the same prompts, but using different styles. The file is available in the resources section of this tutorial, and I'll see you in the next one. 9. Match the style of an existing image with the Firefly reference image gallery: Everyone, and welcome back to the lecture. Now that you know how to apply effects to your image generation process. Let's talk about another way you can influence your AI art creation. Generative match. Generative match allows you to create an image based on the look and feel of a reference image, which you can upload from your own gallery, select in the Firefly reference image gallery, or generate in Firefly, and then use as a reference image for your next project. Let's see how it works. In Firefly, click on the text to image functionality. From here, you can create an image from scratch by typing your prompt into the prompt section. Alternatively, browse through the gallery and click on the image you like to access the project details. That's what I'll do. Next, let's go to the style Match section. From here, you can either select an image from the reference image gallery or upload your own reference image. Let's generate an image using the Firefly reference image gallery. In the gallery, you see a vast collection of reference images in different categories. This includes art mediums like acrylic and oil, watercolor, pencil, as well as different lights and effects, textures, and gimetric patterns. Let's create an image with one of the reference images. I'll type futuristic humanoid robot and click Generate. To see how the image looks before we apply any reference to it. This one looks especially nice. Let me save it to favorites, as we discussed earlier. If you like an image that was generated and want to keep it to come back to its editing later, don't forget to save it to your favorites folder. Now, let's try to apply the following style. Interesting results, completely different from our initial generation. We see a minimalist design with the mixture of human like forms and mechanical parts and joints. The monochromatic color scheme adds a clean and modern aesthetic to the image. Let's test how the image will change if we apply another reference image to the same prompt. This time, I'll choose this image and as always click Generate. Now we still have a robot since we haven't changed our prompt. However, its appearance is very different from the first one. The second robot has more visible mechanical and intricate details, and the color tone has shifted to a cooler blue. Contrasting with them in a chromatic color of the first image. You can also modify the strength of the style mage apply to image by moving this strength slider. Let's move it to the left, reducing the influence of the reference image and see what results we've got. Now, let's increase the style reference strengths to the maximum. Yes, the generative match feature is really cool. But what's the practical application of it? Well, there are plenty of use cases for this feature. For example, it is useful in maintaining your brand identity and aesthetics, including elements such as colors, shapes, and other visual elements. It can be used in product design, as we've just seen with the example of a humanoid robot. You can create concepts and prototypes for new products and quickly come up with visuals that express design tastes, color palettes, or the verle look of the product. Generative match can also be useful for content creation, as you can quickly produce visual elements such as pictures or drawings for articles, blogs or social media platforms, preserving similar content style themes or atmosphere. Of course, these are just some examples of how the generative match can be used. Please share in the Q&A section for this video, how you would use this feature for your work. As always, I've tested some of the reference images from the Firefly gallery and collected the results together with my original prompts in the course handbook that you can find in the resources section of the video. In the next video, we'll continue working with generative match and we'll get even more creative by designing your own reference image that we will use to influence our problems. Here in the next video. 10. Match the style of an existing image with your reference image (part 1): Everyone, and welcome back to the second lecture related to the Firefly generative match functionality. Apart from selecting a reference style image from the Firefly gallery which we tested in the previous lecture, you can also create an image in Firefly and use it as a style reference. Let's see how it works. Here is my current image that I've generated using the prompt bronze statue of a line in a seated pause with a solemn expression. I want to change the texture of the statue. For this, I'd like to create a reference image. Let's create the following prompt. Old marble texture, three D material sphere on a white background. Let's change the image size to square. My test shows that this size works best for the sphere image we want to create. If you select a bigger size wide screen, for example, you could get two spheres, and that's not what we'd like to have. Let's click Generate. All results look great. Now, let me select this image as a style reference. I select edit and click Use as a style reference. Let's write down our previous prompt, bronze statue of a line in a seated pose with a solemn expression. Let me also change the image size to portrait. And I'll click Generate. You see how the texture of the material the statues made from was altered based on the reference image. That's really incredible. I like this picture, so I'll save it to favorites. You can create different reference style images with Firefly and apply them to your picture. The Dema, I've generated several images with different textures, and I now want to use them for new image generation. To upload the reference image, go to the style section and choose upload your image. Select an image that you'd like to use as a reference. We see a warning that you must have the rights to use any third party images and that your load history will be stored as thumbnails. Please also note that according to Adobes documentation, the image you upload to generative image will not be used to train the firefly model. You'll see the image thumbnail located at the left side of the prompt. For this example, I leave the prompt without changes and click Generate. And here we go. We see that the texture of the statue has changed yet again. Now the line is styled with intense striking red and orange colors that really resemble lava. Let me quickly show you several other style images that are used for this prompt. Please note that when you apply a reference image to the original one, a statue of a line in my example, the original image will be regenerated again based on the prompt. At the moment of recording the video, Firefly does not support modifying the exact same image. But perhaps this functionality will be introduced to the product over time. So let's monitor the product updates. 11. Match the style of an existing image with your reference image (part 2): If we want to be really creative, we can create a texture made of flowers and use it as a style reference. Let's try this. I'll remove this style reference and type in a prompt, Anki macroscopic flowers, three D material sphere on a white background. Let's also change the aspect ratio to square, and I'll click generate. All the results are very creative. But I want to use this image as a style reference because it has a clean almost white background. You'll understand why this is important in a moment. I'll go to edit and select use as a style reference. Next, I'll type in my original prompt. Let's also change the aspect ratio back to portrait. Let's see what results we get this time. We've got very creative images I could hardly imagine. Let's start by creating a new reference image. I'll remove the current style reference and type in this prompt, old marble texture. This time, I'll leave the aspect ratio unchanged since we don't have a three D material sphere in the prompt. Let's use this image as a style reference. We see it appears as a thumbnail on the left hand side from the prompt. Now, let's type in our main prompt, bronze statue of a line in a seated pose with a solemn expression. You see that this time we've got an entirely different result. Now the line is depicted on the marble surface, stylized with smooth lines and some gimetric shapes. To create a new style reference image, let's remove the previous one and change the prompt to unc macroscopic flowers with mountains in the background. For this work, I'd like to set the aspect ratio to white screen. I like this image. Let's use it as a style reference. I'll type in my original prompt, bronze statue of line in a seated pose with a solemn expression, and we've got completely different results as compared to our previous generation. Remember that you can also play with the strength of the style applied to your image by moving the strength slider. Let's move it to the left, reducing the influence of the reference image and see what results we've got. Okay. Very nice. I actually like these images better than those we got with the medium style strength. Now, let's increase the style reference strength to maximum. We see that this time, the images are much more stylized and closely resemble the style reference image. Let's set the style strength back to medium. Additionally, you can combine reference style images with other effects we've learned in the previous videos to create fresh and unique work. For example, let's add fantasy and digital art to this prompt to push fire fly to even more creative results. Remember, you can search for the style effect by pressing Command F on a Mac or Control F on windows, and then typing in the style name. Great results. So which image do you like best, please share in the comments to this video. I think with this possibility of creating your own style reference images and mixing them with different effects, lighting, and composition, you can unleash your inner designer, even if you belong to a non art related field from tech to business. And you know what I've noticed. The more you work with firefly and try various artistic movements, styles or textures, the more attentive you become to the details surrounding you in your daily life, whether it is the texture of a coffee table in a coffee shop near your office or the way the light casts shadows in your living room at dusk. AI generators like firefly seem to have the power to fine tune our perception of beauty and design, encouraging us to observe and appreciate the intricate aesthetics in the world around us. This shifting perception can enhance our problem solving and creative thinking skills, so much needed in other aspects of our lives. Not just when we are sitting down to create something. Have you noticed the same? And what do you think about the reference style image feature? Please let me know in the comments below, and I'll see you in the next video. 12. Match the image structure with the Firefly reference image gallery: Everyone, and welcome back. In this tutorial, we are going to talk about a brand new feature of the text to image functionality. The structure reference, structure reference, allows you to set an outline or depth for your image. Let's see how it works. I already have the text to image functionality opened. From this screen, go to the structure section. Similar to the style match feature, you can upload your own structure reference image or use one from the Firefly gallery. In this tutorial, we will work with images from the gallery. For our first example, let me type in this prompt. I'll remove the styles that I had from the previous image generation. Next, I'll choose the structure reference. For this, I will click on Browse Gallery and browse through the gallery of structure reference images. For this work, I'd like to choose this image. Let me also change the aspect ratio to portrait. And I will click Generate. These are the results. Notice how firefly matched the composition of these images with the reference picture and how nicely the pancakes are arranged in a stack. You can also use the strength slider to control how much the generated images will adhere to the reference image. Let's change the strength settings by moving the slider all the way to the maximum. I like the previous results better, but let me also try to move the strength slider all the way to the left, reducing the influence of the structure reference image. These toasts look delicious, but I prefer the results that we've got when the strength slider was set in a medium position. Let me know in the Q&A section for this video, which image you prefer. Let's continue with the D. I will styles and type in the next prompt. And for this example, I'd like to use this structure reference image. With this structure reference, we got images that represent a low angle shot looking up towards the sky which emphasizes the height of the buildings. Let me change the aspect ratio to white screen to see if we can get better variations. Yes, I think this aspect ratio works best for my prompt. You can also combine the structure reference with style reference. For example, let's add this style reference to the prompt. I very much like the sense of fantasy in these images from the vibrant colors to the moon size and position relative to the clouds and buildings. The images have a very surreal quality quite different from theistic blues and whites of the previous images. Let's test other style effects with the same prompt and structure reference. Okay. Of course, you can also add various style effects to the prompt. Let me add fantasy and see what results we'll get. Let me save this image to favorites so that I can access my work later on. I can continue experimenting with the prompt by including additional style effects. And let me remind you that you can search for the effects by using the search function of your browser. Click on Command F, if you work on MAC or Control F, if you work on Windows, and then type in the first several letters of the effect that you like to apply to your work. Okay, we've got very surreal results here. I'm happy with these images, and let's continue our demo. I'd like to test a prompt with which you should be well familiar. If you've watched my previous tutorials, let me clear the styles and type in the prompt. Let me change the aspect ratio to portrait as the structure reference, I'd like to use this one. Let's click on generate. Let's also play with the strength. I'll move the slider all the way to the right, increasing the influence of the structure reference image and click on Generate one more time. Given that we are using a picture of a cat as the structure reference file, you see that the whiskers of the lion are particularly noticeable. They are long and slander. Definitely, with the structure reference image, the lion started to resemble a cat at least for a bit. All right, that's it for this tutorial. In the next video, we continue talking about the structure reference feature, and we'll go over how to use your own structure reference images, including sketches when creating unique artwork. Ilsa in the next tutorial. 13. Match the image structure with your reference image (part 1): Everyone, and welcome to the second part of the tutorial where we are talking about the structure reference feature. In case you missed the first part of the tutorial, please watch it first so that you can follow along with this tutorial. As you might have noticed from the previous tutorial, Far fly provides quite a limited number of structure reference images. There might be cases when you won't be able to find a reference image that works for your project. But the great news is that you can use your own reference images as well. You have several options for how to do this. The first option is to use an image generated with fireflies text to image as a structure reference. For this, select an image that you'd like to use as a structure reference, and then click on Edit. Use as a structure reference, and as a next step, type in your prompt in the prompt field. Another option you have is to upload a structure reference from your local drive. Let's explore several examples of using your own reference image. First of all, you can use sketches as a structure reference. This is a drawing prepared by my daughter, and I'd like to create an illustration based on this image. Here is what I'll do. As a first step, I will upload the image to the reference image gallery. I'll click on upload and choose the image from my local drive. You see that the structure image thumbnail has been changed. Let me remove the effects from the previous generation. I'll leave only the structure reference here, and I will type in my prompt. Coming up with the prompt idea for illustrating your sketch can be challenging when you first start experimenting with the structure reference. After rounds of iterations, I came up with this workflow. I first define the subject matter or what I'd like to see in the image. In my example, I'd like to see a space strip. Then I take a look at the prompt suggestions that Firefly gives me. Remember that to see these prompt suggestions, you need to activate the suggestions here. If I don't like any of the options on the first page, I click refresh and look at the next page. I don't necessarily copy paste the prompt suggestions from this list. But they really help me brainstorm what descriptors I can include in the prompt in addition to the subject matter. If you struggle with thinking of possible prompts for your sketches, keep this workflow in mind. Let me type in the prompt that I came up with using the prompt suggestions from Firefly. Let me change the aspect ratio to portrait and generate one more time. Great results. Notice how the outline of the spacecraft and its position resembles the spacecraft in the structure image. That's really amazing. Let's also experiment with the strengths slider. I'll first shift it to the right, increasing the strength of the structure reference. Next, I'll move the slider all the way to the left. I'll move the strength slider back to medium, let me also include style references. Let's try this one. I also would like to move the strength back to maximum as I think this setting works best for my image. Let me click on generate one more time. And let's try Neon as well. I'll keep the strength slider on the maximum. I'll save this image to my favorites, let me show you another sketch that I've prepared for this demo. Let's try to use it as a structure reference and see what results we'll get. I'll drag and drop the file to the reference section. Let me remove the style reference and type in the first prompt that I've prepared for this demo. D. D. No, no, no. Hello. I'm very excited about this opportunity to use sketches as structure reference images. And of course, apart from children's drawings like in the examples we just watched, you can use your own drawings that you can quickly put together and want to transform into digital art. 14. Match the image structure with your reference image (part 2): All right. Let's continue our experiments. Another case for using the structure reference is for illustrating some complex concepts that are quite challenging to visualize yourself. For example, in the lecture on how AI generates art, I needed to visualize a big invisible map. Yes, I didn't exaggerate when I said that this task is challenging. I came up with this simple picture to illustrate the map, but I think that I can improve it with the structure reference. Let's see how it will work. I will drag and drop the file into the structure reference section. I will remove cartoon and then change my prompt. We had the strength slider at the maximum. Let's shift it to a medium position and see if we can get better results. But before we do this, as always, let me save my file so that I can access this project later on if I need to. After you watch the lecture on how AI generates art, please let me know in the Q&A section for this video, which of these pictures would illustrate the big invisible map best. Those with the strength set at medium or those that we saw earlier with the strength set at maximum. All right. There is one more example I'd like to show you before finishing this tutorial. The structure reference might work well for brainstorming ideas for logos, for your business or product. I've prepared this structure reference image to start brainstorming my company logo. The image includes two capital letters for my company name, future Varsity. This particular image has been created in Canva, a graphic design tool that can be easily used by everyone, including non designers. If you're interested to know how I created this specific file in Canva, I'll leave a link to the tutorial file in the resources section for this video. Alternatively, you can create structure images similar to mine in Adobe Express, a graphic design tool from Adobe, that we are going to cover in the upcoming sections of the course. Please stay tuned. Let me submit this image as a structure reference and type in my prompt. Since we are designing a company logo, the prompt should describe the nature or idea behind the business. I'll future learning concept, logo design, and I'll click Generate. Let me also choose the aspect ratio to square since we are designing a logo and click Generate one more time. Not bad, but the text is unreadable. I would need to delete it. As a heads up, you can easily remove desired elements from your image using firefly generative feel function that we will be working with in the next sections of the course. As always, I can modify the strength to see if I can get better variations from firefly. Let's do this. Let's take a look at other images that I generated before with the same structure reference to access the images, I'll go to F fly home page. From here, I'll click on Favorites. These are the images that I got with the similar prompt Here, I tweaked the prompt a bit. Okay. And here is another variation that I got. This one looks quite interesting. Here, I changed the prompt to show how the logo would look like if I ran a food related business. I like this variation. Let's look at other similar examples. D. Of course, some of the logers that we've seen just now still require some post processing like removing undesired elements or changing the background. But considering how easily and quickly we came up with these pictures, I can say with certainty that using the structure reference for logger design is a very promising idea. Cool. That's it for the tutorial. Please let me know in the Q&A for the video, How you use this feature. Okay. 15. How to exclude something from your image: Everyone, welcome back. Sometimes to get the perfect final output from the text to image model, you have to exclude certain elements, like texts, shapes or specific colors. The good news is that you can easily create negative prompts by telling firefly to leave out specific elements while processing your prompt. However, I need to warn you up front. It's not a 100% guarantee that if you set something for exclusion, it will be removed from the final image. There will be cases when to get to the desired outcome, you have to tweak your prompt rather than excluding something from it. Let's jump straight to the text to image function of firefly to take a look at some specific examples. To create a negative prompt, go to the advanced settings in the text field, add up to ten words that you want firefly to avoid when creating images based on your prompt. Use the comma or the return key to add each word of rase. Currently, you can type in words of rases in English only. Let's take several examples. First, I'll try this prompt, serine beach scene at sunset. Let's first see what firefly generates for that prompt and then decide what we'll try to exclude. These photos look good. Now let's try to exclude buildings. I'll type in buildings in the advanced section and press the return key. Here is what we've got. I like the results. Firefly did a great job. Okay. Let's try another prompt, flower garden in full bloom. This time, I want to exclude garden paths, buildings, and pink color. I click Generate to see the results. Well, I have to say that the job is partially done. I still see garden paths here and there, as well as buildings. There is also a pink color that I wanted to avoid. We can try to refresh the generation to see if we can get better results. They are pretty much similar to the first one. This is an example of when the exclusion function doesn't give the desired results. We have to think about how to modify the original prompt. Let's try this prompt instead, plain garden full of flowers. Let me also remove the excluded words and press generate. Much better. This time, we don't see garden paths and buildings, and the pink color is not as prominent as in the previous example. Let's try another prompt. Vibrant Caral reef underwater scene filled with various marine plants, let's exclude yellow. Definitely not what we are looking for as we still have plenty of yellow colors on every picture. Let's rewrite the prompt into something like this. Coral reef underwater scene, filled with various marine blue, purple red plants. Let's also remove yellow from this list, much better this time. Let's do one more example. I'll Picnic set up in a park with a blanket and a basket I'm going to exclude red. All right, the results are definitely not in line with our expectations. Let's change the prompt by specifying what color we want our blanket to be. I'll add ivory before blanket and also remove red from the excluded words. Perfect. This time, we've got much better results. All right. So after testing the exclusion function, I cannot say that it works reliably well. So to save your time in generating images that you want, try to specify all the details in your prompt rather than trying to exclude something from the image. And that's it for the video. I'll see you in the next one. 16. Top-7 recommendations for creating effective prompts: Hello, everybody, and welcome back to the lecture. As this section of the course is all about generating images from text instructions. We cannot overlook such an important topic as how to create those instructions or prompts. So in the upcoming video, I'll share my top seven recommendations on how to craft effective prompts. Let's get started. Sometimes you'll see solid outputs with simple open ended props, especially if you are open to surprises. However, when you have a specific vision in mind, describing various details can help lead you to perfection. But regardless of the direction you want to take, I recommend starting with a simple prompt and then adding extra details one by one to see how they affect the image. Begin with the description of your subject matter. There is an animal, landscape, fictional character, and so on. Generate your first image and then include extra details or content like its location. Information about the environment and lighting, as well as emotions or moods you like to introduce. Please note that in Firefly, you don't need to use words like generate or create in the description of your prompt. To clarify the idea of what you want to create. It is helpful to ask yourself a series of questions. For instance, here is a checklist that you might use. Decide if you want a poder or an illustration, What is your subject matter? Think of specific effects and details you want to include. Please don't feel overwhelmed or frustrated by the number of details and nuances that you need to know, and remember to craft a good prompt. If you watched previous videos in this section, you already know that all these effects and styles are embedded into fireflies interface. So all you need to do is select a category you're interested in, browse through the effects and choose one or several that you like. Please experiment with the settings to find those effects and styles that you like, and that may work best for your creative ideas to speed up your experimentation process and save your generative credits, please refer to the course handbook where I've collected images created with different effects under the same prompt. So that you can see the differences. You can click on each image to go to the prompt inside Firefly and try it out for your own project. When working with effects, don't try to add many of them in one goal, but rather approach the process iteratively. Seeing how adding one new effect changes your image. Please don't forget that you can adjust the strength of the style effects by adjusting the position of the strength slider. This is quite an effective method. If you want to push firefly to maximize or minimize the application of the chosen effects. As you might have guessed by now, you can go far beyond the current interface of firefly and include additional descriptions in your prompt that may take the AI creative process in a completely different direction. Or add extra flavor or nuances to your images. Here are just some examples of what you can add to the type of photography, environments, emotions and moods, magic words, specific art styles. I've prepared a quick demo where I've demonstrated how final images have been changed by adding the details from the list. Let's take a look. B. All right, let's continue with tips and tricks for creating great prompts. My recommendation number four, by attention to the order of the words in your prompt. Let's take a look at these two examples. You see that I used the exact same words for generating two images. However, the final output is different. In the first example, we see that more emphasis is done on the Snowvy landscape rather than a wooden cabin as I placed snowy landscape at the very beginning of my prompt. So the words at the beginning of the prompt have more weight than the words at the end. Let's move next to recommendation number five. Be respectful of third party rights and don't use keywords that violate copyright, trademark, privacy, or other third party rights. For example, these are the results I'll get if I try to generate an image with the prompt, Tesla style humanoid robot. I'll get similar results if I include such keywords as in the style of artist's name into my prompt. So here is what I get if I ask firefly to create a picture in the style of Salvador Dalm. In case you've experimented with other IR generators before, you've probably used artist keywords in your prompts, so it may feel counter intuitive to remove them from your prompt. However, Adobe is committed to setting the industry standard for the responsible use of generative AI and ethical creativity. And as such, Adobe encourages the use of descriptive prompts, style effects and compositional techniques without directly referencing contemporary artists or copyrighted characters. This approach ensures that all generated artwork respects the rights of creators while promoting an environment of innovation and inclusivity. Cool. Here is my next recommendation. Look for inspiration and examples when crafting your own prompts. Let me explain. If you are just starting your creative journey with AI and you are not a designer or illustrator. I have to warn you that it might be challenging for you to create detailed descriptive prompts. Don't be disappointed and give yourself some time to get used to the new tool. Start by looking at the ideas of others. For example, by browsing through Fireflies community page. Choose the ones you like, look at the prompts that have been used to create them Try to play with them by making some modifications and add or remove keywords, as well as adjust style effects. It is also a good idea to create a mood board or a gallery of images that you like and might use as a reference later on. For example, here is a moodboard I created in notion when working on this course. Here I saved the images I came across in the firefly community. That I might use as a reference or inspiration for my next project. In the board, I saved an image, the prompt that was used to create it, as well as styles and effects that were applied to the prompt. Notion has pre built mood board templates that you can use as well. And I also leave a link to this mood board in case you want to use it as a reference, and last but not least, don't forget to enjoy the process of creating the prompts. At first, it might be hard to see your own touch in the artwork because the EI is the one making it based on your prompts. But without your unique ideas on how to craft those prompts and how to blend them with different style effects, They wouldn't be able to create anything. B yourself, throw your ideas out there and make sure to have fun. These were my key recommendations on creating an effective prompt. To recap, start with the simple prompt and then add extra details one by one to see how they affect the image. Ask yourself a series of questions to clarify the idea of what you want to create. Go beyond fireflies interface and add extra descriptions to your prompt for more creative directions and nuanss in the final images. Pay attention to the order of the words in your prompt. The words at the beginning of the prompt have more weight than those at the end. Respect third party rights by avoiding copyrighted or trademarked keywords in prompts. Look for inspiration and examples in the firefly community and create mood boards with images that you like and might use as a reference. Last but not least, have fun creating your artwork. Later in the course, we will explore some additional more advanced techniques for crafting your proms. So stay tuned. 17. Text-to-Image functionality limitations and restrictions: Everyone, welcome back to the lecture. Now that we've covered the key things about firefly text to image functionality and also discussed the best practices for creating effective fronts. I want to share some use cases where firefly may not perform as expected and needs further improvement. But the purpose of the video is not just to brag about what doesn't work well. But first of all, to set your expectations that some edge cases still exist. And most importantly, to talk about how you can help to make the AI model produce better results. Let's get started. So out of hundreds of experiments I run while preparing this course, I notice that Firefly might not perform well with prompts that describe people, especially when you want to depict a crowded scene, like in this example, I also got some funky results when I tried more exotic prompts like photo of a witch on a broom or bally dancer silhouette. You may also experience challenges when trying to create images for non material objects and processes like software products or software development. The images may look too abstract and confusing. In some cases, I was able to get better results by tweaking and modifying my prompts. But there were cases when I got unsatisfactory results repeatedly. But as I mentioned earlier, the good news is that you can contribute to improving the images generated by AI. Let's see how to do this. If you believe that the generated image has room for improvement, click on the thumbs down icon located at the bottom of the image. Next, click on the feedback. You'll see a screen where you can specify what went wrong. For my example, I'll select poor quality. You can also choose other and type in your feedback. In case you believe that the image violates user guidelines and for example, contains harmful illegal or offensive content, press the report button. You'll see a pop up screen where you need to specify more details about what went wrong with the image. You can choose one or several options from the list or type in your feedback manually. There is another way you can help with the firefly model training and feedback. For this, click on the laboratory flask icon located on the top right side of the screen, and then select Rate firefly images. You'll see a screen with two images and the prompt that was used to generate them. Your job is to select image A or B that you like the best. If you don't want to rate these images, click skip to go to the next pair. If you feel that the content violates the user guidelines, click report. As soon as you rate the current image, Firefly shows you the next pair, you can give feedback on. The process continues unless you close the window or switch to other firefly functions. Another issue that you may come across when using text to image functionality is a problem with generating numbers, letters, and texts. The best way to go here would be to generate an image without numbers or texts. Remember, you can mark them for exclusion in the advanced settings. And once you get an image that you like, You can work on it further in such tools as Ad Express or Canva or any other similar tool of your choice and add text letters or numbers. And if you're using express, remember that it is already integrated with Firefly. So you can open an image generated with Firefly and Express and continue editing it with just one click. In the following videos of the course, we'll have tutorials where I'll show you how to edit the images in Express. All right. Let's continue. There is another feature that's constantly asked about and requested by firefly users the ability to create multiple images with different backgrounds and settings, but the same main person or character. Unfortunately, at the time of recording this video, this option is not supported by firefly. Even though this feature is something that the firefly team is probably already working on or is planning to work on. There is no official information from the team on when it will be released. But in case this feature becomes available, I'll of course make changes to the tutorials and notify you about it. State. And by the way, if you have a feature or idea in mind that you think can improve firefly, you can submit your request. To do this, click on the laboratory flask icon and go to submit an idea. On the next screen, fill in the subject and write down the details about your idea. You can also select texts for your post and drag and drop files or pictures. Please note that you can also submit a product bug or technical issue with the product by selecting the buck category as the conversation type. Another thing that you may come across working with Firefly is that it will try to tailor the results based on where it thinks you are located. For example, one of the users who lives in Netherlands reported getting Dutch themed pots even when it is not in the prompt. Stuff like tulips, windmills, the Dutch flag or people wearing hats in the Dutch flag colors. Another user who is in Australia tested a prompt device used by students and got an Australian flag appearing within the generated images. There are other similar issues that other people are discussing in the community. I haven't found an official communication from the firefly team regarding the issue and how to overcome it. But it appears that Firefly may be looking at your PAddress location and giving you a personalized version of the images. So to overwrite this behavior, you could either modify your prom description or experiment using a VPN or virtual private network service, which can mask the user's IP address and location. Of course, you can also provide your feedback by clicking the thumbs down over the image and giving written comments related to what's happening, similar to what we've covered at the beginning of the video. That's what I've noticed so far. Perhaps you may come across some other edge cases or behaviors that surprised you or when you find a good solution to the problem. Please share this information with other students of this course by writing a comment in the Q&A section of this video. And that's it for the lecture. Let me sum up the key points for you. Firefly text to image functionality may not perform well with prompts that describe people, especially crowded scenes or non material objects and processes. Taken prompts can sometimes improve results, but the right cases of repeated unsatisfactory outcomes. Firefly struggles with generating numbers, letters, and text. So it is recommended to exclude them from the image generation. And add them later, using tools like AdobEpress or VA. Currently, Firefly lacks the ability to create multiple images with the same main person or character. Additionally, firefly may tailor results based on the perceived user location, leading to localized themes in images. You can overwrite this by modifying prompts or using a VPN. At cool Alsa in the next video. 18. Introducing Advanced Techniques in Text-to-Image Functionality: Every one, and welcome to the new section of the course. Now that you have all the essential knowledge on how to work with fireflies text image functionality, it's time for learning more advanced techniques. In the upcoming videos, we will cover how to enhance your work process and come up with even better prompts descriptions using additional tools. These include Adobe stock, a service provided by Adobe that offers a vast collection of high quality royalty free images, videos, templates, three D assets, and more. Users can license these assets for different purposes from graphic and website design to video production, ensuring that their creative work is both professional and legally compliant. Another tool we are going to explore is Chart GPT, an application developed by Open AI, a major tech company focused on generating human like text in conversational formats. It's capable of understanding context, generating responses, providing information, and maintaining coherent and relevant conversations. The latest version of chart GPT, powered by open yes most capable AI model to date, GPT four has the capability to understand and generate descriptions for images. This means it can process both text and image inputs to provide relevant responses descriptions or generate text based on the content and context of the images, and that is what we are going to explore in detail in the upcoming lessons. And the third tool we are going to talk about is Microsoft copilot. Copilt is a chat assistant from Microsoft, powered by GPT four. All of these three tools will help us to find relevant and accurate keywords that you can embed into your prompt description, and as a result, get a more nuanced, detailed image that would be hardly possible to create otherwise. Please note that this section is optional, and you can come back to it later on whenever you have some spare time to master the advanced prompting techniques. All right. That's it for the intro, and I'll meet you in the first video of the series where we are going to talk about using Auto stock in your work. Ilsa. 19. Creating prompts using Adobe Stock image gallery: One welcome to the video. So let's kick off these advanced tutorials by covering how to enhance your process of creating art with AI by using Adobe stock. By exploring Adobe Stock and its many images and texts, you can greatly enhance the results of Firefly text to image feature. Given that Firefly training includes Adobe stocks extensive library, Understanding the types of images and texts it contains helps you to create prompts that align closely with fireflies knowledge base. This alignment between your prompts and Firefly training leads to more accurate output images that reflect what you envisioned. You could navigate to Adobes stock website and search for images similar to the one you'd like to create. As you're just browsing through the pictures and not downloading them, You don't need to have an account with Adobe stock. For example, I'd like to create a picture of a tropical bird. But let's say I don't have enough vocabulary to describe the appearance of a bird. I want to see in my picture. Let's try to find some images from Adobe stock database that we can get inspiration from. For this, I will choose images on the left hand side from the sage bar, and in the surage bar, I will write tropical birds. I'll heat, enter Next, I browse through the results, looking for the pictures I like. This one looks great. Let's drill down to the detailed view of the picture to analyze the information provided about that photo and what words actually describe the image. I usually look at the words used in the title category. As well as keyword section. You can click on view to look at the entire list of keywords. Next, I will select a few keywords that I think are most relevant to the image I want to generate and I use them in my prompt. For this example, I'll definitely want to use Northern Cardinal as the main subject. I can also copy the words that describe the action of the bird like flying up to bird feeder for a bite of sunflower seeds. Let's use this combination of words to get our first images. Let me switch to firefly, from the home page, I will open text to image function. Let me type in my prompt But actually, I can copy the words from here. I'll click Control C and then Control V and make medifications if I need to Let me say Northern Cardinal fline to a bird feeder for a bitle of sunflower seeds. I'm fine with this prompt, and I'll click on Generate. Here are the first pictures that we've got. Let me change the aspect ratio to white screen and click Generate one more time. Sometimes experimenting with the aspect ratio can help in getting the best outcomes. Nice first results. The bird looks quite realistic to me. Okay. Let me save this picture in case if I need it for the future reference. We can continue modifying the prompt by adding some adjectives from the list of keywords from the Adobe stock image to further describe the bird. Let's look at the keywords that we have here. Let's add exotic colorful and we can also add Wild. Okay cute results. I usually do several iterations of including the keywords from the Adobe stock. Until I don't see any more improvements from the extra words I've added to the prompt. Apart from keywords from Adobe, you can also include additional modifications into the prompt, as well as a style effects so that the final pictures fit closer to your vision. Okay. For my example, let me change these words. These pictures are truly cute and exotic and look at the attention to details. For example, here, we even see the individual feathers of the bird. That's really fantastic. I will save this picture to my favorites so that I can continue editing it later on or use it in my other projects. We've just looked at the use case when we refer to Adobtok for inspiration. To find the reference images that we like, see how they're described and use those keywords in our work. Using the Adobe stock database may also come in handy when you're having trouble generating specific things or objects. In case you try a handful of prompts, but still cannot get the desired results. It is always helpful to do a search on Adobto to see if there are references and how they are tagged. Then choose the most relevant texts and embed them into your prompt. Third use case of using adb stock in your work is when you combine keywords from different ADB images into your prompt to get some really cool creative results. That is by far one of my favorite techniques as you would never know where the relevant texts and AI will take you. In the next video, I'll show you this technique in action. So if you'd like to follow along, I'll see you there. 20. Follow-along: Combining several Adobe Stock images into one masterpiece: Okay, a Adobe stock website already opened. From here, you can choose content type, and type in your keywords into the Sage bar. Alternatively, click on the stock photos Card and choose your content type, and include your keywords into the Sage bar. You can also apply additional filters. For this, click on collections. On the next screen, choose filters. On the next top, you can choose such parameters as subcategory, price, artists, background. You can also decide if you'd like to see images generated by AI into the search results. In case you'd like to exclude them, generative AI. If you have a reference image and would like to find similar ones, click on Find Similar and upload or drag and drop your image here. You'll see similar content on the right hand side of the screen. All right. For my work, I'd like to use two Adobe stock images. I'll first search for contemporary art and rows through the images to find something that I like. This image is interesting. Let me open it in a new top. I think it looks fun and humorous. I'd like to pick up on this idea for my work, but I want more fashionable outfit. For the second image. This Tile type fashion style. And let's browse through the images. Many of them look great, but I like this bright image. Let's open it in a new tap as well. Okay. Very nice. We have two of the reference images selected and we are ready for our experimentation. This picture is going to be our main reference. Let's look at what keywords are included in its description. Let's look at the keywords used in the title. Category. And keyword section. Our task here is to try to use a handful of these keywords in our prompt. I think I'll definitely use this description of the subject. I also would like to include something for the outfit. Let's switch to the second image and look for the keywords here. I also look at the header category, and let's check keywords description as well. I'd like to use this one colorful transit jacket, and we can also include vintage retro sunglasses. All right. We're all set. Let's open firefly at firefly dot.com. Open text to image functionality and type in the prompt. A human body Okay. But actually, let me copy paste the keywords. I'll copy these words and paste them into the prompt. Next, let's add rein, and I'll open the second image and copy paste this description, colorful trained jacket and vintage glasses. Okay. All right. Cool. Let's click Generate. I think that for this work, I'd like portray aspect ratio. Let's change that. Let's generate one more time. All right. Here are the first results that we've got from firefly. Let's try to add more keywords into our prompt. Let me shift to the main reference picture. Okay, let's check the header for the strong keywords that we could use. I like so, so let's copy and paste it to our prompt. And I'll click generate. Try to apply one or two keywords at a time to decide if you like their influence on the image. Okay, I definitely see some improvements. I like this image. It's really stylish but fun and humorous at the same time. I like the vibrant colors on the blazer looks nice. We are moving in the right direction. Before we include additional keywords, let's play with the visual intensity. I'd like to move it all the way to the right and see how this change would influence the image. Let's generate and wait for the results. Our new pictures are more stylized than those we generated with the medium visual intensity. I like this texture of the blazer over white shirt and pink tie. Let's start saving our images so that we can refer to them later on. Now I'd like to continue my experimentations and move visual intensity all the way to the left. This is an interesting style as well, but I think I would prefer visual intensity set in the middle. Let's get back to the settings. Let's also try to include vibrant colors to our prompt and see how this change would influence the pictures. I'll go to color and tone and choose vibrant colors. Let's click generate. All right. I feel that the pictures became too bright and for me, it's a bit too much. I would prefer to clear this style and continue our experimentation with the prompt description. Let's come back to our reference images and look for some other keywords. Let's look at the cool teenager image. Here we have description of your style as style of 80s. I think we could add these words to our prompt. Let me copy style of 80s and included before the shocking art, and I'll generate. Let's also check that we have visual intensity set in the medium. Yes, everything is correct. Let's take a look in more details. I like this background. It's simple, but at the same time this gradi adds extra static to this picture. I'll save this image for the reference and continue including the keywords. But before that, let's shift visual intensity to a maximum one more time and see if it's going to improve the image. We've got different background with some landscape and the shirt was changed to hoodie or something like this. Here we have more prominent landscape and mountains on the background. Let's keep visual intensity on the maximum level and add more keywords. I'll return to the cool teenager, and I would include a disco party and see what direction firefly would take us. Let's include the keywords after sunglasses. And click generate. All right, we are getting more and more interesting results. I like this fantasy boca effect in the background of the image. This background is very vibrant and abstract, and here we even have some futuristic city scape and planet on the background. Let's try to include something else. Boom. All right, this vibrant and abstract background reminded me that we could actually try to use a reference image from the Firefly reference image gallery that could probably enhance our image. Let's try to do this. So I'll go to reference image gallery. And from here, I'll select digital illustration. You see that the background of this illustration is similar to the background that we've got on the second image. I'll leave my prompt as it is, and heat generate. Let's come back to the control panel and play with the style strength. By default, the strength is set on medium. Let's shift the slider all the way to the right and generate. Let's set the slider to the minimum as well. I like the medium style strength settings best. So I'll return this slider into the default position. Also, let's remove the style reference and continue including the keywords from two of our reference images. B. Okay, I think we've reached the point when we don't see any significant changes from adding something extra to our prompt. So let's try applying different styles. Remember to apply one style at a time. I like how the fantasy style has changed our image, and I think it is now resemble more of a contemporary art collage. Let's add additional effects such as modern art, psychedelic and surrealism. I'll add one effect at a time. D. D. Okay, I think you would agree with me that we've got very creative, even crazy results, especially if we compare this image to our very first generation. I would prefer to change this background to something more, even monochromatic so that the figure in the foreground gets more attention. I'll show you how to remove the background of this image as well as change its color in the upcoming tutorials for the course. And that's it for this tutorial. Please let me know what you think about this technique in the Q&A section for the video. Of course, share the links to your projects. 21. Creating prompts with ChatGPT: Every one, and welcome to the video. Let's continue talking about some advanced techniques of how you can enhance your process of creating AI art. In this video, we will talk about using such a tool as Chart GPT. As always, I'll first explain the idea behind the technique and then show some examples of its application. Let's get started. In the previous two videos, I showed you how to search images from Adobe Stock and look for relevant texts and keywords that describe them so that you can incorporate these words into your prompt. But what if you have an image that is not part of the Adobe Stock library? Let's say you made a photo and want to use it or part of it as a reference for your firefly project. And you'd like to get some keywords on how the picture can be described. Instead of searching for keywords in Adobtok, you may upload that image to Carpi with the request to describe that image. Next, you scheme through the description of the image and look for relevant keywords that you can use in your prompt to get similar results. For example, for one of my projects, I wanted to create an image that depicts this type of classical architecture buildings with light projection artwork that I observed during a visual art festival in Singapore. However, I wasn't sure how to describe this type of architecture in my prompt. I've decided to use Cj Pi to help me here. Let me open C P to demonstrate my example. Here is the first screen of Carp. If you don't have an account with Open AI, click Sign up. Provide your e mail that you want to use for the registration. Click continue, and then choose a password. I already have an account with open AI, so I'll be using this account for the demo. Now, there is one thing that you should be aware of. If you would like to experiment with the technique that I'm going to describe in this tutorial and try other features that are part of Open AI paid plans. Consider upgrading your account. Of course, you can cancel this subscription anytime. For this demo, I'll be using my plus account. The step one of the process is to upload your reference image to the prompt tab. For my example, I Googled reference image by keywords, National Gallery Sapore I remember I saw the light projections I want to create on that same building. Let's click on images. Here, I need to select the image that I'd like to use as a reference. This one looks good, so let's copy it. Return to chart GPC, and paste it to the prompt tab. To paste an image, click Command V, if you work on a Mac or control V, if you work on Windows. We see that the image thumbnail appears at the prompt top. You can also upload the image from your local drive. The second step will be to specify what you want to do with the picture. I usually write a very simple prompt like the image. But in case if you are interested in getting a description of a certain part of the image, you can specify this as well. For example, describe in detail what is depicted in the foreground of the image. I leave the prompt as it is and click Return key. Okay. Okay, we see quite a lot of keywords from this paragraph. Large neoclassical prominent dome, grant, symmetrical facade, multiple levels. Probably we can use that as well. Let's open firefly and try creating a prompt using some of these words. I'll go to firefly.dot.com. Next, I click on text to image functionality. Here, let's type in the first prompt. And I'll click Generate. We've got some first results. For this work, I think that wide screen aspect ratio would work best. Let me choose it and click Generate. We are getting results that are really close to what I want to see in my project. Okay. I like this image, so let's save it to favorites. If you need more options, click refresh or add more words to the prompt. Let's try to add some additional words. Let me return to CG PT and read the paragraph one more time. Let's try tall windows, evenly spaced across the front. I'll copy the text and paste it to my prompt. Let's add with tall windows, even spaced across the front. Let's click Generate and see if we can get better results. All right I think all the images great. Maybe the first one is a bit distorted, so I won't be using it for my project. I'd like to save this one. Which I think is the best out of all the generated images. You can continue experimentation by adding new keywords into the prompt, refreshing the prompt, and of course, adding styles and effects from the control top. Let me show you the final results that I've created earlier and saved to my favorites folder. I've experimented with different reference images and also changed my prompt several times to find the best outcome. Here are the images that are closest to what I was trying to create for this project. Let's go over another example. Here is a picture of a bronze statue of a line that I made in a coffee shop near my house. I thought it would be cool to use it for some of the prompt examples for this course. I hope that you've recognized the picture. I use it in the lessons where we spoke about how to use reference images. I wanted to get more ideas on how I can describe this statue. I've uploaded this picture to CPT and asked to describe it. Okay. And here are the results that I've got. The next step is to find several keywords that you think describe the picture best and can be relevant for your firefly work and create a new prompt. Let me create the prompt with these keywords. I'll delete the reference style and change the prompt to the new one. Bronze statue of lion, intricate details sense of texture and depth. I leave aspect ratio as it is, and click generate. I like this results, but I can go even further by including more keywords into my prompt or adjust in visual intensity of the image and including additional styles and effects. So what do you think about this technique. Do you like it. Please share in the Q&A section of the video. Personally, I like to reverse engineer the prompting process this way. But I use a paid version of CGPT for my other work. I didn't have to sign up for it just to experiment with this technique. However, if you don't use the premium plan, there is another way you can achieve similar results with another tool, Microsoft copilot without having to pay anything extra. I'll see you in the next video where I'll share more details. 22. Creating prompts with Microsoft Copilot: One. Welcome back to the next video where we talk about advanced techniques for generating images from text prompts. In the previous videos, you learned how to use Adobe stock and Open AI chat GPT in searching for relevant keywords that you could incorporate into your prompt to get superior results. In this video, we are going to discover how to use yet another tool. Copilt to recap, copilot is a chat assistant from Microsoft. Powered by the latest and most capable open AI model, GPT four? Remember, we talked about the fact that you'll need to have access to GPT four available in the paid plans only to upload your reference image. Well, with copilot, you'll have access to GPT four in the free plans, at least at the moment when I record this video. What's the trick here? Why is something you have to pay for in one product available for free in another. With the copilot free plan, you'll have access to GPT four during non peak times. There is no clarification on what peak time means. However, I've tested the free version many times when preparing for this video, and I never encountered the situation when GPT four was not available. The copilot free plan should work well for our needs. But if you'd like to explore the pro paid plan, which is also available, I leave a link in the resources section for the video. All right. Before we dive into co pilots functionality, let me explain the project I have in mind. I'd like to create an image of a futuristic cityscape. The city we could live in a distant future, like 30 50 years from now. I have these reference images. I really like this futuristic building, and I want to get some ideas of keywords on how it could be described, and by the way, these pictures are real. Probably some of you have recognized this place. This is Singapore Marina Bay, and the building I want to use as a reference is the first of its kind Apple Store, appearing to float on water in Singapore Marina Bay. Let's see what futuristic cityscape we could create with this reference. We are going to start our work in copilot and then switch to Firefly. Similar to Chat GPC, copilot is available in web as well as mobile versions. For my workflow, I usually use copilot on mobile, so I'll use it for this demo. All right. Let's begin by downloading Microsoft copilot. I'm using an Android device, so I'll use Playstore to download the app. By the way, while we are here, let me show you another use case for fireflies text to image functionality. I created this wallpaper in Firefly as well. I used this prompt and portrayed aspect ratio to generate the image. I think this background looks pretty unique. Keep this use case in mind. I'll open the store and search for copilot. Here is the app. Let's click Is style. All right, the installation has been completed, so let's click Open. I'll choose continue. Copilot wants to access this device location. I'll click deny since this won't be required for our demo. All right. Here is copilots mobile interface, which is very clean and easy to use. Begin by selecting P four, which is open capable model, at least at the moment when I record this demo. Next. You see some ideas of how you can start the conversation with the chart board. You can also sign into the app to ask more questions and have longer conversations. However, this won't be required for our use case, so I'll keep this step. Next, you can either write down your question or record your voice. Before we ask anything, let's upload an image. Copilot wants my permission for taking pictures and recording video. I'll choose while using the app. Since I'm using a mobile version, I can take a picture and it will be automatically uploaded to the chart board. For my example, I'll choose an image from my gallery. Okay, I need to allow copilot to access photos and medias on my device, so I'll click. Next, I'll go to Phos downloads. From here, I'll select this picture. Let's write down some simple requests like describe the image. And I'll click Enter. All right. Here are the first results that we've got this text is not very informative, so we have just several keywords that we can use. We see that op tells us that the picture was taken in Singapore, which is correct. And we also have the information that the building is a large sphere floating on the water. I think it's not enough of descriptors for our purposes. So let me try to ask copilot to give us more details regarding this picture. For this, I can either choose from several questions that we have in the copilot interface or I can type in my question. So I will ask describe the floating sphere in detail, and here we go. Now we have a lot of information about the building with the keywords that we definitely can use in no prompt. We see here such descriptors as self supporting glass and steel dome. We definitely can use this. We also can use something from this sentence. It has an oculus at the top. That let's in natural light and ring of sunshade rings that provide shading and light effect. Yeah, definitely, we can use some of these keywords or maybe all of them in our firefly prompt. I really like these results. But in case if you would like to get more keywords from copilot, you can always submit additional reference images if you have them. So let me try to submit another image and see if we can get better options from copilot. Okay. Well, we've got some extra keywords like a modern spherical glass building near the waterfront. So we could probably use them as well. But to begin with, I would rely on the description that we've got in the previous iteration and see what picture we can generate with firefly text to image. Let me open firefly to continue with the demon. All right. So let's open far fight text to image function, and I'll type in my first. As you can see, I use the keywords that we've got from the copilot. So I will click Generate to see the first version that we get. Before we proceed with modifying our prompt. Let me move visual intensity to a maximum as I want fire fight to gives us something really futuristic, the building that we could see in the distant future. Let's generate and wait for the results. Yeah, that's much better. I was looking for those kind of futuristic results. I like this image best, so let me save it to my favorites. In case if I'd like to come back to this image later. Let's do a couple of more iterations. So I will add extra keywords to the prompt. All of these keywords was suggested by Microsoft copilot. We are getting more and more futuristic results. I really like the direction where we are going. Of course, apart from defining the prompt description, we can also apply additional style effects or maybe try out different fly reference images or use our own reference image to see if we can get even to our vision for this project. All right, cool. That's it for this tutorial. I hope that you enjoy working with another tool, Microsoft copilot. Now, I encourage you to try out this tool for yourself. Please let me know in the QNA section for this video. What do you think about working with Microsoft copilot as your assistant in generating effective prompts for firefly. I'll see you in the next video. 23. Taking a deep dive into the world of AI: Everyone, and welcome back to the new section of the course. Imagine stepping back in time and telling Leonardo Da vinci that one day machines could create art. And now we are very close to making it a reality. So let's dive into how I is not just learning from us, but also creating with us, challenging our understanding of creativity and innovation. In the upcoming three lectures of this section, let's cover some I basics, given you essential knowledge that will enhance your creative journey and deepen your understanding of AI capabilities and potential. We will start with an overview of the AI landscape as of today, speak about how the art generation technology evolved over time. And why it is recently became so popular with the explosion of new software and tools. Of course, we will discuss how the art generators work and how you can go from a simple text description to something like this. This section is optional. If you'd like to focus on the practical application of R generation technology with firefly, feel free to skip it for now and come back to it later whenever you are ready to learn more about AI. And for those of you who have decided to fall along, I'll see you in the next video. 24. AI landscape of today: Everyone, welcome back. In this lecture, we will go through an overview of the AI landscape as of today. First of all, let's define what AI is. In simple terms, AI is the ability of machines to learn, understand, reason and interact in ways similar to us humans. This allows machines to solve new sets of problems they could not before. For example, AI powers voice as systems like Syria, recommends movies on Netflix and help doctors diagnose diseases. AI encompasses a range of technologies from simple automated rules in everyday gadgets to advanced systems that learn and adopt. While AI can perform specific tasks at or above the human level at the moment of recording this video, it does not possess general intelligence or consciousness. Recently, I has also made significant progress in creative fields, generating art, music, and literature. That's exactly what we are going to explore in this course. Now that you understand what I is, let's discuss how machines actually learn. At its core, machine learning a key component of AI involves teaching computers to recognize patterns and make decisions based on data. This process is somewhat similar to how humans learn from experience. But instead of learning from life experiences, machines learn from data. Machines learn in different ways, mainly categorized into three types. Supervised learning, unsupervised learning, and reinforcement learning. These are what we call the foundational learning methodologies. Each of these methodologies has its own approach to learning and is used for different kinds of tasks. Supervised learning involves training AI models on labeled data. Labels are identifiers associated with input data. For example, they can be textual. In a dataset of animal photas, each phota input would be labeled with the name of the animal output, like cat, dog, et cetera. Another example is numerical labels that can be used to predict house prices based on features. Supervised learning essential for applications where the model learns to predict outcomes based on provided examples. This includes speech recognition, image classification, and expert systems, AI systems that mimic the decision making abilities of a human expert in a specific domain. Unsupervised learning focuses on finding patterns or structures in unlabeled data. In other words, it discovers the underlying patterns in the data without explicit guidance. The unsupervised learning is pivotal in domains like recommender systems, Systems that predict user preferences and suggest relevant items accordingly. It is also used in certain aspects of computer vision that focuses on enabling machines to interpret and respond to visual information from the surrounding environment. Third methodology is reinforcement learning. It focuses on training models to make decisions through trial and error, receiving feedback from the environment and learning optimal actions through rewards. It's key in robotics, autonomous vehicles, and some planning and scheduling tasks like resource management and automated scheduling systems. Please note that most application areas rely on a combination of different learning methodologies to leverage the strength of each. This approach often gets better performance and more robust solutions. For instance, many modern recommender systems integrate all three methodologies to leverage their strength. Supervised learning provides accuracy based on historical data. Like predicting and recommending new movies or products, a user might like based on historical data with user preferences or ratings. On the other hand, unsupervised learning offers insights into users, which might not be apparent through ratings alone. Clustering algorithms, a type of unsupervised learning technique that organizes data into clusters or groups based on similarities. Might find that certain groups of users tend to watch similar genres of movies even without explicit ratings and recommend movies based on these clusters. Finally, in case you want the recommendation engine to be dynamic, and adopt the recommendations based on how users interact with different content. For example, by browsing, watching trailers, selecting and watching movies, reinforcement learning comes into play. This system will learn by interacting with users over time and adjusts its recommendations based on user engagement and feedback. All right. Our overview of the AI application areas won't be complete without the other tool that also leverage all three foundational learning methodologies. These application areas are natural language processing or NLP and generative AI. NLP implies understanding, interpreting and generating human language and is used in such applications as language translation, sentiment analysis, chat boards and voice assistance. And finally, generative AI, the term that has become extremely popular in 2023 and that you probably have heard of before. It is an umbrella term that includes various techniques focused on creating new original content that never existed before that mimics or inspired by real world examples, I generation, something that we will be doing in this course with the help of Adobe Firefly, specifically refers to the use of generative AI techniques to create artworks. It is a niche within generative AI focusing on visual creativity. AI models in R generation learn styles, patterns, and artistic elements from existing artworks using foundational learning methodologies and then use this knowledge to generate new images, paintings, or visual content. In the following lecture, we will learn more about how the IR generation technology evolved over time and why it was taken by storm recently. But before we proceed, let's sum up this lecture. AI is the ability of machines to learn, understand reason, and interact in ways similar to us, humans. A key component of AI machine learning involves teaching computers to recognize patterns, and make decisions based on data. Machines learn in different ways, mainly categorized into three types or foundational learning methodologies, supervised, unsupervised and reinforcement learning. Supervised learning teaches AI with labeled data. Unsupervised learning finds data patterns without guidance and reinforcement learning involves learning via feedback. Most application areas rely on a combination of these learning methodologies to leverage the strength of each. Generative I is an umbrella term that includes various techniques focused on creating new content that never existed before, inspired by real world examples. That's it for the lecture and ACA in the next video. 25. Evolution of AI art generation technologies: Everyone, and welcome back to the series of lectures where we cover AI fundamentals. In this lecture, we will find out how I R generation technology came from being able to do this to this, and even this. Let's begin. Generative I research traces its history back to the 1960s. However, generative AI began to develop into something similar to its current form in 2006. The first significant paper in the field by Geoffrey Hinton and his co authors titled a Fast Learning Algorithms for Deep Belief Nets. But the first major breakthrough in the field of image generation happened back in 2014, with the introduction of a novel framework called Generative Adversarial Networks or GAs prior to introducing GANs, Is focus in the visual domain, was predominantly on image recognition and classification. Gans shifted this focus towards the generation of entirely new images. Initially, they weren't used for turning text into images, but rather for creating realistic images from random chaotic visuals. The gans then gradually transformed these initial iotic visuals into coherent and realistic images. Gans brought a significant shift in how machines could create visually detailed and realistic images. Laying the groundwork for more sophisticated text to image applications. Let's try to go a bit deeper and see how the model works. In a nutshell, the framework implied two neural networks, a generator, and a discriminator working against each other leading to improved quality of generated images. This description is quite technical. Let's simplify it for a bit. Imagine two people. Let's call them the artist and the detective. They play a game to trick each other but in a fun and creative way. The artist loves to draw pictures. But instead of drawing from real things, the artist tries to create drawings that look like they could be real even though they are completely made up. Think of the artist trying to draw a unicorn that looks like it could exist in real life. The detective is really good at figuring out what's real and what's not. When the artist shows a drawing, the detective tries to guess if it is a real thing or just make believe drawing by the artist. The artist keeps making new drawings, trying to make them looks as real as possible and the detective keeps trying to guess correctly. As they keep playing, both the artist and the detective get really good at their jobs. In the world of computers, the artist is like one part of the g that creates things a generator, and the detective is the other part that checks if they are good enough or not a discriminator. By working together and challenging each other. They end up creating really amazing things that can sometimes trick even humans into thinking they are real. Following the breakthrough of generative adversarial networks in 2014, the development of text to image models has seen several key milestones. In 2017, researchers at Google introduced transformer models that revolutionized the field of machine learning. Particularly in tasks involving natural language processing. Over time, researchers discovered that transformer models could be applied to visual data as well. One of the key features of transformer models is the ability to process entire sequence of data simultaneously, unlike traditional models that process data point by point. This feature enables transformers to understand the context and relationships within the data more effectively, making them particularly powerful for tasks in natural language processing and image processing. This became possible due to the attention mechanism, which is a cornerstone of transformer models. This mechanism allows the model to focus on different parts of the input data prompt and understand how each part relates to the others. You can visualize the transformer model as a super smart robot that can look at all the pieces of a huge duga puzzle at once and figure out how they fit together really quickly. It is especially good at solvent puzzles with words or pictures. Understanding which pieces are important and how they all connect to make up the whole picture. If you tell a story or show it a bunch of drawings, it can quickly make up a new story or picture that fits everything together perfectly. Since its introduction, the use of transformer models and image generation was largely experimental. Researchers were exploring how to adapt these models originally designed for texts, to handle visual data. Everything has changed in 2021, when a major tech player company called Open AI has introduced Dali, an AI system that can create realistic images and art from a description in natural language. Dali represented a significant leap in the capabilities of AI, for creative tasks. It demonstrated an unprecedented level of proficiency in generating diverse, complex and contextually accurate images from textual descriptions. This showcased the potential of AI in creative domains far beyond what was previously thought possible. Seeing the capabilities of Dali, developers and tech companies were inspired to explore similar technologies. This led to a wave in innovation and development in generator software, as more entities thought to leverage the underlying technology for various applications. The most prominent art generator tools that have emerged since the introduction of Dali include mid journey, stable diffusion, open art, firefly, Leonardo AI, and others. Around the same time when open y announced dal another major breakthrough in the image generation field happened with the introduction of diffusion models. These models showed impressive results in generating high quality detailed and coherent images, often rivaling or surpassing the quality of images generated by gans. Diffusion models unique process involves starting with an image that is entirely noise. Think of a TV static screen. Over multiple iterations, this noise is slowly reduced. With each step bringing the image closer to a realistic and coherent final picture. According to the information from Adobe, firefly uses diffusion models for its image generation features. Since our course is on Adobe firefly, we will speak more on how the diffusion model work in the next lecture. After watching the lecture, you will know exactly how the generator software works and how it translates a text prompt into the standing images. But before we go there, let's sum up what we've learned in this lesson. Generative Is history dates back to the 1960s, evolving significantly with Jeffrey Hinton's 2006 paper. The introduction of gans in 2014, marked a major advancement, shifting the focus from image recognition to generation. Gans user generator and discriminator network to improve image quality. Similar to a game between an artist and detective, transformer models introduced in 2017, revolutionized the field of machine learning and were applied to both language and visual data. Dal AI system that can create realistic images from textual descriptions showcase the potential of I in creative domains, far beyond what was previously thought possible diffusion models, another breakthrough in the image generation field, and those used in firefly generate images by refining them from noise to detailed pictures. That's it for the video and ICA in the next one. 26. How AI generates art: Everyone, welcome back. In this lecture, let's explore how AI R generation actually works. What happens after you enter a text prompt and at generate, and if and how can you influence the R generation process? Let's begin. So the R generation process consists of the following four steps. Let's talk about each step in more detail. Step one, data gathering and pre processing. The process begins with collecting a large dataset of images. Here we are talking about really large datasets comprising hundreds of millions of images, as well as the textual descriptions. The larger the dataset and the more diversities, like having images with a wide range of subjects, various artistic styles, lighting conditions and compositions. The more varied and nuanced the learning can be. And so as the final generated images. It's like giving an artist a broader range of experiences to draw inspiration from. Once the data is gathered, the images are then preprocessed to ensure consistency in the data fat into the model so that it can learn faster and more efficiently. Imagine if some images were blurry or oddly colored. Without preprocessing to normalize this, the model might learn incorrect or inconsistent visual patterns, like mistaken a blurred image for a style. Once the data is collected and preprocessed, we are ready to go ahead with the step two of the process, model training and latent space formation. While it is technically feasible for an IR generator tool to use multiple types of models like gans transformers, or diffusion models we've covered earlier. Such a system would be complex to implement. Most current tools tend to focus on a single model type. For example, Dali primarily uses a transformer model, and Firefly is a diffusion model. Whichever model is chosen, it undergoes a training process where it learns to interpret text proms and generate corresponding images. During training as the model learns from the data, it creates a multi dimensional latent space. This space abstractly represents the various features of the images, like style color or content in a lower dimensional format. It's gating technical. Let's take an example. Imagine we are training a model on a dataset of different animal pictures. During training, the model develops a multi dimensional latent space. Let's simplify this concept by imagining it as a big invisible map. Each dimension in this space represents different features of the animals. Think of one dimension for color like brown, white, black, another for size, small, medium, large, one for type mammal, bird reptile, and so on. Every point in this space is a combination of these features. One point might represent a small brown mammal like a squirrel, while another point could be a large white bird, like a swan. It's hard to visualize many dimensions, let's simplify it further. Picture a two dimensional grid. The horizontal axis represents size. Left is small, right, is large, and the vertical axis represents color. Bottom is dark and top is light. A point on this grid shows an animal size and color. When the I wants to create an image of a large, dark colored animal, it moves to a point on the grid that is towards the top right corner. Large and dark. This point in the latent space corresponds to the features of the animal, it will generate. The actual latent space is far more complex with many more dimensions than just two or three. Often in the range of hundreds or thousands. It's not something you can see with your eyes. It's more like a mathematical concept stored inside the machine I is running on. However, I asked firefight to visualize the multi dimensional latent space for me, and that's the picture that I got. Not bad. The AI learns to navigate this complex space during training. By moving around in this space, it can generate a wide variety of images. Now example, animal images, each with different combinations of features. After the second step is done, we are ready to go ahead with the fun part. Step three, generating art from a text prompt. The AI model uses your text prompt to navigate the latent space, finding points that correspond to the desired features or styles. Every model has its own unique mechanism for navigating the latent space and generating images. Since firefly is a diffusion model. Let's see how it works when it comes to generating images from text prompts. The model starts with a canvas of pure noise, a random arrangement of pixels. When given a prompt like a cat on a sofa, the model uses this input as a guide, setting the direction for the transformation process. The model then iteratively refines the canvas by navigating different regions or cardinates within the latent space and introducing and sharpening features relevant to the input prompt. This is what is called reverse diffusion or removing the noise from the initial canvas. As the noise diminishes, features start to emerge. Based on the prompt and its learned knowledge, the model begins to introduce elements like the shape of a cat, the texture of a sofa, so on and so forth. This stage represents a back and forth between the noisy and less noisy states where the model decides which features a align with the given prompt. The final image is a coherent and realistic depiction of a cat on a sofa. This image represents a specific combination of features encoded in the latent space, the size, color, and pose of the cat, the style and color of the sofa. All influenced by the initial prompt and the models training. Now you may have a very reasonable question. But how come the model decides on what's the size of the cat, its color or pose? These details come from the input prompt. In case you have a simple prompt like a cat on a sofa. The model might default the final image to more commonly seen or average representations of cats based on its training. So by adding more details to the front, for instance, including such descriptors as cute for a cat and antique, for a sofa, you are pushing the model to navigate through more specific regions of the latent space that correspond to these specific attributes. Cute might relate to certain aesthetic features of the cat while antique could involve particular styles or patterns associated with sofas. As a result, the model generates an image that not only includes a cat and a sofa, but also reflects the specified characteristics of being cute, and antique, this leads to a more nuanced and contextually rich image. Of course, there is also an element of randomness in how the model navigates the latent space, leading to creative and varied outputs. Using the same prompt might give you different images each time, because of the random way the model navigates its latent space. We are almost at the end of image generation process from the text proms. The final step includes post processing when we can enhance the image, by, for example, adding text adjustin contrast, or even adjusting parts of the composition. The number of steps of how you can alter the image depends on your creativity and the final result you want to get. Needless to say the post processing step is optional. You can use the image generated from I generator software without any modifications if you like it as it is, C that's it for the lecture, and as always, let's recap on what we've just learned. The IR generation process consists of four steps. The process begins with collecting and pre processing vast dataset of images and their descriptions to teach the AI. The next step includes model training and latent space formation, where this space abstractly represents various features of the image, such as style, color, and others. Step three involves generating art from a text prompt with the EI using prompts to navigate the latent space and generate images. Adding specific descriptors to prompts guides the EI to produce more detailed and contextually reach images, the final and optional step involves post processing activities, allowing further customization of the I generated images. At cool, Alsa in the next video. 27. Introducing Firefly Generative Fill functionality: Hello, everyone, and welcome to the new section of the course. Here, we will cover another fireflies functionality, generative feel. Unlike the text to image functionality that helps you generate brand new images from textual descriptions, generative feel works with existing images. It supports creative tasks like modifying images without the need for detailed graphic design skills, and it is created to be very intuitive and user friendly. In the upcoming videos, we will talk about the following use cases for using generative fiel, to add new objects to an existing image, to remove undesired content from images, to replace image backgrounds. These are the most common cases when generative feel is used, but we'll go beyond these three and also talk about such additional use cases as stylizing your existing image with digital effects as well as old photo restoration. The great news is that you don't need a separate account to access generative feel. All you have to do is sign into the Firefly web app with your existing Adobe account or the one you created at the beginning of the course. Once signed in, you can directly manipulate images by uploading them and specifying your editing needs through text. I hope that you are as excited about the functionality that we are about to explore as I am and let's get straight into the first video of this section. I'll see you there. 28. Adding new content to your image: One and welcome to the first video of the section where we learn about fireflies generative feel functionality. We'll begin by learning how to add new objects to your existing image. For the demo, I'll be using images that I've prepared in advance and saved to my local drive. In addition to this, you can also access generative feel directly from the text to image functionality with just one click as both of them are seamlessly integrated. So you can first generate an image with the text prompt and then get straight into generative feel to make any necessary adjustments. How convenient is that? Let's open firefly to see the generative feel in action. Let's open firefly at firefly dot.com and from here, go to Generative feel. To start working with generative feel, you can either upload your image from the local drive, or you can choose one from the FarflyGallery. For this demo, I'll upload an image from my local drive. I'll click upload and then choose an image from this demo folder. Here is my picture of a beautiful desert landscape. As we are experimenting with generating new objects, let's try to add skyscrapers on the background of this image. To add an object, select insert on the left hand side panel, and on the control tab below, select Add. Let's also change the size of our brush. For this, I'll click on settings, and you'll see this slider that you can move to modify the brush size. For my work, I will reduce the brush size to about 30%. You can also change the brush size by clicking on the right and left brackets on your keyboard. Let me show how it works. Right now, I'm clicking on the left bracket on my keyboard and you see that the brush size became very small. That's very convenient. I will use this method moving forward with other tutorials for generative feel. Let me adjust the brush size again and let's select an area where we want a new object to appear. I'll select rectangle and paint over the entire area. Let's also remove these trees from the background. Okay, I'm fine with this shape and let me write down my first prompt modern urban Skycrapers and I'll generate. We've got pretty decent results for the first group of images. I really like them and you see that the skyscrapers were placed exactly inside the area that we painted out. But for my example, I'd like the skyscrapers to be of different height. Let me click cancel and change the selected area. To modify your selection, you can click on subtract and then you can completely erase the selection or just the parts that you want to remove. Let me remove the entire area and painted over again. I'll click Add and draw more complicated pattern. And I'll paint over the entire area to show Firefly that I want my new objects to be placed inside the entire area. Okay, I'm happy with this prompt, so I leave it without any changes and I'll generate one more time. Very cool results. Now our C scape look more complex and sophisticated. I really like this first generation. But if you feel like you want more options from Firefly, you can click more to get other three images you can choose from. All images look great. I like this best, so I'll click keep to continue working with the same image. Now, let's add something on the foreground of our picture as well. For this, I'd like to increase the brush size for a bit. I'm pressing on the right bracket on my keyboard. Let me choose an area where I want a new object to appear. I'll choose this Rtgal brush it over. And I'll type my next prompt. Several camels are resting on the sand. Let's generate very nice results. Let me press cancel and add a few more descriptors into the prompt. I'll write with saddles and blankets in red white and green. I'll click Generate. Amazing results, isn't it? Let's click more to see if we can get even better variations from Firefly. Okay. I really like this striking contrast between the natural desert landscape and man made urban development. The pictures we've got are very realistic and I think they don't look like they were generated with AI. I like this image best, so I'll click on keep. Of course, you can continue including more objects if you feel like it or if you are done with your image, you can download it to your local drive. For this, click on download. We have a warning regarding content credentials that as always will be applied to our picture. I'll click continue and all done, we see that picture has been saved to our local drive. Let's also explore what other actions are available for you here in the generative feel. If I click on the three icon, we see that you have several more actions that you can do with this picture in Adob Express. We will explore all of these actions later on in the course. If I click on Share, I can continue working with the picture in Adob Express. And this is a new icon that replaced an old one, a laboratory flask icon that you've seen in the previous tutorials on text to image functionality. Yes, firefight team is actively working on their product. As you can see, they introduce some modifications to their product interface. Let's see what you can do when clicking on this icon. You can submit an idea or feature request to firefight team you can also report a bug or a problem when using Firefly, you can join Firefly community, rate firefly images or join the group. For this tutorial, I've prepared a couple of other images where I want to add some new objects with generative field. Let's do this. Boom. A. No, no, no. No, no. All right. That's it for the tutorial on how to add new content to existing images. I hope that you like the generative field functionality. Please let me know if you think it will be useful in your work process. I'll see you in the next video where we will talk about how to remove undesired content from your image. 29. Removing undesired elements on your image: Hello, everyone, and welcome to the new video where we talk about the generative field functionality of Adobe Firefly. In this video, we will learn how to remove undesired elements from your images. As always, you can work with the images you created upfront and saved to your local drive or those you just generated with the text to image function and want to modify before downloading to your computer. Let's open generative feel to go through the demo. As always, we start at firefly main page available at firefly.dot.com. From here we go to Generative feel. I have my images saved into my local drive, so I'll choose upload image and select the file that I've prepared for this demo. This is an image that I made during my trip to Cyprus. We see here a bland of urban landscape with historical architecture. I think that this picture will be so much better if we delete this modern building. Let's try to do this. To remove an object from an existing image with generative feel. Click on remove on the left hand side panel. Next, adjust your brush size. For this, you can use left and right brackets on your keyboard. If you don't like to use the keyboard for adjusting the brush size, you can always click on settings and then move the slider to the left or to the right to change the brush size. I'll reduce the brush size with my keyboard and let's select an object that we want to remove. So I'll paint over the entire building, as I want firefly to completely remove it from the picture. We don't need to provide any further instructions or prompts to firefly, as we already selected remove as an action that we want firefly to perform. All we have to do here is to click on remove button. As always, firefly gives us four variations of the images we can choose from. If we don't like any of them, we can always click on more to ask for three other variations. All results look nice. Notice how firefly looks at the things that surround the removed object and tries to replace it with something similar. I like this image, so I'll click on keep. We can continue working with this image by removing something else or inserting a new object. Or if we satisfy with the results, we can click on download to save an image to our local drive. I have another example that I'd like to show you for this demo. Let me open the generative field function from scratch and upload my second image. With this example, I want to illustrate a situation. I'm sure all of us are familiar with. Remember when you came to a beautiful place and wanted to make a picture of yourself. But unfortunately, many other people had exact same idea as you. So as a result, you'll have a picture of yourself surrounded by other tourists. Now you can adjust all of those pictures with generative field remove function. This example is more challenging than the first one that we've just seen as ideally, we want to remove just the figures of people, and we still want to see this metal walkway inside the picture. Let's see if we can do this. The first step I'll select remove on the left hand side panel and I will reduce the brush size to a minimum. Next, I will choose the first figure and brush it over. You don't need to be super precise when brushing over something that you'd like to remove. For my example, this loose rectangle will do its job. Let's click remove and see what firefly gives us. Very nice results. I like this variation best. You see that firefly not only removed the figure from this picture, but also added parts of the metal walkway so that it doesn't look damaged. If we want our walkway to be completely empty from other people, we can continue removing them from this picture. But before we proceed, let's click on keep to save the results from the first generation. This time, I will try to remove a group of people from the image. Let's see if we get good results from firefly. I brush over the entire square and click. Very nice. I think the first picture is the best, so I'll keep it and I'll continue with the next group of people. And here is our final result. I really like it. I could add some extra objects to this image, but I like it as it is. I'll click download to save it to my local computer. That's it for this tutorial, but we have a lot more to cover. In the next video, I'll show you how to remove or replace the background of your image. If you're interested in learning how to do this, I'll meet you in the next video. 30. Replacing image background: Hello, everybody, and welcome back. Let's continue our exploration of the generative feel function. Apart from generating new objects and removing undesired things from your image. You can also remove the image background and change the background. When I first tested the background removal with Jen fiel, I was surprised by how easy it is. I remember how time consuming the process was before. So you needed to use professional complex image editing tools like photoshop to get a decent quality. Hire someone to get the job for you. I used to test some online editing software as well that promised an easier background removal process, but I rarely got results of good quality. If you're like me, you will be impressed with what you're about to test. Let's open firefly to see the background removal in action. Let's open generative fiel function. From here, choose an image we are going to work with. I'd like to remove a background from this image. It's very easy to do. Just click on background and wait for firefly to do the job. Perfect. We can enlarge the image to see if there are any imperfections and there are parts that need to be removed. Everything is good if I'm satisfied with this image, I can either continuum replacing the background with the text prompt, or I can save the image to my local drive. You can also invert your selection by clicking the invert button. For my example, I want you to remove the image background, so I'll click on invert again to get back to the initial picture generated by firefly. Okay. Let's save the image to the local drive. Okay. And continue working with background removal using another example. I'd like to remove the background from this photo. As you remember, we've generated this picture in the advanced tutorials for text to image function. Let's remove the background and then generate the new background for this image. Let me show you how you can access generative feel from text to image functionality. I'll go to my favorites From here, I'll choose a picture that we've generated earlier with the text to image function. I click on the picture to go to the project details. Then I select a picture that I want to modify with generative fiel, go to ADD and choose generative feel option. From here, we can make any modifications that are available for us in generative feel function. Let's remove the image background. As the first step, click on the background removal button. If we enlarge this picture, we see that some parts of the background are still present in the picture, which may lead to some imperfections when generating the new background for this image. Let's try to remove those parts. For this, first of all, reduce the size of the brush to a minimum. I'm using left bracket on my keyboard. I also would like to increase the brush hardness. Okay. Brush Hardness makes the brush edges. So I'll increase it to a maximum. Click on the settings again and start removing the parts of the background. The adjusted image looks really nice. And as the final touch, let's add a new background to our image. I'll write historical architecture, buildings, blurry background, and I'll generate. Great results. I like this one, but if we need more variations, we can always click more to ask fire fly for other options for the same prompt. I like this one best, so I'll click keep and then download the picture to my local drive. That's it for the tutorial on how to remove and replace the background of your image and I'll seea in the next tutorial. 31. Stylizing your image with digital effects (part 1): Everyone, welcome back. We've already covered some of the core features of generative feel, including how to add and remove objects as well as how to replace the image background. But these are not all that generative feel is capable of. Another cool thing that you can do is to stylize your existing image into a digital illustration, allowing a significant amount of the original image or background to show through. So if you wonder what your portrait would look like as a three D cartoon character or as a digital watercolor illustration or minimalist digital line art. You can use generative feel to achieve just that. Let's dive in to figure out how to create the illustration. We begin from the far fly home page. From here, we go to generative feel function. I will upload an image that I would like to stylize for this demo and I will begin working with this image by setting the opacity level to about 15%. To set the opacity, click on settings. From here, adjust the opacity slider to about 15%. Opacity level allows us to adjust the visibility of the original content at 15% opacity. Much of the original image remains visible and the generator fill acts small like a nuanced modification. Lower opacity is great for subtle enhancements or when you want to gently merge i generated elements with your original image, and that's exactly what we are going to do in this tutorial. High opacity, on the other hand, is suited for bold changes or when the aim is to significantly alter or replace parts of the image with new content generated by the EI. Please note that the exact level of opacity is something that will vary from one image to another, and that's something you should experiment with. Try to set opacity at 50% and see what the results are going to be. Then gradually decrease the opacity level to find the number that gives you the best results. Next, I brush over the entire image to tell Firefly that I would like to modify the whole image and not just a part of it. To do this, I will enlarge the brush size by clicking on the right bracket on my keyboard and brush over the image. As a next step, let me write down my prompt. I will write three D cartoon character. Before we hit generate, let's also check the settings tab. I will click on the settings icon, and we see the settings tool bar on the right hand side of the screen. Here we have the preserved content settings that indicate how much of the original content will be kept in the new image. For our example, this setting doesn't make much sense. Given that we've set opacity to a low value, clearly indicating that we want a significant part of our image to remain visible. So if we set it to original, we won't see any changes made to the image. I've experimented with moving this slider gradually to the maximum while the opacity level was set at a lower volume. I love the images that I generated with the preserved content set to a maximum. But I also achieved some good results when the slider was in this zone. So for the first iteration, let's move the slider all the way to the right. Let's quickly speak about the guidance strength setting that affects how closely the I follows the prompt versus exploring its own creative interpretations. For this use case, when I'd like to stylize my existing image, I love to increase the guidance strengths to a maximum. That's what I'll do for the demo. But as usual, experiment with these settings as well to find out what works best for you. Let's move guidance strength to maximum to tell Firefly that we would like the generated content to be very close to our prompt. We are done with the settings and let's generate to see what we'll get. Okay, the first results look interesting, but I want to see the variations. Let's click more. For my work process, I have a rule that I generate no more than three times so that I have nine images in total I can choose from. This helps me to run many experiments when I play with the settings and and see what results are the best. Of course, the limits set on how many generations I do for one test, help me save my generative credits and my time as otherwise, both could run out very fast. I also recommend you set some rules and boundaries for your work process. All right, after several iterations, these are the pictures that I've got. I like this one the best. So I will click on Keep. Next, I'll click download to save this picture to my local drive. Okay. All right. That's it for the first part of the tutorial. Right now, you are all set to stylize an existing image on your own. I've also recorded a second optional part of the tutorial where you can see my experiments working with the same image and adjusting guidance strengths and pacity to different levels to see how they would influence the final image. I'll also make a demo for some other styles I've applied to my portrait to provide you with some ideas on how you can give your image a nice digital touch. So if you follo along, I'll see you in the next part of the tutorial. 32. Stylizing your image with digital effects (part 2): Everyone and welcome to the second optional part of the tutorial where we learn how to stylize your existing image. In this tutorial, I'll continue working with the same image and prompt. But I will modify the preserved content and guidance strength settings, as well as opacity to find out how the changes would influence the final image. Let's begin by modifying the preserved content settings while keeping the guidance strength at the same level. I'll drop the preserved content settings to about 75%. Guidance strength stays the same. Let's close the setting tab and check the opacity level. Currently, the brush opacity is set at 15%, I'll keep this and let's generate. And here are some pictures that we've got. I can clicking on more to see other variations. But since we are experimenting with the preserved content settings, I'll click on cancel and continue defining the preserved content settings. This time, I'll shift it to about 50%. All other settings remain the same. Let's generate and see what we'll get. D. So, as you can see, when I dropped the preserved content settings to a minimum Firefly didn't do any changes to the original picture. Let's click Cancel to go to the settings again. To sum up, we've got the best results when the preserved content settings was set above 50%. Personally, I like the results that we've got when it was set to a maximum. But let me know in the comments below what you think. Now let's experiment with the guidance strength. As you remember, this setting tells Firefly how closely the generated content needs to be kept to the prompt. Right now, we have guidance strengths set at the maximum level, meaning that the generated content will be to our prompt description. But let's see what will happen if we lower the guidance strengths to about 75%. I'll keep the opacity setting at 15%. I'll generate o. O. Let's return the guidance strengths to the original settings to ask fire fly to keep the generated content close to the prompt. The final experiments that I'd like to run is to change the brush opacity setting. Currently, we have brush opacity set at a low level. Meaning that we would expect just a slight subtle changes made for the original content. But what will happen if we would increase the opacity level? Let's find out. As the first experiment, I'll set the brush opacity to about 30%. I won't make any further medifications in the preserve content and guidance strength settings. And I'll hit January. Bo. M. 33. Old photos restoration: One. Welcome back. Let's explore another use case for generative feel. You might have an old phota that you want to retouch, and you can use generative feel for this purpose. If you're using photo shop, you have other tools at your disposal, such as the phtaRstoration neural filter that has recently been added to the product. So in that case, you can stick to the workflow you have and skip this tutorial. However, if you're like me and don't use photoshop, but still want to give a second life to the old photos you have. This video may come in handy. Let's dive into firefly generative field and see how it works. Okay. All right, let me open generative field function from the far fly home page. From here, as always, I will select an image that we are going to medify from my local drive. This is an image that we are going to restore. As you can see, the bottom part of the image is quite damaged, so we'll try to change that. Let's check that we have in stored selected, and function is active at the bottom toolbar. I will reduce the brush size for a bit by clicking on the left bracket on my keyboard and paint the rectangular area where I want firefly to make a restoration. When you make a selection, ensure that you include some parts where existing image is clearly visible. This is to give far fly some context of what I would expect to see on the new image. In my current example, I'd like to see the men's hands and part of the jacket that resembles the original one. Let me make the selection by taking some parts of the existing image as well. Something like this. I will brush over this entire area. I will enlarge the brush for a bit. All right, great. We don't need to provide any prompt here as we want firefly to fill in the space based on the surroundings. I will leave this field blank and click Generate. We've got some decent results. Of course, we can generate several more options to choose from. But before we do this, let me show you what will happen if I just highlight the missing part of the image only without selecting any overlap area with the visible part of the man's hands and jacket. I will click on cancel. Here, I click Clear to ask F fly to remove my selection. I will reduce the brush size and let me select just the missing part of the image with the minimum overlap with the visible part. Something like this, and let's generate. You notice the difference. For example, for this image, we see that far fly tries to restore the original frame of the image. These two generations look quite similar to the original one, so we don't see any improvements. Let me click on cancel and restore my original selection. I will brush over the existing part of the image as well. Let me remove this selection by clicking subtract. All right. Let's generate and see what firefly gets us. This image looks much better than original one. But I'd like to see other variations as well. I'll click more to ask Firefly to generate three more images I can choose from. All right. I like this image best, and it's really great that Firefly not only restored parts of the jacket, but also draw this frame very nicely. So please don't forget to brush over parts of the surrounding area of your image to give Firefly something to work with. I will keep this image by clicking on keep, and then I'll download it to my local. All right. Let's do two more examples of retouching old images. No, no. No, no, no, no. D. D. No. No, no, no, no. All right, that's it for this tutorial. Please let me know in the QN for this video if you find this possibility of restoring the image f, and I will see you in the next video. Okay. 34. Introducing Firefly-powered functionality in Adobe Express: One and welcome to the new section of the course, where we are going to talk about firefly powered capabilities that are not available directly in the Firefly web app, but rather accessible in another Adobe product. For those of you who are not familiar with Adobe Express, formerly known as Adobark, It is a tool from Adobe designed to create graphics, web pages, video stories, and other assets effortlessly. Express is created for users without professional design or video editing skills. It has an intuitive interface with a wide range of templates and design options that help easily create high quality visual content. Firefly and express are integrated seamlessly. You can first generate an image using that text to image functionality, and then proceed to add the image in express with just one click. You can apply extra filters and adjustments. Remove the background text or extra graphics to your firefly generated image with express. In addition to this, you can also access fireflies features, text to image, generative feel directly in express. This is the beauty of working with Adobe products as their generative capabilities are available for you in not just one product like firefly, but they spread across the entire family of products. Helping you to bring your complex projects to life with ease and convenience. However, please note that at the moment when I record this video, the latest firefly model model two is available only in firefly.dot.com in the text to image module. So other products, including express are using the previous model. You might expect different results when running the same prompts in far fly and other products. Let's continue with an overview of the generative AI capabilities of Adob Express. In this section of the course, we will explore the text effects feature that allows you to creatively enhance text by applying styles or textures through simple prompts. This functionality provides a straightforward process for generating kitchen text for various purposes, such as marketing materials, educational content, creative designs for websites and loggers and personal projects. There is one more feature that is not available directly in the Firefly web interface. That we're going to focus on. I'm talking about the text to template functionality that allows you to create your custom made unique templates, tailor it to your specific use case. For instance, you can build a template for an Instagram or Facebook post for flyers, posters, parts invitations, book covers, Linkedin profile covers, and so on. However, you don't need to spend time searching for relevant templates in the gallery of pre built templates or design it from scratch. All you have to do is describe what you are looking to create and then choose from the proposed options. I'm really excited to show you this functionality and how I used it for this course to create templates for the course handbook cover, which I hope you're already familiar with and some of the course resource files. We will begin with an overview of the express functionality and its generative AI features. This video is optional, feel free to skip it if you already know how express works. Next, we will have a tutorial where I cover in detail how to create templates using simple text rums, how to additionally modify them in express, including how to automatically translate your text to more than 30 different languages. Lastly, we will talk about how that text effects feature works. If you decide to follow along, I'll see you in the next video. 35. Adobe Express: Overview: Hello, and welcome to the first tutorial of this section. Here, I will give you a brief overview of what you can do with Adobe Express, as well as what firefly powered generative AI capabilities are available in this product. If you follow along, let's get started. Before we jump straight into the tutorial. Let's check if you have access to AdobiExpress. If you have a subscription to Adobe Cloud all apps, then you all set for this tutorial. However, if you have subscribed to the FireFly web app only, let me briefly walk you through the options you have to gain access to Express. Option one is to sign up for Creative Cloud, all apps plan that gives you access to more than 20 creative cloud apps, including Ado Express, premier pro that I will be using in the subsequent projects of the course, as well as other applications. Creative Cloud all App subscription comes with a seven day free trial period. You can sign up for the free trial first. Follow along the tutorials and do the course project and then decide if you would like to continue using Creative Cloud, all Apps. The second option you have is to sign up to Adobe Express application. The app comes with the following subscription options. I recommend you choosing the premium plan as it comes with a 30 day free trial period, and this is more than enough for you to not only follow along the tutorials and do your course project, but also experience the entire functionality of Adob press and decide if you would like to use this product further. The plan comes with 250 generative credits per month that you can use for generative AI tools available in express. I leave the links to both of these subscription plans in the resources section for the video so that you can study them and decide which plan you'd like to sign up. Now that we've settled the question of getting access to express. Let me briefly walk you through its interface and the different options on how you can start working with the product. If you are starting to work on a new project and would like to create a social media post, a presentation or a video, you can begin by clicking the plus button on the left hand side task panel. On the next screen, choose a document that you would like to create or select a custom size. Another option of what you can do here is to click on quick actions, browse through the list of actions, and choose one relevant to your current project. Notice that here, we have such generative AI actions as text to image and text effects. We'll talk about generative AI capabilities available in express a bit later. For now, let's return to create page and select Instagram square post. This will bring you to the editor where you start creating the post by adding pictures texts. Design elements and changing the background color of the post. M. Notice that every object you add to your design, whether it is text, image, video or something else, will be represented as an individual layer in the layer stack here on the right of the canvas. Layers allow you to establish a visual hierarchy and effortlessly adjust the position of individual objects within your design. For example, if you want to change the background, select the background layer, and then go to elements, backgrounds and choose one of the new backgrounds from the gallery. You can also bring a layer forward or ascend it backward by selecting the layer and moving it up or down. This adjusts the position of the corresponding object in your design. Let's look at an example. To move this rectangle backward, I will select the relevant layer and move it down beneath the photo. If you are happy with your design, click on download, select file format. And then click on Download. Apart from downloading the image, you can also share it directly in your social media. All right. Let's return to Express home page. Please note that all the assets that you create with Express will be saved to your staff folder. You can browse through your folders, as well as files and choose any of them to continue working on the document. I will return to the homepage Again. Now, let's explore other options on how you can start a new project with Firefly. If you prefer starting from a template, select an asset type that you want to create under what do you want to make? Let's choose a video for a change. Next, select one of the actions that you want to do with your existing video, or if you're creating a new video, select the video type. You can scroll horizontally to see the types of videos and types of actions that you can make in express. I would like to create an Instagram reel, so I'll choose browse templates. This brings you to the editor from the second tab, you can search for a template, you'd like to start from Click on the template you'd like to use for your video. This adds the template to your canvas. Now you can edit this video template or add additional scenes. Notice here a group of layers. A layer group helps tie together closely related design elements. Sometimes multiple objects in a template such as text and shapes coat in a group to make it more convenient to move them without disturbing their form. You can ungroup the layer by clicking on its image and then choosing group. You can create your own layer by holding down shift on your keyboard and selecting multiple objects. And then pressing the group button. Let's make some modifications to this group. You can also use the replace icon to swap out template videos for your own or choose another video from the gallery. As always, when you are done editing the video, click on download to save it to your local drive or click on share to share the video directly to your social media. Let's return to the express home page and explore what firefly powered generative AI capabilities are available in Express. For this, click on Generative I under what do you want to make? You will see a list of all the generative AI features currently available in express. You are already familiar with many of them as we've learned in detail how to work with text to image, generative feel in the previous sections of the course. There is also a new feature text to template that we are going to explore in detail in the upcoming tutorials of the course. Let's try out how the text to image function works in express. I'll type a simple prompt. And here are the results. I think you've noticed that the interface that we see here is not the same as the one we worked with in the Firefly web app. For instance, we cannot choose an aspect ratio, visual intensity, style strength, color and tone, as well as other settings. Apart from this, as I mentioned earlier, the latest firefly text to image model is not yet integrated into express. You might not get results of the same quality as if you would generate an image in the firefly web app. Given these limitations. In case I want to use a picture generated in Firefly in my work. I prefer to begin working in the Firefly WebApp and then continue editing the picture in Express. Let me show you how to do this. I will return to Firefly WebApp. Here is the picture that I would like to continue editing in Adobe Express. For this, click on Edit. From here, you can choose one of the editing actions from the list. I use express to add texts to my pictures most of the time. Since, as you remember, currently, Firefly is not good at generating texts, let's choose it. All right. We go back to Adobe Express, where we can continue editing this image. Let's explore what generative AI capabilities available for us to make the changes. I clicked on the image, and on the secondary tool bar, we see these three generative AI functions that we can use to make the changes. Let's try to use the generative feel first. I'll click on the generative feel card and then select an area that I'd like to change. D. The cupcakes look very yummy. I like this image best, so I'll click to keep the changes. Let's see what else we can do here. You can click on era to delete specific areas of your content. Please note that unlike generative field function, the erase feature allows you to just delete parts of the objects, so you cannot replace them with the new ones. I will click Cancel as I like my original image. Let's see what else we can do here. Another option available for you is to remove the image background in just one click. Let's test how it works. Decent results. So now I can change the background color. Alternatively, I can choose background from the elements. Let's also add text to this image. All right, cool. It was an overview of the key features available in Express. In case you'd like to further explore its functionality, I'll leave a link to the user guide from Adobe in the resources section of the video. I'll meet you in the next tutorial where we are going to talk about the text to template function in Express. I'll see you there. 36. Generating Social media editable Templates with simple prompts: One and welcome to the second tutorial where we explore and use firefly powered features available in Adobe Express. In this video, you will learn how to generate editable templates for social media, as well as digital posters, flyers, invitations, and more. Let's get started to create a template, click on Generative AI what do you want to make? Next, go to text to template. I'd like to create a template for an Instagram post to share information about the upcoming launch of a new course with my students. So I'll type the following prompt. Okay. Notice that you cannot choose the type of the image that you want to create on this screen. In my example, I'd like to create an Instagram post. You will be able to choose the image type only after you make the first generation. I think this is an oversight by the firefly team and I hope it will be fixed soon. Let's click on generate. Before we start browsing through the results, let me change the image type. Poster has been selected by default. However, I'd like to create an Instagram square post. I will choose this type from the drop down menu, and I click Generate one more time. Here are the results that we've got. Let's do several more iterations to see if we can get even better results. If you like a specific template, you can click on C variations to ask Express to generate four additional variations of that image. You can click on image that you like to drill down to the editor and continue modifying this image. For example, changing or including additional design elements. Alternatively, click on back to all results button. Okay, now you may probably have a reasonable question. What's the point of generating additional templates? If express comes with so many nicely designed professional templates you can choose from. Yes, you're right. You can definitely start with the template if you'd like to save time or need an initial reference for starting your creative design process. However, with the text to template functionality, you can create unique templates tailored to your specific project or use case in a matter of seconds. Let's say that I would like to announce the launch of a specific course on a specific date, and I need a template that I can start from So I will modify my prompt by adding additional information. Okay. You see how fast this process is right. I really love this idea of starting with the template highly personalized for my needs. It's amazing how Firefly picked up all the essential information from my prompting as well as a date of the. Of course, you can modify any of the templates, and we will see how to do this very soon. But before that, let's generate several other designs so that we can have more options to choose from. I usually do two additional iterations. For this, I will click on generate more results. Notice that you can rate every image by clicking on thumbs up or thumbs down icons to help the model to get better. All right. I like this image, so I'll click on it to drill down to the editor. You see here that the template is organized in a stack of layers. You can medify any of the layers here or add new elements. I find that sometimes it is useful to take ideas from several templates that I got from text to template into my final design. Let me do some modifications here. O. A. All right. Another cool thing you can do here is to translate your text into more than 30 different languages. Let's see how to do this. Click on translate and on the secondary tab, select one or several languages, you want your document to be translated to. You can choose up to 20 different languages for one translation. All right. I'm fine with this list, and let's click on duplicate and translate. Here we have the translated versions of the Instagram post. That's very convenient, especially when you work with international audience. Okay. Sometimes the translation is not perfectly accurate. Please keep this in mind. So you may still need to make slight medification of the text. Let's do one more example. This time, I would like to create a flyer, so I will select this document type. And let me also add a few more descriptors into the prompt. Okay. I. I like this image best, so I'll click on it to go to the editor to make further changes. Okay, let me do some modifications of the text as well as background image here. D. D. I can also include some assets I worked on before to this image. For this, click on your staff folder and browse the gallery of files and other design elements. Oh. For my example, I'd like to include a logo for the future diversity school. It is saved on my local drive. To add the file to my work, I will go to media and choose load from device. I will reduce the size of the image and place it into the bottom right side of the picture. When you are done editing the images, click on download to save them on your local drive. You need to choose a file format and click on download. In case if you'd like to post this image to your social media, click on Share and choose social media where you want this image to be published. All right. That's it for the tutorial on how to use the text to template feature in express. Now it is your turn to open Express and create your first template. As always, if you have any questions, just pause them in the Q&A section for the video, and I'll see in the next tutorials. 37. Generating eye-catching Text from simple prompts: Hello, everybody, and welcome back to the detailed tutorial on the Fireflies Text Effects feature. To access the text effects, open Adobe Express, and then go to generative AI. From here, scroll down to the text effects feature. You can start from scratch by describing the text effects you want to generate. Alternatively, select an image from the gallery and click on the image to go to the editor where you can make further changes. That's what I will do. Okay. Let me start by changing the text. I will double click on the text field, and I will write my new text, AI art. Next, I would like to change my prompt on the secondary tab. Go to the prompt field and type in your new prompt network of neurons. I will click Generate. Interesting first results. You can also add various style effects to your text. Let me choose neon and click Generate. Very creative results. I think I could use this text in the logo for this course. Adobe creates several variations of the text effect. Click on each image to see the style in more detail. In addition to style, you can also add tint to the color of your text. Let's see what other modifications you can do for the text using their secondary tab. You can customize the text by defining how tightly the design fits within the outline of the characters. A loose text effect will allow the effect to be very creative and expand more within the character bounds. Let's try this. I will choose loose and heat generate. Let's enlarge the text to see the effect in more detail. The tight effect does the opposite job. Let's try it out as well. I'll choose tight and click on Generate. For my work, I like the loose effect more, so I'll choose loose and click on generate one more time. Let's explore other settings available on the secondary tab. You can also change the font. Click on font and scroll down the font selection on the next screen. Let's try this one. Very nice. Apart from defying the font, you can also change the background color for your image. For this, click on the background color and choose the color that you like. Let's try black. And I will reduce the image size look very nice. You can also enlarge your text and change its alignment. All right. If you struggle to think of ideas for your prompt, look at the sample effects. Let's try them. I will change my text as well as background color. Next, I will click on and browse through the Gathery of sample effects. Let's choose this one and see what we get. If you'd like to have more variations, come back to the main menu, and from the results section, choose one of other options. And as always, you can change your prompt description. If you don't like any of these four variations, click Load More to get additional four options. Let's try a couple of more examples. B. Okay, we've just covered things that you can do with the text effects feature. However, if you'd like to continue working with this text, you can come back to the main editor screen by clicking on B. From here, you can change the text layout. As well as font font type and its size. And in case you'd like to come back to the text effects feature, click on the text effects tab at the bottom of the page. So as you can see, you have endless possibilities to play with text and various to get unexpected creative results. Please let me know in the Q&A section of the video, what you think about this feature and whether you plan to use it in your projects.