Photoshop AI - Generative Fill -Beginner to MasterClass

1. Course Introduction: Imagine the moment when you'll be able to fill the empty space in your photos with realistic and stunning details, all with just a single mouse click. When you can correct, edit, and transform your photos in mere seconds in ways you've never been able to before. When you can rejuvenate old photos with a single click, you can erase, recolor, or add objects to your image. After this course, your imagination will be the only limit to editing your photos because I'll provide you with all the knowledge and techniques you'll need and even more. Because I'm not just a graphic designer, I'm also an AI expert. I've been working with artificial intelligence and machine learning for years. This allows you to understand Photoshop, A I from a perspective you've never had the chance to before. It doesn't matter if you're an absolute beginner or have been using Photoshop for many years. In this course, we'll take it step by step so you can master the generative fill. If you just want to edit your family or community photos, then learning the introductory section of the course will suffice. But if you're committed and aim to become a Photoshop AI prompt engineer in the future, you can also study the master class section where you'll get to know the artificial intelligence behind generative. Phil, join us and be a part of this exciting journey. I look forward to seeing you in the course. 2. Starter Line – What is Generative Fill?: Welcome to the Photoshop I Generative Fill course. I'm glad you've joined us. What is generative fill? Generative fill is a technology that allows you to automatically fill in missing or unwanted parts of your image with content that matches the style and mood of the picture. Generative Phil can be used for various purposes such as image repair. If there's something on your image you'd like to remove like a stain, a crack, Noise or another object. Generative fill helps you restore the image to its original state. Image editing, if you want to change something in your image, like colors, lights, shadows or effects. Generative Phil helps you give your image a fresh look. Image creativity, if you want to create something new on your image, like a new object, a new character, or a new scene. Generative Phil helps you express your creativity. What makes generative Phil revolutionary is that it doesn't just simply copy and paste existing content into the picture. Instead it intelligently creates new content. Generative Phil utilizes AI and Deep Learning to recognize the features and style of the image producing content that matches it. It also takes into account the lighting and perspective of the image, ensuring that the new content appears natural and realistic. Additionally, it lets you quickly and easily repair, edit, or create new images without much time and effort. Generative Phil takes digital art and photography to a whole new level. Let's dive in, download the images needed for the lessons and then install the Photoshop beta. 3. How to Download Photoshop with Generative Fill: Hi there. In this video, we will install the Photoshop beta which includes the generative fill feature. The Photoshop generative fill is a new text to image tool currently only available in the Photoshop beta version. You can type a word or sentence and the artificial intelligence will create a digital image of what you've written. To download the Photoshop beta, we need to create an Adobe account, visit Adobe.com and click on Sign In located at the top right. Once the page loads, click on Create an Account Here. Apart from the traditional e mail registration, you also have the option to sign in with a Google account, Facebook account, or Apple ID. Create an account. Once you've created your account on Adobe.com navigate to the Creativity and Design tab in the top menu, and then click on What is Creative Cloud. As you scroll down this page, you will see the Adobe programs. You can choose one or subscribe to bundles. The main thing is that your chosen package includes Photoshop. I'm going for the first one. A Photoshop subscription is $20 per month, but if you're a student or a teacher, you get additional discounts. The Photoshop subscription comes with a seven day free trial. If you just want to try Photoshop for this course, you can do it for free. Just remember to click on this icon. Once you've selected the appropriate package, click Continue. Enter your e mail address and credit card details. After purchasing the package, you can download and install Creative Cloud from the main page. Once you've installed the Creative Cloud software on your computer, sign in in the top menu. Select Apps in the left side menu. Click on beta apps. Find Photoshop beta and click Install. Wait until the Photoshop beta installation is completed, then launch Photoshop. 4. Learn the Photoshop AI Generative Fill Taskbar: Once you've installed the Photoshop beta, we can start creating. But before we get started, let's familiarize ourselves with the new taskbar and its settings. First, let's create a new project. Click on the new file button in the pop up menu. I'm now going to choose the suggested custom document. In this lesson, it doesn't matter what size and property image we create. I'm going to make a slightly smaller image with a width of 964 pixels and a height of 543 pixels. For the color mode, I selected eight bit RGB color. But if you want, you can also make a black and white image. Once we have set the parameters, click on the Create button. Once we have created our document, we will see a white sheet in the lower task bar. There's an import image button with which we can insert a new image In the adjacent menu, we can position the task bar. We can hide or reset its position, or fix it with the pin bar position option. If this task bar disappears, or we accidentally hide it by pressing the hide bar button. Don't despair. In the drop down window menu above, click on the contextual Taskbar, and it will reappear by clicking on the last button of the Taskbar. It brings up the properties of our document, which we can adjust at any time as we wish. Click on the fourth icon from the top in the left hand menu and select the object selection tool. If you have a different icon there, right click on that icon and select it from the drop down menu. Now, using the left mouse button, select the entire screen. Once we have selected it, the task bar changes and will offer other options. Our first button will be the generative fill with which we can generate new things with the help of AI. The second drop down menu is the selection tool, where we can refine the selection. We can increase contract or even round its edges. The second is the Invent Selection, where we can reverse the selection. The third is the Transform selection, where we can reshape the selection as we like with the fourth. We can create a new mask with the fifth icon, a new adjustments layer. With the sixth, we can fill the selected area with color or gradient. The seventh option we already know with this we position the task bar and can hide it. Don't worry if you don't yet know the tools I just listed. We will talk about these in later videos. As the course progresses, you will gradually learn to use them and become a master. Now click on generative fill. In the now changed task bar, we received an empty text box where we can type in what we want to generate. The text entered here, we will call prompt. It's a command with which you can detail and specify to the artificial intelligence the AI what you want it to generate. Next to the text box is a generate button with which the AI executes the instruction given by you. We will not provide a now, so we will not write anything in the text. We don't give anything, and the selected area is empty, Then the AI generates an arbitrary image. Click on the Generate button. The process of generating the image has started. The process can take some time. It can be a few seconds, even a few minutes. We have to wait a bit until the green stripe reaches the end. Computer has finished generating the images. It generated three images by itself. Here on the side, in the Properties menu under Variations, we see the images where we can choose whichever one we like, whichever one we want to keep. If none of the three are to your liking, you can regenerate them as many times as you want. Let's generate three new ones. Click on the Generate button again, the newer variations will appear above the previous ones in the Variations menu. If you like one, just select it. We will see the selected image in the layers and we can start working with it. Once the images are generated, three new buttons appear on the taskbar. These buttons are used to send feedback to Photoshop about the generated image. You can like dislike or report the generated image. These buttons will probably disappear after Photoshop beta. That's the end of this lesson. I hope you enjoyed it. The next video will be much more exciting. See you next time. 5. Create new Images: Hi there. In this lesson, we will learn to write prompts and create an entirely new image. We won't go into the topic deeply yet. We're only learning the basics for now. Let's get started. Let's create a new document. Click on new file. Unlike the previous lesson, I will specify the width and height. I'll first change the unit to pixels and then set the width to 12 80 pixels and the height to 720 pixels. The resolution can remain at 72 DPI, just like the last lesson. In our case, it doesn't matter whether it's 72 or 300. However, the color mode should be set to RGB, color and eight bits. Once everything is set, click the Create button. We now have our white blank image. Select the object selection tool from the left and use it to select the screen. After that's done, click on generative Fill. We'll start by generating something very simple type house into the text box, then click the Generate button. We'll have to wait a bit for the computer to create the image. We now have three images, which you can see under the Variations tab in the Properties menu. All three images are good depict houses. But this isn't quite the house I wanted. I had in mind a White House on a hill. Let's modify our prompt and write White House on the Hill. After typing that, click the Generate button and wait a bit for the images to be produced. The images are ready. But it generated something entirely different from what I wanted. Artificial intelligence thought. I meant the presidential residence in the United States known as the White House. To get what we want from AI, we must be much more specific and accurate with our prompt. If we don't want the AI to generate the American White House, we need to expand our prompt, right? Switzerland and little at the beginning of our prompt. Then click on Generate. We have received our three new images. This is exactly the picture I wanted. They are perfect. All we did was add one attribute to our prompt, Switzerland. With this new information, we completely ruled out the White House in the United States and got what we wanted. But what if I wanted much smaller details. I envisioned a river in front of the house, cows grazing in the field and birds in the sky. Let's type that in and see what image it will generate. I will extend the prompt with cows in the pasture, birds in the sky River in the middle of the picture. Cows in the pasture, birds in the sky River in the middle of the picture. Then I'll click on Generate. We wait a bit again, the images are done, but sadly they turned out to be terrible. The first image doesn't have the house on a hill, the cows are distorted and there's not a single bird in the sky. In the second image, the White House looks more transparent than white. The third image has deformed cows, strange houses, and no birds in the sky. But the current Photoshop, A I text interpretation is very weak. And it works best with prompts shorter than eight words. If we want to enhance our image with cows, a river, or birds, we'll need to use a completely different technique. But I'll show that in future lessons, this is just the surface of prompt writing. In later lessons, we will delve into it and you can understand how an A I like this works and how to write prompts at an engineering level. But first, let's practice the basics. I hope you enjoyed the lesson. See you next time. 6. Selecting Tools: Before we proceed in our journey to get to know Photoshop AI, let's familiarize ourselves with the selective tools. These tools will help us specify the area of the image we want to work with, making their understanding crucial. We won't be creating a new image. Let's open the image number one which we downloaded at the beginning of the course. Click on the Open and find the numbered one image in the Res folder. Once the image is loaded, let's start getting to know the selecting tools by right clicking on the second icon. From the top. We will find the first selection tool, the rectangular marquee tool. This tool allows us to make rectangular selections. Let's select a square in the sky and then click on Generative Fill and type in cloud for the prompt. After typing, click on the Generate button. Now the A, I will generate a cloud within the selected square. It's done. The first version turned out perfect. Let's now switch to the elliptical marquee tool. This selection tool allows us to make circular or elliptical selections. Let's select a portion of the sky. Then click on Generative Fill and type in cloud For the prompt, after typing, click on the Generate button, it's done. I like the third version the most. We go back to the second icon from the top and right click, we will see two more selection options. Single row and single column. These only select a single row of pixels. We won't be using this selection tool, so I won't demonstrate them. Let's now go one icon down and right click on it. Select the Lasso Tool. This tool allows us to freely select an area as if we were drawing. Let's select a portion of the sky. Then click on Generative Fill and type in cloud for the prompt. After typing, click on the Generate button. We will be using this selection tool quite frequently, but don't be afraid of it. It's very easy to use. I finish the images, the first one turned out great. Let's go back to the icon and select the polygonal lasso tool. It works similarly to the previous one, but here you can make selections with straight lines. Let's try it out, Draw a shape in the sky, then click on Generative Fill and type in cloud. For the prompt, after typing, click on the Generate button. The images are ready. Choose the cloud you like the most. Let's go back to the icon and select the magnetic lasso tool. With this tool, we can select the edges of an object. Let's try it on a cloud and attempt to select it. Once we have, click on generative fill and type in dark cloud for the prompt after typing, click on the Generate button. This selection allows for much more accurate selections. The three images are ready. The shape of the cloud didn't change, but it got darker. Now let's select the fourth icon from the top on the left side. Within that, the object selection tool, this will be another selection tool that we will use quite often. With this tool, Photoshop automatically recognizes and selects objects in the image on the right side, select layers at the bottom, then scroll down and left click on the background layer. We need this because we created a new layer with each generated cloud, and now we want to select objects in the original image. Once selected, hover over the tree on the left side of the image. When you move the mouse over it, the edges will glow in pink. Left click on it and Photoshop will select the entire tree. Once we have click on generative Fill and type in apple tree for the prompt after typing, click on the Generate button. Once it finishes generating, it replaces the entire tree with an apple tree. It's not perfect, but remember we can generate new images as long as we want. Our next tool is the Quick selection tool. This tool is similar to the magnetic selection tool as it also finds edges. Let's select it before we start working with it. Right click on the background in the layer section and merge the base image and all AI generated layers into a single new image. Click on Merge Visible. Once done with that, select one of the Clouds. If we accidentally selected too much, choose the icon with the minus sign in the menu above. And we can erase parts of the selection. Once we have selected it, click on Generative fill and type in dark cloud for the prompt. After typing, click on the Generate button. It has transformed our cloud. Our last selection tool will be the magic wand tool. With this tool, we can make selections based on color. If we click on the blue sky, it will select the entire blue sky for us. What's important in this selection is setting the tolerance. With this, we set how tolerant the selected color will be. Although it may not seem obvious at first, there are multiple shades of blue in the sky. If we set the color tolerance to 80, it will select the entire sky. However, if we set it to only two, it will select a very small portion because the rest of the sky is either lighter or darker blue than the color we clicked on. With this tool, we have finished the selecting tools. We are also done with the basics and the boring part. We will continue working with the original image used in this lesson. If you save the image we created now, please do it under a different name. I hope you enjoyed the lesson. See you next time. 7. Introducing the Beginner Class: Welcome to the Photoshop. A I beginner class. I'm thrilled that you've joined us on this course. We will dive deep into the world of generative fill in Photoshop. If you're new to Photoshop, don't worry. The beginner class material is specifically designed for you. In this course, content, we'll learn how to generate backgrounds, enhance and correct images, remove and replace objects, create realistic objects, adjust the colors of items, produce panoramic photos, and finally, craft a unique image by merging multiple photos. By the end of the beginner class, you'll be confidently creating and editing photos with generative Phil. I promise it will be an exciting journey exploring the possibilities within Photoshop and generative. Phil, let's get started. 8. Generate new Background: From our previous lesson, we learned the basics. Now I'll show you how to apply them. In this lesson, we'll create a new background for an image. Click on open and open the image file numbered one located in the Res folder. This is the same image we worked with in the last lesson. We'll be transforming the blue background of this image using artificial intelligence. From the left side menu, select the Object selection tool. Hover over the blue sky with your mouse and click on it with the left button. Having selected the background, now click on Generative Fill in the prompt box. Type, mountains in the background, mountains in the background. Once you've typed it in, click on Generate. We have to wait a bit. Here are the images I personally like the third one the most. If we look closely at the images, there are a few minor flaws. By turning off the back ground layer, we can see that it didn't only affect our selected area, but also altered the house in the image. This caused certain parts of the house to change. We can fix these parts, but in upcoming Photoshop versions, the AI's object detection is likely to improve eliminating such errors. If you're interested in learning how to correct these, you can find a lesson on this topic under the Practice menu in the course content. Let's further enhance our background with clouds. Click on the blue area above the mountains and then click on Generative Fill for the prompt, type in Clouds, and then click Generate again. We'll have to wait for the three images to be generated. The AI has finished generating the images. The first picture was perfect, comparing our final image to the original. The transformation is truly astounding the style of the generated background matches the base image. If we overlook the minor issues for now, we've achieved a perfect picture. That's how simply we can change or modify a background. I hope you enjoyed the lesson. See you next time. 9. Extend Images with Generative Expand: Hello there. In today's lesson, we'll learn how to expand images using the generative expand feature. This is an incredibly useful tool that you can frequently use even for family photos. Let's get started with the lesson. Click on Open, Open the image labeled as number two in the Res folder. We'll be transforming this image into a larger version with a 16 to nine aspect ratio. Select the fifth icon from the top. If you have a different icon in the fifth position, right click on it and choose the crop tool. I'll zoom out a bit so we can see the image better. The most crucial setting in the Croup tool is defining the aspect ratio. Set it to 169. Here, drag the white edges of the frame as shown in the video to expand our workspace. Also ensure that in the film menu, generative expand is selected. Once you've set everything up and enlarged the workspace or the image size, click on the Generate button. We'll have to wait a bit for the AI to generate the images. They are done. Let's see how they turned out. All three images are astonishingly good. We didn't input any prompts for these images. We left it to the AI to recognize and generate the surroundings. Now let's try typing Forest into the prompt field. After entering it, click on the Generate button, wait for the AI to generate the images. They are done and the results are amazing. The A, I recognized the original photo as a beach scene, but due to the prompt we entered it, added forests in the background. There are a couple of minor errors in the images, but they can easily be corrected. However, by the time you start learning this tutorial, the AI might not even make those small mistakes. Now without saving, close the image in the main menu. Click on Open again. This time open the image labeled as number three in the Res folder. It contains a small forest and a lake. We'll see how it can enhance a landscape image with our previous image. The task was simpler as the background was blurred and only the girl in the foreground was sharp. But in this image, every detail is crisp and clear. Select the fifth icon from the top. The croup tool which we used for the previous image. Similarly, pay attention to our aspect ratio, which in this case is 169, and then expand the size of our image as shown in the video. Once done, type in forest and lake. For the prompt, after entering, click on the Generate button and wait. Ai has generated the three new images. They are phenomenal, the enhancement is sharp, and the style remains consistent. The AI did an outstanding job. Consider that we're only at the beginning stages of this technology and it can already made images with such good quality. There are many more exciting lessons ahead. I hope you enjoyed today's content and I'll see you next time. 10. Remove Objects: Hi there. In this lesson, we'll learn how to remove objects from photos. This can be particularly helpful when improving or retouching photos meant for family albums or social platforms. Let's get started. Click on Open and within the Res folder, choose the image labeled six. We'll begin with this beach side photo. First we're going to remove the tablet lying on the blanket. Click on the zooming tool and magnify the image. Then using the lasso tool, draw around the tablet. Once done, click on generative fill. Don't input anything for the prompt, simply click on Generate. We'll have to wait a bit. As the AI generate the images the AI has finished generating, the second image looks the best. Now I'll zoom out using the zooming tool to view the photo in its original size. The edit turned out great. Let's remove the sunglasses as well. From the left menu, select Lasso Tool and encircle the sunglasses. Once done, click on Generative Fill. You don't need to type anything in the prompt field, just click on Generate and wait for the AI to finish. First one looks perfect. Let's zoom out and watch the whole picture. But what if we want to remove something bigger like the girl in the photo? Let's see if the A, I can handle it using the lasso tool draw around the girl for accuracy. Try to zoom in on the image. It will make the task easier. Once you've outlined the girl, click on generative fill and without any prompt, hit the Generate button. Let's check the generated images. The first image didn't work out well, the second one isn't as expected either. However, the third image is fantastic. The girl was removed flawlessly from the photo. The AI is even capable of removing objects or people from more complex images. Close this file, then click on open, open the five labeled image in the Res folder. This is a sharp high quality photo of two young girls. Let's remove the girl on the left from the photo. Select the lasso tool and start outlining the dark haired girl on the left. Try to be as precise as possible. Again, zooming in will help. Once done with the outline, click on Generative fill, and without any prompt, press Generate. We just need to wait for the AI to generate the images. It'll have a tougher chal***ge compared to the previous image the AI finished generating the three images. The results are astonishing. It's almost impossible to tell. The photo was edited. The girl took up a significant portion of the image and there was a metal railing behind her. Nevertheless, she was removed perfectly. If you want to remove something from an image, all you have to do, draw around it and generate with generative fill without any prompt. If the first three attempts don't succeed, don't despair. You can generate as many times as you wish. I hope you enjoyed this lesson. See you next time. 11. Replace Objects: Hi there. In this lesson, we will learn how to replace objects within a photo. Let's get started. Click on Open in the Res folder. Look for the image labeled four where there is a brunette woman holding an ice cream. We will be modifying this image first, I'll replace the vase. Zoom in a bit to make it easier to select. Next, choose the lasso tool and start outlining the vase. Always visualize what you want to replace the object with. I want flowers on the table. When I outline, I leave more space at the top to ensure a bouquet can fit is crucial because the AI will work with the space you designate. If you allot too little space, an entire bouquet might not fit. Once you've finished outlining, click on generative fill in the prompt field type flowers. After entering it, click the Generate button. Let's see what images the AI has generated for us. The second and third ones are perfect. The AI did an excel***t job. However, I now want candies instead of flowers. Delete the generative layer we just created by right clicking. And select delete layer After removing it. Re, select the vase on the table using the lasso tool. But this time don't leave as much space on top as before. Adhere more closely to the vases contours. Otherwise, if you give it too much room, the AI might generate enormous candies and lollipops for us. After outlining, click generative fill in the prompt field type candies. Once entered, click Generate. Let's examine the generated images. They're perfect, although they came out a bit larger than expected. In such cases, it's simpler first to erase the vase from the table and then generate candies onto it. Now let's replace the ice cream with a microphone. Click back on the background layer and start outlining the ice cream using the lasso tool. Since a microphone is roughly the same size and shape as the ice cream, try to stick closely to the ice creams contours. Once outlined, click on Generative Fill in the Prompt field type microphone. After typing, click Generate. Let's see how the replacement turned out. The images are fantastic. Typically, the AI struggles with hands and fingers, but all three images worked out the first time. The first image does have an extra ring, we can easily correct that. Now let's try something more chal***ging. Close this image in the menu, click on Open. Then in the Res folder, select the image labeled two. In this photo, we will replace the swimsuit with a wedding dress. Select the lasso tool and start outlining. While outlining, visualize the girl wearing a wedding dress. Consider how it would look on her and the space it would occupy in the photo. Once you're done, click Generative Fill in the prompt Field type wedding dress. After typing, click Generate. The AI has generated the images. All three are outstanding. There are some errors though. The AI struggled with the girl's right hand, but we can easily correct it later. Always remember when replacing objects ensure the designated area can accommodate the new object. If the new object is smaller, it might be best first to erase the original from the photo and only then generate the new one. In the practice materials, you'll find several videos where I completely transform an image utilizing all of Photoshops functions. If you want to practice, check out the videos under the practice tab. I hope you found this lesson helpful. See you next time. 12. Generate Realistic Objects: Hi there. In this tutorial, we'll be enhancing our images by adding new realistic objects. Let's dive right into the lesson. Click on Open in the Res folder. Select image number seven. We're going to enhance this pub scene with some realistic objects. First, we'll generate some food and drinks on the table from the left side menu, choose the lasso tool, use it to circle a large area above the table. Once done, click on Generative Fill in the prompt box, type Soda and Breakfast. After typing, hit the Generate button. Let's check out the images the A I has created for us. They look good. I especially like the third one, the wall looks a bit bare. Let's add a painting. I'm imagining a rectangular painting, so we won't use the lasso tool this time. From the left side menu, select the rectangular marquee tool and highlight a large section on the wall. Once that's done, click on Generative fill in the prompt box type paint. After typing, press the Generate button. Let's see the images the AI produced. Only the second one had a minor flaw, but the other two turned out great. One of the strengths of Photoshop AI is its ability to accurately gauge the light sources in the image and generate objects that don't look out of place. Let's give the AI a more chal***ging task, adding a cat to the image. I'll stick with the rectangular marquee tool and mark an area on the floor. After marking it, I click on Generative Fill and type cat In the prompt, once you've typed it, press the Generate button, the images are ready. Let's look at the cats. They're not perfect. This is a tougher chal***ge for the AI. Let's generate three more and see if we have better luck. The next three are done. The first one turned out pretty well. Generating living creatures is still a weak spot for Photoshop AI. But as it's further developed, it will get better month by month. Now let's try something entirely different. Close this image in the menu. Click on Open in the Res folder. Select image number four. We'll generate a diamond necklace for this girl holding an ice cream to make the selection easier. Zoom in on the image in the left side menu. Choose the lasso tool and draw around the girl's neck. Imagining the shape of a necklace. Feel free to get creative with the shape, it's up to you. Once done, click on generative fill in the prompt box type diamond necklace. After typing, hit the Generate button, the images are ready. Let's take a look. All three necklaces turned out great. Choose the one you like best. Then use the zoom tool to see how it looks at its original size on the image. I think it turned out really well with Photoshop A I, you can add virtually any object to your image. The AI takes into account the quality of the picture, the lighting, and the colors, and can create entirely realistic objects. It's only limited by our imagination in terms of how we modify and what we add to our images. I hope you enjoyed the tutorial. See you next time. 13. Change Colours: Hi there. In today's lesson, we will learn how to change the color of an object in Photoshop. Let's get started with the lesson. Click on Open in the Res folder. Select the image numbered four. This image features a girl with brown hair holding an ice cream. First, we'll change the color of the ice cream to make the selection easier. Use the zoom tool to zoom in on the image. Once zoomed in sufficiently start outlining the ice cream using the lasso tool. After you're done, click on the generative fill button in the prompt box type red ice cream. Once you've typed that, click on the Generate button, Photoshops. Ai has advanced object detection. Most times just typing red would suffice to generate a red ice cream. However, more precise descriptions yield better results. The AI has finished generating the images. All three outcomes are great, but the third one is the best. We now have a lifelike red ice cream. Now let's color the girl's lips red. This will be a bit harder to select. You are advised to zoom in even more. Use the lasso tool to outline her lips in the same manner. Once done, click on the generative fill button in the prompt box type red lips. After typing that, click on the Generate button. When working with such details, the selection is crucial. If the selection is too small, the AI will generate smaller lips. If too large than larger lips, the AI has finished its task. Again, the recoloring looks great. Let's continue practicing. We'll give the eyes a new color outline one eye with the lasso tool, hold down the left shift key on your keyboard, and outline the other eye. After this, click on the generative fill button in the prompt box type C, Blue eyes then click on the Generate button. The outcome is once again fantastic. The brown eyes have transformed into beautiful sea blue eyes. Let's zoom out to see the whole picture. The results are quite impressive. Now let's chal***ge the AI a bit more. Zoom in slightly and start outlining the girl's hair using the lasso tool, leaving some space between the selection and the actual hair. When you are done with the selection, click on the generative fill button in the prompt box type blonde hair. After typing, click on the Generate button. The AI has finished generating the images. These blond hairs look amazing. Personally, I prefer the second one the most. Zoom out to view the whole image. It's amazing to see the transformation. Perhaps she would look even better with red hair. In the prompt box, rewrite blond hair to red hair. After making the change, click on Generate. The images are now ready. Needless to say, the outcome is outstanding. The red hair truly suits the girl better. Selection plays a key role when recoloring objects. I personally prefer using the lasso tool and allowing some space for the objects. However, if you don't want to manually select using the lasso tool, you can use the object selection tool. Choose it from the left panel, move the mouse cursor to the top left corner of the ice cream. Hold down the left mouse button and drag the selection across the entire ice cream. The AI will automatically detect and select the ice cream. This method works for simpler objects, but not working for complex ones like hair. I prefer to manually make the selections with the lasso tool because it lets me control the maximum area The A I should consider for each part. Proper use of the lasso tool can yield better results during image generation. I hope you found this lesson helpful and enjoyed it. See you next time. 14. Make Panorama Picture: There In today's lesson, we will be creating a panoramic image using generative. Phil, let's dive right into the lesson. Click on Open in the Res folder. Select image number one. This image features a large white house. Since the background is quite empty, let's generate mountains. Using AI, select the Object Selection tool from the left hand menu. Click on the blue background. After it has selected the blue backdrop, click on Generative Fill in the prompt field, Type in Mountains. Once you've typed that in, click on Generate. The mountains are generated. Pick the mountain that appeals to you the most. This mountains makes our background less barren. As we're creating a panoramic image, we need to increase our image width. In the top menu, click on Image from the drop down. Select Canvas Size in the pop up window. Increase the width. I'm setting mine to 2,600 pixels. But if you're going for a real panoramic shot, you can set it to 3,750 pixels. Note that this will take much longer to generate images after entering our desired width. Click on the okay. Our image is now much wider to see the whole image. Use the zoom tool to zoom out. For easier layer management, it's best to merge them in the layer's menu. Right click on background from the pop up. Choose the merge visible option. Now we have just one layer. On the right side of the background layer, you see a padlock. Click on it to edit the layer. If you do not click on it, you will not be able to edit this layer from the left menu, select the rectangular marquee tool, then select the right edge of the house. After selecting, as shown in the video, click on the topmost icon in the left menu and pick the moving tool. Drag your selected area over to the left side of our image by holding down the left mouse button. Once done, again, choose the rectangular marquee tool from the left menu. Now select the remaining part of the house. Once done from the left menu, choose the moving tool again with the left mouse button held down. Drag the selected area to the right side of our image. Be careful not to move the selected area up or down. If you hold down shift while dragging with the left mouse button, it will move the element only along one axis. This can help in achieving precise movement. After aligning the edges precisely, again, select the rectangular marquee tool from the left menu with it, select the empty space between the two images. It's a good idea to overlap a few pixels into the images. No space is left unfilled after generation. Once selected, click on generative fill this time, don't type anything in the prompt, leave the AI's image recognition ability to fill in the gap. Click on Generate. The AI has generated the images. None of the generated images are quite to my liking. The AI doesn't always do a great job at generating houses. Let's try generating a forest instead, in the prompt field type in forest and mountains. Once done, click on Generate. The new images are a bit better. Of course, you can keep generating images until you get a perfect one. This isn't the simplest method for creating panoramic images, but it's essential for you to learn how to move and re, position certain parts of your photos. Now let's close this image. Click on Open in the Res folder, select the image number eight. The image depicts a forested beach. We need to increase the width of the image in the top menu. Click on Image from the dropdown, Select Canvas Size in the pop up window. Increase the width. I set it to 3,750 pixels wide, but feel free to set it to an even larger width. We won't be selecting and sliding the edges of the image this time. Instead we'll opt for a much simpler solution. Go to the filter menu at the top. In the dropdown menu, click on Other and then on offset, the pop up window, you can easily slide the image both to the right and the left. Slide the image horizontally by roughly 1,000 pixels. Then click Okay. Next from the left menu, click on the rectangular marquee tool. Use it to select the white empty space, ensuring the selection slightly overlaps into the image. Once selected, click on generative fill in the prompt field. Don't write anything this time. And click on the Generate button. The AI has generated the images. They look good, but let's verify. First, merge the layers. Right click on the background layer in the pop up menu. Click on Merge Visible. This allows us to move the image as a single entity. If we don't do this, the background and our AI generated layer will move separately under the filter menu. Click on Other and then on offset in the pop up window. No matter where we move the image, it remains seamless. It's that simple to create a panoramic image. You can even view our panoramic image on Renders Stuff.com To do this, save our image in the File menu. Click on Save As and save the image in Jpeg format. I saved mine as test. Once saved, open your browser and go to Renders Stuff.com on the site. Click on on line 360, Panorama Viewer. Upload our image by clicking on the upload image from file button. That's how our image looks in panorama view. To achieve a better resolution, it's recommended to create an even larger image. I hope you enjoyed the lesson and I'll see you next time. 15. Fixing Pictures: Hi there. In today's lesson, we'll be fixing and retouching images using generative fill. Let's dive into the lesson. Click on Open in the Res folder, Select number ten. On this image, we see a very common mistake on the left side. A small part of my finger is visible in the photo. Such image flaws can be corrected with the generative fill in a flash from the left hand menu, choose the lasso tool and encircle the finger. Once you've done that, click on Generative Fill, leave the prompt field blank and click Generate. The AI did a great job fixing the flaw. You can't even tell the image was edited. Let's now fix another common mistake. Close this image in the menu. Click on the open button in the Res folder, select image number 11. Blurred objects are also common mistakes in photos. This too can be easily corrected from the left hand menu. Choose the lasso tool and encircle the wooden spoon. Once you've done that, click on Generative Fill, leave the prompt field blank and click Generate. The blurred spoon is gone. The A. I did an impressive job again, but there's another way to fix this image. Right click on the generative layer in the pop up menu. Click on Delete layer using the Lasso tool. Select the spoon again. Once selected, click on Generative Fill in the prompt field type wooden spoon. After typing, click Generate the AI generated Beautiful Spoons. For us, the image looks so much better than with a blurred spoon. Close in the menu. Click on Open in the Res folder, select number nine. It's almost a given nowadays that people only post perfect photos on social media. This would be a perfect photo, if not for that pesky finger. If we just want a minor touch up, let's remove the white discoloration from the nail. Select the lasso tool and circle it. Once done, click on Generative Fill, leave the prompt field blank, and click Generate. At the end of the image generation, we received an error. The generated images were removed because they violate guidelines. Of course, we didn't violate guidelines, We just wanted to remove the white discoloration indicating vitamin deficiency from our nail. This part of Photoshop isn't quite reliable yet, but we can easily outsmart the AI. Click on generative fill again in the prompt field, type a dot and click Generate. This prompt worked without triggering the error. Over time such errors won't occur, but it's expected that in the next update, writing a.in the prompt field won't work. If you want, we can also replace the entire finger using the lasso tool. Circle the finger. Click on generative fill again in the prompt field, type a dot and click Generate with the woman's finger prompt. I received an error. I use the dot prompt, but by the time you learn this lesson, you might be able to type woman's finger without error. Our generated images are complete. The entire photo looks so much better with the corrected finger. With generative fill, we can easily correct nearly any flaw in a photo. For fixing these flaws, use the knowledge you've gained so far and your creativity. I hope you enjoyed the lesson. See you next time. 16. Create a Unique Social Media Image by Merging Photos: I there. In today's lesson, we'll create a unique image by b***ding two photos together. Let's get started. Click on Open in the Res folder, Select the image numbered 16, and while holding down the left control key, also click on image number 17. After selecting both images, click open. This way we will open two images simultaneously. We'll enhance this photo and superimpose a bear in the background. Let's start editing. Switch to image number 16. Comparing the two photos, we can see that the bear's background is too dark, contrasting heavily with our main image where a man is squatting. We need to adjust the background of the bear image. First, from the left side menu, select the Object selection tool and click on the background. This should select most of the image except for the bear. Click on Generative Fill in the prompt Type forest, and then click Generate. We might have gotten better results with autumn pine forest, but this could still turn out fine once the generation is done. Let's see the images produced by the AI. They aren't perfect, but I think the second image will do. Let's merge the layers. Right click on the background layer, then use the merge visible command. After this, we can copy the bear to the other image. Choose the rectangular marquee tool. Select the bear with a larger selection. Then go to the top Edit menu and select Copy from the dropdown. Now switch back to image number 17. Let's Paste the bear. Click on the edit menu at the top and choose Paste. To fit the bear in the image, we need to expand its width. On the right, click on the lock icon next to background layer To make our primary layer editable. To view our main image better, temporarily hide the bear's layer layer one. Once hidden, click on image in the top menu. Then canvas size in the pop up window. Set the width to 2,400 pixels or larger. And confirm with, okay, I'll place the bear on the right side. So let's move the man to the left side. Select the move tool and start dragging the image to the left. That looks good, enable visibility for the bare layer layer one and click on it to edit. Before we start editing, re, arrange the layers so the bear isn't in the foreground. It's good to get used to ordering layers based on their spatial arrangement. Drag layer one below layer zero using the left mouse button from the left menu. Click on the move tool and adjust the bears position, ensuring layer one is selected at the bottom, reduce the size of the bear. Click on Edit in the top menu. Then select Scale under Transform. Adjust the bear's size by clicking on the image corners. Begin removing overlapping parts from the images. Select the eraser tool and choose layer zero for editing. I'm increasing the eraser tools size to 105 pixels and starting to erase. Switch to layer one and continue erasing, ensuring we don't erase any part of the bear. The goal is to leave some space between the two layers. Helping the A, I generate a seamless transition. I'll slightly reduce the bear's size again. Click on Edit in the top menu, then scale under Transform. Adjust as desired. Now that I've resized it, I see a small portion left that needs erasing. Use the eraser tool to correct any imperfections. When corrected, all the errors in enough space exists between the two images. Merge the layers, right? Click on layer on E. Then choose Merge Visible from the top right. Click on the fourth icon and choose the magic wand tool. Once selected, click on the empty space. Before using generative fill, expand the selection. Click on the icon next to generative Phil, and choose Expand Selection In the pop ups set, expand by to 80 pixels. This gives the AI more freedom to generate the transition, making the result more realistic. Click okay, then generative Phil, leave the prompt, field blank and click on Generate. Let's look at the finished pictures. I think the first one is the best. It looks quite good, especially given we only used AI generation once. Still, we can improve it slightly. Let's merge the layers, right Click on layer one, then click on Merge, Visible in the top menu. Click on Image, then auto tone, auto contrast, and auto color. I'll also adjust the light on the bears hindquarters, outline the bears as with the lasso tool. Then generate it without a prompt. It looks much better now. Wonderful. It's that simple to create a realistic and entirely unique photo using our own images. I hope you enjoyed the lesson. See you next time. 17. End of the Beginner Class: Hi there, You did an excel***t job. Congratulations, you've successfully completed the beginner class course content. In this material, we've explored Photoshop and its generative fill capabilities and controls. By now, you possess the skills to transform any image into whatever you desire. But this is just the beginning. If you feel like you'd like to practice the basics more, I recommend checking out the practice videos. If you feel confident about your basic skills and want to dive deeper into understanding Photoshop, I continue with the course and start the advanced class. Happy Learning, and see you next time. 18. Introducing the Advanced Class: Welcome to the Photoshop AI Advanced Class. Having mastered the basics in the beginner class, we'll now delve deeper into understanding generative fill, and learn a plethora of techniques. In this module, we'll explore the use of the partial mask. We'll transform a damaged painting into a photograph and then turn our photo into a drawing. We'll learn to perfect the blur effect, how to merge multiple images into one, how to create rain or snowfall on our image, and much more. By the end of the advanced class, there will hardly be any image you can't produce using the generative fill happy learning and perseverance with the advanced class. 19. Layer Mask editing - How to use multiple generated images for one picture.: Hi there. Today we'll be creating a new image by merging two generated images and editing the layer mask of the generated image. Start by opening image number three in the res folder using the open command. This image features a forest and a lake. Once opened, select the crop tool, Expand the image size in all directions. For this image, I've used a 16 to nine aspect ratio. Make sure to verify that generative expand is selected under Phil. Once you've expanded the image as shown in the video type forest and lake into the prompt after entering click on Generate. Let's wait a bit for the AI to finish generating the images. There we go, The images are ready. I'll zoom in a bit for a better view. All three images turned out fantastic. Choose the one you like the most from the three. My favorite is the third one. Now, right click on the generative layer in the Layer menu and select Duplicate Layer. This will copy the layer. I'll now switch back to the previous generative layer. I change the second one to another image then and turn off the visibility for the first one. I prefer the left side of image two over the left side of image three. I'll erase the left side of image three and replace it with the left side of image two. To do this, we need to erase the undesired part in the layer mask, restore visibility to all the layers. Then click on the layer mask thumbnail of the topmost layer to edit. Select the eraser tool and start erasing the parts from image three that you didn't like. It's advisable to choose a smaller eraser size if you want a more precise eraser. You can also adjust the opacity of the topmost layer. This makes the top layer slightly transparent, allowing you to see what the underlying layer contains. Once you've finished erasing, you'll notice some areas where the difference between the two layers is visible. We'll smooth these out using the smudge tool. Choose the smudge tool from the left menu. Set the appropriate brush size and strength. Once you've set them, start smoothing out the edges of the layer mask. We're almost done. Check the difference and temporarily turn off the top layer mask this way. If there are any remaining errors, you can easily spot them. I think it turned out great. This is how you can combine two generated images to create a new one. I hope you enjoyed this tutorial and I'll see you next time. 20. Restore a damaged old paint to make a photo: Hi there. In this video, we'll be restoring an old damaged painting to photo quality. Let's get started. Click on open and open image number 13. We will be restoring this image to make it photo quality. Let's use the generative fill to correct the white spots where the paint is chipped off. Let's start correcting them with the Lasso tool. If quality is a concern, it's advisable to correct small sections one by one, rather than fixing all the white parts at once. We will delve deeper into the reasons behind this. In the master class sessions, use the lasso tool to outline all major flaws and repair them without any prompts. I speed up these monton parts a little in the video. Don't miss this large crack. Circle it, and without any prompts, click on Generate. The tooth didn't come out perfect, but let's deal with that now. If you want to select multiple things simultaneously, hold down the left shift button continuously while drawing. We're done fixing the major errors. Click on the back ground layer from the pop up menu, select Merge Visible to combine all layers. On the background layer, I unlocked the image editing by clicking on the lock. But this isn't crucial for us. Now if you click on it, our layer named Background will be renamed layer zero To further enhance the image, we need to create a partial mask. Click on Edit in Quick Mask mode. From the left menu in the top menu. Click on the Edit tab and choose Fill. From the dropdown in the window that appears under Contents, choose Color. Adjust the blue channel to 30% The higher you set this percentage, the more the generated image will differ from the original. Once done, click okay and then click okay again. Now we can see on the image that the partial mask is ready. Now let's exit the Quick mask mode by clicking on this icon from the left menu. Even though the red layer disappears, it's still there, just invisible. Don't select anything, just click on Generative Fill. Leave the prompt field empty and click Generate. Because the image size is large, it will take some time to generate the images. The AI has finished generating the images. For me, the first image turned out best. Select the image that you find best. Then on the right side under the layer tab, adjust the opacity. The image I generated looks best at 80% But how much you adjust is up to you. The painting already looks much better. Merge the image into one layer by right clicking on layer zero and selecting Merge Visible. Now let's create another partial mask. Follow the previous steps. Click on it and quick mask mode from the left menu. Then in the top menu, click on the edit tab and choose Fill From the drop down in the window that appears under Contents choose Color. Adjust the blue channel to 30% Once done, close the windows with okay buttons and exit the quick mask mode. Now click on Generative Fill. You can leave the prompt empty, or if you wish, experiment with prompts like realistic. Then click Generate. Once the images are generated, select the best one and adjust its opacity. When done, you can merge the layers again. You can repeat this kind of image generation as many times as you want. The more images we generate, the closer the quality of our image will be to that of a photo. If you set the partial mask not to 30% but to 20% or less, it will take longer to reach photo quality, but the result will deviate less from the original image. It's also worth correcting minor errors that might arise during the process. Here, for instance, the eye looked odd, which I fixed in the top image menu under the adjustment section. Don't hesitate to make changes. Be daring and experiment. I achieved this image quality after five generated images. I was a little impatient, so her mouth became a little different. But I believe the result speaks for itself. I hope you enjoyed this tutorial and I'll see you next time. 21. Unique Composite Images with Special Effect: Hi there. In this lesson, we'll learn how to transform a mundane photo into a stunning image using generative fill. Let's get started. First, click on Open and select the image numbered 15 in the Res folder. Once the image loads in Photoshop, navigate to the top menu, click on File, and then click on Open in the Res folder, open the image numbered 14. What we're aiming to do is extract the ballet dancer from the photo and superimpose her onto the phone image, giving her a holographic style on the left toolbar. Click on the rectangular marquee tool and select the girl in the photo. Then navigate to the top menu. Click on it. From the dropdown, select Copy. Now switch back to the image number 15 in the top menu. Click on it, and from the dropdown, select Paste. I imagine the image as her dancing in holographic form above the phone screen. The phone is at the center of the image, but I want the girl to be the focal point. Hence, we'll lower the phone a bit. Let's get started. For now, let's hide the new layer by turning off its visibility on the back ground layer. Click on the lock icon to make it editable. From the left tool bar, choose the move tool and drag the image downward just a bit as a result of moving it down, The top part of the image is now empty. We can fill this using the generative fill. Using the left toolbar. Select the rectangular marquee tool and mark the empty space at the top. Click on generative fill and generate the missing part without a prompt. The first generated image looks good. Let's merge the phone image with the newly generated right click on Layer and see Visible. Now our background is ready and we can begin working on the ballet dancer. Turn on the layers visibility and click on layer one to edit it. First, let's remove any unnecessary parts of the image from the left toolbar. Choose the Object selection tool and click on the girl invert the selection by going to the top menu, choosing Select, and clicking on Inverse from the dropdown. We've now highlighted all the unwanted parts of the layer, delete them using the Dell key, or navigate to the top menu. Choose Edit. And from the dropdown, select Clear, Deselect the selection by clicking outside the image, or go to the top Select Menu and click Deselect. Move the girl to the center of the image. Select the Move tool from the left menu, drag the girl into position. I would like to slightly reduce the size of the girl. Navigate to the top menu, select edit, then choose transform from the dropdown. And within it, opt for scale. I've only slightly reduced her height. Now we can begin transforming the girl from the left toolbar, select the object selection tool and click on the girl. Once selected, click on generative fill and type in the prompt green matrix. After entering the text, click Generate. You can use a different prompt. I have tried a prompt that makes glowing random outlines. The AI is done generating the images. I like the first one best reduce the opacity, reduce the transparency of the image. Then right click on the generative layer and choose duplicate layer. I set the new layers opacity to 100% generated a new layer using the same green matrix prompt. Once the AI finishes generating, I opt for an image with illuminated outlines. Among the b***ding modes, I select Color Dodge. Feel free to experiment with the b***ding modes. Once the b***ding mode is set, adjust the opacity. Click back on layer one and reduce its opacity as well. The image is starting to take shape. Let's correct some minor imperfections. Click on the topmost generative layer Mask thumbnail. And then use the zoom tool to magnify the girl's head. I don't like the see through effect on her head. Let's fix this on the left side, select the eraser tool and begin erasing this layers section around her head. Once done, click on the layer mask thumbnail of the layer beneath and begin erasing the part around her head there as well. And we're done. Within a few minutes, we've managed to create such an incredible image. I hope you enjoyed this lesson. See you next time. 22. Covert your Image to a Paint: Hi there. In today's lesson, we'll be transforming a photo into a painting using the generative fill feature. Let's dive right into the lesson. Click on Open within the Res folder. Select image number six. This image features a blond haired girl on the beach. If we wish to modify an existing image, our approach will differ from what we've done in the past. Merely selecting an area and typing paint into the generative Phil prompt will create an entirely new painting unrelated to our original photo. Let's give it a try. Select the rectangular marquee tool, highlight the entire image, then click on Generative Fill in the prompt box type in watercolor painting. Once entered, hit the Generate button. Ai generated the images. As you can see, it bears no resemblance to the original photo. Right click in the bottom right menu under the Layers tab on the generative layer. Choose delete layer to remove the layer. To ensure our actual image is only modified, we need to create a partial mask first from the left menu. Choose Edit in Quick mask mode. Once selected, you'll notice that the image under layers has turned red. Next, click the top Edit menu and choose the Fill option in the window that appears for contents select color. Adjust the blue channel to 30% one set. Click Okay and then okay again. We're done. And we can now see the completed partial mask on our image. Exit the quick mask mode by clicking its icon in the left menu. Although the red layer disappears, it remains present but invisible. Now click on generative fill again and type watercolor painting into the prompt box. After typing, hit the Generate button. While the results might not be perfect, it's evident our photo has been repainted in water color. If we desire a better image, we can keep generating images with the AI as many times as we wish. Pick the best looking image. We're now going to frame this image. Click on Image in the top menu. Then from the dropdown, select Canvas size. Increase the width and height of our workspace by 200 pixels each. The width will be 1,400 pixels height, 2000 pixels. After adjusting hit, okay, now the image has a white frame. While our image looks great with this white frame, let's generate a proper frame for it. Use the zoom tool to shrink the image for a better view. Choose the magic wand tool and click on the white area to select it. I'd like the selection to be a bit larger by clicking on the icon to the right of the generative fill. I'll expand Selection. This lets us increase the selected area in the pop up. Specify how many pixels to enlarge the selected area by then click Okay, I expanded mine by ten pixels. Now turn off the visibility of the layers on the bottom right. This is necessary because if they remain active, generative, Phil will consider these layers when generating the image frame, which we don't want. In the video, I created a new layer, but this is only necessary when our image has just one layer. However, in our current situation, besides the background layer, we already have a generative layer. Hence creating a new layer isn't essential. Click on generative fill, then in the prompt box type old painting frame, once entered, hit the Generate button. Now our frames are ready. The first one is clearly the best. After choosing your preferred frame, restore the visibility of the other layers. There you have it, our framed watercolor painting. With this lesson, you've learned how to repaint images using the partial mask. We can not only repaint images but also use it for many other functions which will explore in upcoming lessons. I hope you found this lesson useful. See you next time. 23. Perfect Background Blur effect: Hi there. In today's video we'll learn how to easily blur the background of an image and how the generative, Phil can assist us in this task. First, click on Open and load image number 14 from the Res folder. We'll be blurring the background of this image. Our first step will be to move the girl to a new layer. If you click on Select Subject, it should ideally select the girl. However, the A I failed to fully do so this time as it only partially selected her left hand. If this happens, go to the left menu and choose the object selection tool. Then click on the girl, especially on the missed out parts. The selection was recalculated and looks much better now, but there's still an error. It missed out on some of her hair above her ear. We'll need to fix this. Manually zoom into the girl's head using the zoom tool. Right click on the fourth icon from the top and select Quick Selection tool. Choose a smaller brush size and start selecting the missed areas. Zoom out on the image. It's better to click directly on Fit screen instead of zooming manually, like I did in the top menu. Click on the Edit tab and select Copy. Let's create a new layer and paste the girl onto it. On the bottom right, click on the icon that resembles a plus sign. Once the new layer is ready, go back to the edit tab in the top menu and choose Paste. We've now created a new layer with only the girl. Let's create another layer, this one just with the background without the girl. First, turn off the visibility of the new layer we just created with the girl. Click on the background layer from the left menu. Once again, select the object selection tool and click on the Girl. Now we'll be erasing the girl from the background layer. It's a good idea to slightly over select the subject you want to remove. So the generative fill can work more efficiently on the generative taskbar next to the generative fill icon, expand selection and expand the selected area by 40 pixels. That looks about right. Click on generative fill in the prompt field. Don't type anything, just click on Generate. The AI will now produce a layer without the girl showing only the background. Let's check the images created by the AI Images 1.2 are perfect, select the best looking image and then merge the two layers. Click on the background layer and select the merge visible command. Now we have two layers, one with only the background and another with the girl. Turn the visibility of layer one back on so you can see the girl in the image. It's time to blur the background. Ensure the background layer is selected and then click on the filter tab in the top menu. There are various ways to blur the background. The most commonly used is ***s blur. Within the blur option, this will blur the entire image. However, this won't work for us, as it would blur the ground beneath the girl, resulting in an odd look. We need a different type of blur. Click on cancel again, choose Filter in the top menu. But this time go to Tilt Shift under Blur Gallery. This effect creates a horizontal band that blurs the image vertically. Click on the white dot and drag it down to the bottom of the screen. That's much better. Once you've adjusted it, click Okay, Our image already looks fantastic. But let's refine it a bit more. Turn off the visibility of the background layer and make some corrections on the girl. The upper part of her left hand isn't perfect. Zoom in and use the eraser tool to erase the gray areas between her fingers. I didn't find any other major errors on the girl. Select the zoom tool, click on Fit Screen and restore the visibility of the background layer. The most noticeable issue right now is a slightly pixilated outline around the girl. We'll fix that next. The left menu. Select the object selection tool and click on the girl. Once selected on the generative taskbar next to the generative fill, choose select border in the pop up window. Input two pixels for the width. And click okay. We've now selected the girl and the area around her. Click on Generative Fill, and then without adding a prompt, click on Generate. Let's check the results. Select the best one and play around a bit with the opacity. The result is flawless. It's that easy to enhance our images. You can further work on the image if you want. For instance, you could give the girl angel wings or add extra effects to the image. The only limit is your imagination. I hope you enjoyed this tutorial. See you next time. 24. Make a hand drawing from your photos: Hi there. In this lesson, we'll transform a portrait into a hand drawn image using generative fill. It's a quick and simple way to turn your photos into sketches, making them a perfect update for your social media profiles. Let's get started. Start by clicking on Open and select the image number 18 in the Res folder. We're going to create a sketched version of this beautiful girl. First, let's zoom in on the image. Click on the Zoom tool, then select Fit Screen. From the top options, I had a simple black and white sketch in mind. We'll begin by turning our image graycale from the top menu. Select Image in the dropdown. Within adjustments, choose the black and white option in the pop up window. You can leave everything on default settings, but if you wish, you can also select different colors. Now our image is in gray scale. But before we can use generative fill, we need to create a partial mask. Let's do that next. In the left menu, click on Edit in Quick Mask mode. From the top menu, choose Edit. And then fill from the dropdown in the window that appears under contents choose color. Change the blue channel to 30% in the pop up. Close the windows by clicking Okay. If you see this red layer, you've successfully created a partial mask. Now exit the quick mask mode. Editing from the left menu. Now click on Generative Fill in the prompt type charcoal face illustration detailed pen and then click Generate the A I generated the images for me. The third one looks the best. Don't hesitate to generate a new one if none of the options appeals to you. Once you've chosen the best result, set the b***ding mode of the generated image to overlay The image. Already looks great at this point, but we can refine it further by creating another partial mask. Click back on the background layer, then enter the edit in Quick mask mode. Again from the top menu, select Edit, and then fill under Contents, Pick color again, change the blue channel to 3% in the higher this percentage there, the generated image will differ from the original. For now, I'd like it to deviate only slightly. Close the windows by clicking okay and exit the Quick Mask Mode editing. Now click on Generative Fill. This time leave the prompt field blank and just click Generate. The AI is done with creating the images, but this layer became the second layer. We hardly see the result from the first layer to see the result, turn off the visibility of the first layer. As we can see, it barely differs from the original. Choose the best image. Then turn the visibility of the first layer back on. Use the left mouse button to drag the recently generated layer to the top. It's the uppermost layer. Set its layer opacity or transparency to 50% There you have it. Our image is complete and looks fantastic. Simple to create a perfect sketch of yourself with this technique. You can also make oil paintings or even transform your photos into cartoon characters. Your imagination is the only limit. I hope you enjoyed the lesson and I'll see you next time. 25. Creative Advanced Selection: Hi there. In this lesson, we'll be b***ding two photos with very different lighting conditions. It won't be an easy task, but AI will assist us. Let's dive into the lesson. Click on Open in the Res folder. Select image number five now, while continuously pressing the left control key, Also click on image number 17. After selecting both photos, click on Open. This way we'll open both photos simultaneously. We'll be placing the two girls from another photo onto this one. Begin by removing the man from this photo. Choose the lasso tool and outline the man after outlining him, click on Generative Fill in the prompt. Don't enter anything, just hit the Generate button. The AI has removed the man from the photo. Choose the best image, then switch to image number five with the two girls. We'll be copying these two girls onto the other photo. First, let's select the girls for copying. Click on Select Subject below the A. I recognized the subjects, but the selection isn't perfect. It didn't capture the left girl's shirt and the hair selection is flawed. We can get a much better result by choosing the cloud service for the selection. Let's try it out from the left menu. Pick the Quick Selection tool in the upper toolbar. Next to Select Subject, you'll find a downward arrow. Click it and select the Cloud option. Once chosen, click on the adjacent Select subject. This gave us a much better selection. It even detected the area under the hand. We can refine the selection further by choosing Select and Mask next to generative fill. Here you can clearly see which parts of the image are selected. There are many options to refine the selection. It's worth trying out all the settings. Once adjusted to your liking, click on the OK button. Now let's copy the selected girls onto the other photo. From the top menu, select the edit menu. Then from the dropdown, click on Copy, switch back to image number 17. Then from the top menu, select the edit menu and click on the Paste option. We've pasted the girls onto the photo, but their size isn't right. We need to enlarge them to fit the image at the top. Click on the Edit menu. Again, in the dropdown under Transform. Choose the scale setting. Now we can freely adjust their size, enlarge them, and drag them to the appropriate position. This will do, it's evident that they don't b***d well into the image. The two photos were taken under very different lighting conditions. The light source is different and even the sharpness of the image varies. However, using AI and a few techniques in just a few minutes, we can b***d the two photos quite well, if not perfectly. But first, let's start by converting the background into a single layer. Turn off the visibility of layer where the girls are so we don't b***d them into the background for now. Once done, right click on the background layer and select the merge visible command. Let's try to bring the two images closer together. Select layer one for editing. Once selected in the upper menu, click on the image option in the dropdown, click on Adjustments. Here you can adjust almost all aspects of the image. Don't be afraid to experiment and explore all the options. Consider visiting levels and curves to fine tune brightness and contrast. Adjust the color balance to match color temperatures. With hue and saturation, you can adjust the saturation. And with selective color, you can fine tune specific colors. The match color and photo filter options are also very useful. But we won't use them for now. You'll see them in the next How to fix video. For now, I'd like you to manually adjust the colors of the image as this will provide a deeper understanding. Once you've adjusted the image settings, you can proceed to the next step. Generate the legs of the girls using generative Phil. Choose the rectangle tool from the left menu. And select the empty area below the girls, where we'll generate the legs. After selection, click on generative Phil. Don't enter anything in the prompt. Just click Generate. Generating Legs and arms with AI is chal***ging, especially when subjects are very close together. The AI has generated the images, None of them are particularly satisfying. I'll generate them again and hope for better luck. These turned out even worse. I see potential in the sixth image, so I'll choose that one. But it certainly needs tweaking to correct this. First merge the two layers, turn off the visibility for the background layer. Then right click on layer one and choose the merge visible option. After merging, restore the visibility to the background layer. From the left hand menu, select the lasso tool and circle the flawed area and the girl's arm. If you wish to adjust the selection continuously, press the left shift key to add to the area, or the left Alt key to reduce it. Once selected, click on generative fill in the prompt box. Don't type anything, just click on the Generate button. Let's see how the image is turned out. The second one is perfect, merge the layer again. Turn off the background layer. Then right click on layer one and choose the merge visible option. What I don't like is the green light on top of the brunette girl's head. It completely stands out from the image. So let's fix that. Circle it with the lasso tool, then click on generative fill. Again, don't type anything in the prompt box. Just click on the Generate button. Let's see the results. The AI did a great job making it look much more natural. Let's merge the layers. I don't want to make any more corrections to the girls, so we can also merge them with the background layer to make the image look even more natural. I'll generate another layer to further balance the lights and colors. For this, we'll create a partial mask in the left hand menu. Click on Edit in Quick Mask mode in the top menu. Click on the Edit tab and select Fill from the drop down menu in the pop up window. Under Contents choose Color. In the next window, adjust the blue channel to 7% Click okay to close the windows and exit the Quick mask Mode Editing. Now click on Generative Fill. Again, don't type anything in the prompt box, just click on Generate. The AI will recreate the entire image using its patterns, making the generated image look much more natural than mine. However, I've set the partial mask to just 7% It makes only a very slight change the AI has generated the images, We can see that the heads appear distorted, but we don't need to worry about that. Set the b***ding mode of the newly generated layer to hard light and reduce its opacity to 15 to 20% With hard light, the images get a strong contrast effect. And it'll be less noticeable that the two layers were different, making it look somewhat more natural. Of course, we can add more partial masks if desired, but the image might distort over time. Merge the layers using the merge visible command. Now that we have one layer, we can start fine tuning the entire image using the adjustments under the image menu. We're done. In just a few minutes, we managed to create such an amazing image from two completely different pictures. It's almost impossible to notice that it was created from two images, even though we didn't use all available tools. Further refining of the images will be covered in the how to fix on. Exceptionally save this image as we'll use it in the next video. I hope you enjoyed the lesson. See you next time. 26. How to Fix after Generative Fill: Hi there. We've already learned how to fix minor flaws in our photos. Now let's look at how we can enhance our merged images made from multiple photos to make them even more realistic. Click on Open in the Res folder, Select the image numbered 17, and then click open if you've opened it, now open the image we created in the previous video. If you haven't saved it, you can find my version in the practice folder in the top menu. Click on file, then open inside the Res folder, Open the practice folder, and select the image named Pan two. Then click open. We'll refine this image further. When you merge two images, the most noticeable difference between them often lies in color and shade variations. Photoshop allows us to transfer the color settings from one image or layer to another. Let's try it out in the top menu. Click on Image. In the dropdown, under the Adjustments tab, choose Match Color in the pop up window. For the source, select the image numbered 17, and then you can start customizing settings under Image Options, Feel free to deviate from my settings. Once done, close with, okay, let's see the difference. It looks much more natural. It's worth starting with this setting when b***ding to images even before you begin editing them. There's another useful setting that can greatly enhance the image to make it look more natural. In the top menu, click on Image. In the dropdown, under the Adjustments tab, choose Photo Filter. This option applies a color filter to the image, altering its color, temperature, and hue. The difference is clearly visible. Choose the cooling filter, LB B, and adjust the density, 10-15% then close with. Okay, the improvement is clear. It looks almost like an original photo. Apart from colors, there's a minor issue with sharpness between the two images. The girls and the background differ in sharpness. We can tweak it a bit in the left menu. Click on the Quick selection tool, the dropdown besides select Subject chose the Cloud option. Then click Select Subject. We've now selected the girls, but we need the background. Invert the selection in the top menu, click Select, and then inverse from the dropdown, now we've selected the background. I reduce the selection by one pixel contract selection, but it's not crucial. Now that we've selected the background, let's slightly adjust its blur. In the top menu, choose Filter. And under Blur Gallery, select Tilt Shift, Start Adjusting. I only want a subtle change, so I set the blur to three pixels. One set. Click. Okay, let's compare the final result. I think it's stunning in just a few minutes, we've seamlessly b***ded two very different photos. But not only blur can be utilized. The sharpened tool can be equally useful for sharpening parts. There's one more image to practice on. Close these images and return to the menu. Click open inside the Res folder, open the practice folder, and select the image named Pan one. Then click open. You can find the making of this image in the practice video section. After opening the image, also open the original in the top menu. Click file, then open in the rez folder. Select image number 19 and click open switch back to the image named Pan One. Start again with match color in the top menu. Click Image in the dropdown, under the adjustments tab. Choose Match Color in the pop up window for the source. Select image number 19 and begin customizing settings under Image options. Once done, close with, okay, now set the photo filter as well in the top menu. Click on Image. In the dropdown, under the Adjustments tab, choose Photo Filter for this image, the warming filter looks better. I'll use that upon closer inspection of the image, there's an issue with sharpness. The bear's body and its surroundings are too sharp compared to the archer girl. Unlike the previous image, we can't use the same blur filters. In such cases, it's best to blur manually. The left menu select the blur tool. Adjust the blur intensity to low visualize the bear and its surroundings in relation to the girl and her surroundings. If you see the difference and can visualize the bear in space, start blurring the layers. First, I'll start with the surroundings of the bear, which are too sharp. Then the bear's back side, which should appear more blurred than its head. It looks much better. Now, another tool I often use when editing images is the Camera Raw filter. In the top menu, Click on Filter and choose Camera Raw Filter from the drop down. Here you'll find nearly all the adjustments you need to make the image even more beautiful and realistic. It's worth going through and trying out all the settings. Once you're done, click Okay. Let's check the difference before and after the change is evident. It turned out incredibly well. I hope you enjoyed this lesson. See you next time. 27. Reflection and Rain: Hi there. In this video I'll be demonstrating how to create reflections and rain using generative. Phil, let's get started. Click on Open and select image number 16 in the Res folder with generative Phil. It's a breeze to create a reflection or bring rainy weather to our image. In this tutorial, we'll be transforming this bear image. The bear will have a lake in front of it and it'll be raining in the picture. Once you've loaded the image, select the lasso tool from the left menu and outline an area in front of the bear. Once you're done, click on Generative Fill in the Prompt Field Type in Reflection, and then click Generate. For the prompt, you can also add lake or puddle next to reflection. But for forest type images like this, the AI will almost always generate a lake unless you specify otherwise. For city images, it might create a puddle. I generated six images with the AI, and this one turned out the best. Check out the bear's reflection. It's stunning generative fill can create lifelike reflections in just a few seconds. The reflection for metals isn't perfect yet, but I'm sure Photoshop is already working on improving that. But it's not just good at generating reflections, it can also make realistic rain and snowfall. Let's make this image rainy. To create rain, we need a partial mask. First, merge the layers, right click on background and choose the Merge visible option in the left menu. Click on Edit in Quick Mask mode. In the top menu, click on the Edit tab and select Fill From the dropdown in the window that appears under Contents choose Color in the pop up. Adjust the blue channel to 12% Click Okay to close the windows and exit Quick Mask Mode Editing. Now you can click on Generative Fill in the prompt Field, type in in, and then click Generate. I generated nine images with the last two being the best, which was what I had in mind. If you also managed to generate a great result, set the b***ding mode of the layer you just generated to lighten. There you have it. Our image is complete. It turned out incredibly well. The picture is entirely realistic. The AI even created the ripples from the raindrops on the water. We can create stunning images in just a few minutes with generative fill. I hope you enjoyed this lesson. See you next time. 28. End of the AdvancedClass: You did an excel***t job. Congratulations, you've successfully completed the advanced class material. In this course, we delve deeper into the generative fill technique. You now possess confident knowledge and can create or modify any image. If you feel you'd like more practice, I suggest you watch the practice videos if you wish to explore Photoshop AI even more deeply. I recommend the master class. In the master class, we explore the workings of AI and how such a system is structured, wishing you continued, enjoyable learning. 29. Introducing the MasterClass: Welcome to the master class. This course is designed to help you better understand the system behind generative. Fill the artificial intelligence itself and how it functions compiling. This material was a chal***ge for me as well. Although I've been working with AI and machine learning for several years, every system is unique. Furthermore, Photoshops AI is not open source. We have no direct insight into its workings. In essence, we have very little information about Photoshops artificial intelligence. Given this, much of the content is based solely on my own experience and knowledge and does not include external sources. There might be some inaccuracies. However, I trust that you will still learn a lot from it and gain a deeper understanding of the artificial intelligence behind generative Phil Happy learning. 30. Database Structure: Generative fill leverages artificial intelligence to alter content within a selected area. This AI was trained on vast amounts of image data and text. This enables it to discern the relationships between text and images. Using this knowledge to generate images. Let's delve into how Photoshops AI operates upon entering a prompt, say hat, into the generative fill and clicking on the AI first analyzes our image. It identifies every crucial element in the picture. It then tries to comprehend the prompts, meaning and determines the type of image it should create. To do this, it taps into its internal knowledge and data, such as vocabulary, grammar, semantics, and visual concepts. Using the information it's formulated, it searches its database for suitable images. Once located, it attempts to seamlessly combine images, for instance, of a hat and a bear. Adjusting their positioning, orientation, lighting, and perspective to make them appear as a coherent whole. Finally, it smooths, sharpens, and colors the image. From this, it's evident that the task it performs is more intricate than popular AI, like mid journey and stable diffusion Photoshops. Ai likely employs a reverse diffusion process, similar to stable diffusion when generating images. This begins with pure noise, which is gradually removed until a clear image emerges. Noise associated with the core is sampled from a Gaussian distribution. At every step of the diffusion process, it's added to the image. A neural network governs the reverse diffusion process, predicting the amount of noise to be eliminated at each step. However, Photoshop, AI uses a new version of the diffusion model, the generative fill in. Photoshop uses an algorithm that breaks down an image's structure into multiple elements. This means the algorithm can recognize different elements in an image, such as objects, textures, colors, lights, and perspective. This allows for a more accurate and lifelike fill compared to traditional algorithms. This technique has both pros and cons. The advantage is its non destructive nature. The fill doesn't alter the original image content. It takes into account the image's structure and style and can more easily generate complex images. However, a significant drawback is that generating an image demands more time and resources than other AI, like stable diffusion or mid journey. Therefore, to avoid extended image generation times, Photoshop trained its database using smaller image sizes. With stable diffusion, the size of the images used in an average Laura or checkpoint model can range 512-1024 pixels. While Photoshop can use a fraction of that. I believe that during the Photoshop beta phase, the A I database was trained with images of perhaps 128 or a maximum of 256 pixels. However, to better quality another AI up scales these images, this meant faster AI training, lesser resource requirement for image generation, and quicker overall process. But there's a downside. More complex things such as faces become distorted when upscaled to larger sizes. Let's illustrate with an example. If I select a smaller area on an image and input the prompt girl, the results are clear faces with no distortion. If we select a larger area and input girl, even without selecting a significantly large section, the faces and bodies are distorted. Larger images make the distortion even more noticeable. However, if we isolate the girl's face with a smaller selection and generate using the girl prompt, we get a clear and normal face because the AI didn't need to upscale the face preventing distortion. The same can be done with a hand. If I selected in a small box, input the prompt hand, A non distorted realistic hand with five fingers is produced. Photoshops AI is capable of creating entirely realistic, flawless human figures and other intricate objects after the Photoshop beta phase. They'll likely use a database trained on larger images, resulting in better quality images. But you can still use this technique, just you can select twice or even larger areas. I hope this gave you a deeper understanding of Photoshops. I see you next time. 31. Max Sample Size: High there. From our previous video, we already know that the trained images in Photoshop Beta have a very low resolution. However, we can still get a high quality generated image thanks to upscaling and by generating multiple distinct elements within a selected area. Think of it this way. When we generate a forest, the AI pulls individual images and patterns of trees from the database and places them side by side. The resulting forest might be 800 pixels wide, but it's made up of many 128 pixel wide trees standing next to each other. There's another limitation to the generative fill though. The maximum generated size cannot exceed 1024 pixels in width or height. What does this mean in practice? If we generate an image on a selected area larger than this, Photoshop will stretch it out. If we generate an image that's 2048 by 512 pixels, the Photoshop I will create a 1024 by 256 pixel image and simply stretch it afterwards. This results in the image becoming much blurrier. This limitation was necessary because the larger the generated area, the more time and resources it takes to produce the image. Hence, they set such a limit. Let's see this in action. I'll increase the image side by 1024 pixels. I've selected an area that's 2,400 pixels high and 1039 pixels wide. Since the longest side is significantly more than 1024 pixels, it's sure to be very blurred. Indeed, the generated image confirms this. Let's try it with a 1024 pixel square. Generate a square that's 1024 pixels in both width and height, and observe the difference. The image generated within the 1024 pixels square is noticeably sharper, although it's still a bit blurry compared to the original. This is because the AI has to upscale the size, but at least it didn't need to stretch the image, resulting in a better quality image. To avoid upscaling, I'm going to generate an image inside a 128 pixels square and then compare it with the previously generated layers. The image in the 128 pixel square came out as sharp as the original. It's clear that the AI doesn't need to upscale the model here. Of course, it would be quite time consuming to generate an entire image from 128 pixel squares, but if we want the best quality, we might have to do it. If we're okay with lesser quality, we can use the 1024 pixels square. I expect that after Photoshop beta, the trained images will be of higher quality and at least 512 pixels in size. I believe the final generative fill will allow for a generated pattern size greater than 1024 pixels. The tests I've just demonstrated should be done after every update. Because with each update, the model database used by the AI and the maximum width of the generated image might change. I hope you found this lesson helpful. See you next time. 32. Prompt and Word Weight: Hi there. In this video, we will discuss prompts and delve into their analysis. Let's begin with defining what a prompt is. A prompt is an instruction or description given to the AI based on which it generates the image. How does AI interpret the instruction written in the prompt? During A I training, we associate keywords with every image. The accuracy of these keywords is vital for the AI to generate what we desire. Another cornerstone of prompt interpretation is the I's capability to understand. When we say the AI interprets a prompt, it means it can grasp the meaning and semantics of words and sentences in text. To image, generating AI to work efficiently, it must comprehend human language. This capability is achieved through deep learning models that learn from vast amounts of text data. This understanding allows the AI to grasp complex commands like draw a green dragon sitting on top of a mountain. Merging these two cornerstones leads to innovative applications. For instance, the AI could create a full illustration or even a detailed landscape based solely on a brief description. To better understand and manage the generative fill, it's essential to recognize these two aspects. If we aspire to become prompt engineers, we need to understand how the AI analyzes and interprets our texts. It requires numerous hours of familiarizing and testing. Currently in Photoshops AI, we don't have as many customization options for prompts as we do in mid journey. However, this will change in the future. Just consider this simplified mid journey cheat sheet over time, many of these features will be integrated into Photoshop AI. But discussing them now is premature as we lack specific details. Technically, the only thing we currently need to focus on when providing a generative fill prompt is text weights. We don't even have an option to adjust this separately. It's embedded in the logic of AI's prompt interpretation. What does text weights mean? The A I tries to understand your prompt by assigning a value or weight to each word. The term with the most substantial weight becomes the primary factor in image generation as demonstrated in the video. If we write cat nine times and dog once at the end, the AI will always generate a cat. This is because the weight of the word cat is significantly more than that of dog. With similar prompts, you can test how the AI determines what to generate and understand how Photoshops words weights function. From my tests, it's clear that the weight of words decreases linearly. The first word has a higher initial value than the second. The second words value is greater than the third, and so on. Therefore, it's always advisable to start our prompt with the subject and mention the least essential aspect at the end. What else should we pay attention to? We get the best results if our prompt is 3-8 words, thus always try to be concise and accurate. It's also essential to consider the relationships between words. Although not the best illustration, it helps depict the interrelation of terms. If we input the prompt as wings fly, the AI is very likely to generate a bird. This is because both words relate to the bird word. In more advanced neural networks, relations also have values. In this case, the words wings and fly strongly correlate with bird. If you wish to become AI, prompt engineers, your main task is to experiment extensively and observe how the AI behaves. You understand the basics now, All it takes is persistence and a lot of practice to fully comprehend. Remember, an AI prompt engineer must continuously learn, since an AI can change considerably in just a matter of weeks, I hope this helps in deepening your understanding of prompts and AI functionality. See you next time. 33. Practice - Make a windows background (Beginners level): Hello, in this video, we're going to create a wallpaper. In the practice folder, you'll find an image titled, I Generated Art Mountains, open it. This image is much smaller than the resolution of my computer. Go to Image Dropdown menu and the canvas size option. I'm going to change the image size to 1920 pixels wide and 1080 pixels height. These values match the resolution of my monitor. Once we click okay, we get a large white border around our image, which we're going to fill with generative fill. First, we need to select the area. We can manually select the area using the rectangular marquee tool. After selecting an area, press the shift key, and this way we can add new areas. However, a much simpler solution is to use the magic wand tool. If your fourth icon at the top is different from mine, right click and select the magic wand tool. Before clicking on the empty white space, make sure that the tolerance setting is very low. This way we can select the entire area with one click. Then it's worth expanding the frame with expand selection so that there won't be a faint white border between the new generated content and the image. And the generated content will better match our image. I'm expanding the selection by 20 pixels here. Once we're done with this, click on the generative fill. Leave the prompt blank. We'll have to wait a bit. Now Photoshop has finished generating the images. Click here to watch them. All three. Turned out really good, but I like the first one the most. After selecting one, I merge the layers. I write, click on the background and select the Merge Visible option. Now I've merged all layers into one layer. Then I'll do some touch ups on the image. From the image drop down menu, I select Adjustments, then click on the Brightness and Contrast option. In this setting, I'm going to increase the brightness and contrast values. After doing this, I select Auto Tone from the down menu, which automatically optimizes the colors, now looks much better. We can further improve the image by fixing these minor errors. Select the lasso tool, draw around the error, and remove them using generative fill without a prompt. After we finished with the errors, we merge the layers again. Right click on the background layer and merge the layers using the Merge Visible button and it's ready to save. Choose either a JPG PNG file type and it's ready to use as your new wallpaper. 34. Practice - Expert Merge (Advanced Level): Hi there. In this video, we'll practice b***ding multiple images together. Let's get started. Click on Open in the Res folder. Select the image labeled 16, then while holding down the left control key, click on images 19.20 as well. Once all three images are selected, click on Open. This will open all three images at once. We'll overlay the bear onto the image where the archer girl. We'll also replace the bear's head with a much fiercer and bloodthirsty one. First, we'll work with image 16. Click on it. If you look at image 19, where we intend to place the bear, it becomes clear that the background is totally different. If we want a quality b***d, we shouldn't just cut out the bear but also the surrounding area. However, this background is too different from the other. We need to generate a new one. Click on Select Subject at the bottom. This selects the bear. The selection looks great, but we don't need the bear, we need everything else. In the top menu, click on Select. Then in the dropdown, click on Inverse. This inverts our selection, selecting everything except the bear. Now we can regenerate the environment. Click on Generative Fill in the prompt field type in forest. Once entered, click on the Generate button. Let's see how the image is turned out. The first one will do. Let's cut it out from the side menu. Choose the rectangle tool and make a large selection around the bear in the top menu. Click on it. Then from the dropdown, click on copy switch to image 19. I don't want to place the bear next to the girl. So we'll enlarge the image. Use the zoom tool to shrink the image for a better view. From the top menu, click on Image. Then from the dropdown, click on Canvas Size in the pop up window, change the width to 2,700 pixels. Since specify which direction to increase the image size in the canvas size settings, it centered the archer girl. Let's fix that. From the left menu, select the move tool and drag the arching girl to the left. To be able to edit the background layer, we need to unlock it. Click on the lock icon in the bottom right corner. Now you can drag the girl to the left side of the image that looks right. Now let's Paste the bear. Click on Edit at the top, then from the dropdown, select Paste. The bear is too large, so we need to reduce its size. Click on Edit in the top menu. Then under Transform, click on Scale by grabbing the corner of the selection scale down the bear and drag it to the right corner of the image that looks good. The color of the bear's image differs slightly from the archer girls. Let's make some adjustments. Instead of adjusting the bear, I'll tweak the girl, make sure layer zero is selected in the top menu. Click on Image. Then within the adjustments tab, fine tune the settings, adjust the brightness and contrast. I'll also slightly modify the curves, vibrance and saturation. Finally, adjust the levels that looks much better. Now let's add the aggressive bear head to the bear switch to image 20 in the left menu, select the lasso tool and use it to draw around the bear's head. Then in the top menu, click on Edit. And from the dropdown, click on Copy, switch back to image 19, and paste in the bear's head. It's too large, so we need to scale it down. Before doing that, we should rearrange the layer order. As you can see, the bear's head is behind its body. Grab layer two, and while holding down the left mouse button, drag it above layer one. Now in the top menu, click on it. Then in the drop down under transform, click on scale. Scale down the head to fit the bear's body. It's ready, but the hardest part is yet to come zoom in on the bear's head. Using the zoom tool, I've reduced the opacity slightly to ensure that the head aligns perfectly with the body. It looks good. Now we can remove the original head of the bear. Turn off the visibility for layer two. Then on the left hand menu, click on the Eras tool. Remember to select layer one as that's the one we'll be working on. Now let's begin erasing. Once you're done, turn on the visibility for layer two and turn off the visibility for layer zero. And right click on layer one. Merge the layers using the merge visible command. The next step is to b***d the new bear's head with the bear's body. Using the generative fill, make a selection around the bear's head using the lasso tool. Once done, click on the icon next to Generative fill and choose border Selection. If you've selected, like I did, a pop up will prompt you to enter a value type in 34 pixels. The aim is to cover the gap between the two bear parts. It's okay if it's not perfect. I have some missing areas too. For those, use the lasso tool to add areas to the selection draw around them while holding down the left shift key. Holding the left Alt key will let you remove areas from the selection. Make sure not to select the center of the bear's head. Once done with the selection, click on Generative Fill, leave the prompt field blank, and just hit the Generate button. Our images are ready. They look decent, but there's a small issue. The bear has lost its ears. It looks odd, even from a distance. We definitely need to fix it. Merge the layers, but ensure layer zero isn't selected. Once merged, draw two ears for the bear using the lasso tool for the second ear. Don't forget to hold the shift key. Click on generative fill type brown bear ears in the prompt field and then hit Generate. I had to generate a number of ears to get a decent one, but I'm still not satisfied with the bear's left ear. Let's merge the layers then, using the lasso tool outline the bear's left ear. Click on generative fill. This time I generated the ear without a prompt and got a pretty good result. Now that looks good, let's merge all the layers using the merge visible command. Don't forget, restore the visibility of layer zero like me, choose the lasso tool and start outlining the image. Make sure no part of the bear is selected. After outlining, click on generative Fill and hit Generate without entering anything in the prompt. The images are now ready. The second image turned out perfect. We can now merge all the layers. The image already looks great, but there are a couple more techniques we can apply to enhance it. Let's create a partial mask in the left hand menu. Click on Edited in Quick Mask mode. In the top menu. Click on the Edit tab and select Fill from the dropdown in the window that appears choose color for the contents in the pop up. Adjust the blue channel to 7% Confirm by clicking okay, and exit the Quick mask mode. Click on Generative Fill. Click on Generate. Without a prompt, the AI will recreate the entire image using its patterns, making the generated image more natural looking than mine. But I've set the partial mask to only 7% It only tweaks it slightly. Both the bear and the archer girl are distorted. But doesn't matter. We just need the lighting and colors from this image. Set the b***ding mode for the generated layer to hard light and reduce its opacity to somewhere 20-30% With hard light, the images gain a strong contrast and the difference between the two layers becomes less noticeable, making it look slightly more natural. We can add more partial masks if we like, but overdoing it might distort the image. Our image looks fantastic, but it's still not perfect. To learn more about refining this image, check out the how to fix lesson in the advanced class. I hope you enjoyed this lesson. See you next time. 35. New Feature - Reference Image: Recently, Photoshop has introduced a new feature called the reference image option. With this new feature, we can provide a reference image to guide the AI in generating the desired image. Let's try it out in practice. We will generate this women's top onto the model we saw earlier. It's important to select exactly as much of the original image as needed. For this image, we need to pay attention to the shoulder part of the women's top. Let's start by selecting the area in the original image where we will generate the new top. After clicking on the generative fill button, an icon will appear next to the prompt field. Clicking on it will bring up the reference image option. Allowing us to choose a reference image. I will select the women's top we viewed earlier. Even with the reference image provided, it's useful to add a prompt for more precision. I'll just type Top and click the generate button. I chose a fairly complex and detailed pattern for the reference image, to see how well this option performs. Three images have been generated. They are not bad at all. Although we can see that they differ from the original image and are not exact replicas. Currently, this feature is not yet capable of precisely copying more complex objects. For such detailed and intricate items, it can only replicate the style. Let's generate more images. These are also quite similar. The style is captured, but it's not identical to the original. It's worth noting that the entire reference image is used as a pattern. This is why a necklace appears on the generated images. Better results can be achieved if we remove everything else from the reference image. So in our case, only the clothing should be in the picture. Now, let's try something much simpler to see how the AI handles it. We will use these glasses as a reference. The image contains only the glasses which have a unique but simple shape. I mark the location for the sun glasses on the original image and provide the previously shown glasses as the reference. The result is much better than with the top. The third image is almost identical to the reference image. However, compared to the previous example, it looks less realistic in the photo. The prompt processing is still not perfect while you can experiment with prompts. My experience is that more complex prompts result in a greater deviation from the reference image. Therefore, it's better to refine the image afterward through post processing to make it more realistic. I hope you found this video useful. Enjoy your photoshop experience. 36. New Feature - Remove Background: A new remove background feature has been added to photoshop. With just one click, Photoshop can detect the subject and background of an image and remove the ladder instantly. Let's try it out. Click on the new remove background button. The background has disappeared. Let's check the result. I'll zoom in a bit. Due to the stray hair, this was a challenging task for photoshop. In a few places, the cutout isn't perfect. But overall, I'm satisfied with the result. After removing the background, we can generate a new one with another click. Click on Generate background. In the prompt field, type a simple prompt. I'll just type mountains and a river. When generating the background, photoshop considers the lighting and direction on the subject and its colors to create a background that fits well. Since this is an urban image, it's not an easy task. The results aren't perfect, but maybe the second image is the best. I tried several simple prompts, but the results weren't convincing. This might change over time. To get the best background for your image, it's recommended to provide a prompt similar to the original background. This graffiti brick wall image turned out particularly well. I hope you found this lesson useful. See you next time. 37. New Feature - Generate Similar: In this video, I will introduce a new photoshop feature called Generate similar. Let's see what this new option can do. First, select an area where you want to use the generative fill feature. Then click on Generative fill. I will give a simple prompt Gentleman's hat, and then click Generate. Three generated images are now ready. The second image looks very good, but let's work with the white one. I select the image with the white hat and click on the three dots next to generate. In the drop down menu, you'll find the generate similar option. Click on it. You can see that all three hats are now white and very similar to the original. Let's try the same with a black hat. Now, all three hats are black. However, there is currently no feature to specify how similar the generated images should be to the original. Additionally, there are other limitations. Let's change the prompt to party hat and click on the generate button. Select the second image and click Generate similar a few times. You will notice that our party hat starts to look more like a crown. This shows that generate similar does not use our prompt. It only samples from the generated image and begins to alter it. I hope this video was helpful. See you next time.