Take your Photoshop Skills to the next level with Generative AI

1. Introduction: Welcome to this course on generative AI in Adobe Photoshop. My name is max and I'm a photographer and content creator. Before we're five years of experience in the industry. In this course, we'll be exploring the use cases of generative AI in Photoshop and how you can use it to create stunning at unique results. We will start by diving into Adobe Firefly, the Generative AI engine in Photoshop and explore its various features and tools. Later, we will work on real-world examples like removing objects from a photo, adding new elements and expanding pictures. Or if the power of generative AI, whether you're an experienced Photoshop user or just starting out, you can get access to the Generative AI features of Photoshop by downloading the newest Photoshop Beta. And if you don't have a subscription, don't worry. You can make use of a seven-day free trial and follow along with dead version. You can even try generative fill on the Firefly website for free. So let's get started and take your Photoshop Skills to the next level with Generative AI 2. Firefly: Hey, there. In this video, we'll be talking about Adobe Firefly, the Generative AI engine behind the new Photoshop features. We will talk about what it is, how to use it, and what to use it for. What is Firefly? Firefly is a suite of AI models that can generate texts, images, and other media in response to prompt the user. Initially, it is focused on image and text generation, but later it will expand to video creation, 3D workflows and much more. The image creation engine is comparable to Dolly mid journey as stable effusion. You can access Firefly by going to firefly does adobe.com, and selecting the text to image mode. There, you can look at some examples created by the AI model. And when you hover over the image, you can see the prompt that was used. This is perfect for getting an idea of what kind of prompts cost the desired results. When it comes to text, to image prompt, there are some general guidelines to follow. Text prompt should be at least three to seven words long, detailed and concrete. They should include a subject, a location, descriptors, aesthetic and style keywords can add the finishing touches to the image. So in conclusion, the structure for the prompt should be subject, location, descriptors, and style. This structure doesn't matter for the result, but I find it quite useful for remembering what I should include in a prompt. So as an example, let us try to recreate this picture as close as possible. At first, we will use a general prompt and Z the results. So let's start by entering woman sitting on a beach into the text field. Then we click Generate. As we can see, we're getting some pictures of a woman sitting on a beach, but they're not replicating our picture at the results are not consistent. So let's try a more detailed description following our guidelines and structure, let's enter side view of a wide young woman sitting on a TVA on a sandy beach against blue sky with a lake in the background. Woman wears white clothes, has brown hair. Looks relaxed. Photo 50 mm. They can add in our and there you go. The results are much better and much more consistent. With some jobs in Photoshop and Lightroom. We could get this picture very close to our initial photo. At the right side of the window is an interface that you can use to refine the results. You can change the aspect ratio, pre-select the style, and select colors, lighting, and composition. There are some limitations to the image creation engine. It is difficult to get realistic looking hands and feet. The results are not photo-realistic. It looks rendered more like a concept Art. So keep those limitations in mind when creating in Photoshop. Now, let's talk about some use cases for firefly. It can be used as a text to image engine for brainstorming looks, ideas, pictures, to apply styles or textures to text, to recall a vector artwork, and much more. Furthermore, it is natively integrated into Adobe Photoshop, but that we will explore in the next few videos. Thanks for watching. See you in the next one. 3. Removing objects: Hey there, welcome to this video on using generative field to remove objects from an image in Adobe Photoshop. In this video, we'll be exploring how to use Generate, fill. The difference between generative and content aware fill real-world use cases and examples. Firstly, using generative field to remove objects is fairly simple. Just select the area you want to remove with the Lasso tool or any selection tool you're comfortable with. Then on the new hot bar, click Generate, Fill or go to Edit, generative fill. Leave the prompt blank and just click Generate. And there you have it. They objects should be removed. Now, let's talk about the difference between generative field and content aware fill. Let us try to remove a subject from an image using both methods. This image, we would like to remove this player. So let us click Select Subject. Then we will, by holding the Option key, will remove this player from our selection. And under this menu, we can expand the selection by ten pixels. So the AI has something to work with. Then we will go to Edit Content-Aware Fill. And there you have it. We will output it to a new, a new layer. And this the result of the Content Aware Fill. So now let us try the same method. But with Generative, we will select this object. Once again. Remove the player with the ball, expand the selection, and click Generate, Fill, generate. There you have it. This is the generative fill layer. And this is Content-Aware Fill. Under the Generative layer, we can also select some variations to see if another variation suits this use case better. But with the variation, they invents some kind of poll back there. And that is not what we want. So this is the final result. Overall, we can say generate the fill is perfect for recreating natural patterns and complicated backgrounds. While Content-Aware Fill is better for smaller objects, we have an even background. Generative AI tends to invent stuff just like with this poll. So often it is faster to use Content-Aware Fill. So now let's look at a few examples. So you get the idea on this first image. I would like to remove this flower from the image. So let's go ahead and select it with the Lasso tool kits like so, generative, fill, generate. Let's look what kind of results. But again, just like that, we have a perfect re-creation of this cake. Now let's look up some variations. Here. Here he is trying another kind of cake. Maybe that's better. And here another, there's a difference in the background. So let's look at the background and the variations that think. The second one, I like most. There you have it perfectly recreates the cake. And with the background it has some problems, but pretty much it is flawless. Let's look at another example here. I would like to remove this subject and the reflections in the water. So let's see what AI does. I'm selecting subjects, giving it a little bit of room to work with. And then we will add the same selection, we will add the shed. So okay, let us try generative fill, generate. Now I'm really curious if it is able to recreate this shadow when the water. But let us see. And there you have it. She's gone perfectly. Let us look at some variations. Find the second one the most realistic. Let us turn this layer off and on so we can see what the AI does. Pretty impressive. In this example, I would like to remove people from this chart. So let us go ahead and select this object. Expand the selection by 20 pixels. Okay, generative feel. Let's go. Okay. What is AI trying to do? See this happens sometimes where I select this object and click Generate, Fill. And on some subject it completely removes them. And on other subject it does something weird. So let us try entering a prompt and write remove the subject. Let's try again. Okay? No, it doesn't want to remove those subjects. So I think that would be a case for content aware fill. So as our last example, we're trying to remove those knives from an image. This time we're going to use the polygon lasso tool to get some straight lines. Drawing a rough shape around the divs that generate the fill degenerate. Okay. We can see that it doesn't understand, but we'd like to do it, puts some other knives there and not removes it. So let us try entering a prompt and writing, remove the null. Generate. Okay, there you have it. Removes the knives and replaces it with a little leaf. Just awesome. Here's the before. Here is the after. In conclusion. The new generative field is a powerful tool that enables edits. Take a lot of time and knowledge. In seconds. It will only get better with time because right now the results are a bit and consistent. On one picture, it works great. And on another one, it completely screws up. So that is all on generative field for removing subjects. In the next video, we will explore how to expand a photo. Thanks for watching, and I'll see you in the next one. 4. Expanding images: Hey there and welcome to this video on using Generative fill to expand the aspect ratio of images. In Adobe Photoshop, we'll be exploring how to use it, some real-world examples and discover the limitations of this powerful tool. So how do we use generative field to expand pictures? It's fairly easy. First, use the Crop tool and select the desired aspect ratio while holding down the Option key. Drag out a frame and confirm the cropping. Then with the rectangular tool, select the image, the selection, and choose generated film from the hot bar. Or I go into Edit generated feel, relieved the prompt blank and click Generate. Just like that. Picture is expanding. So let's take a look at some examples. So in our first example, let us try expanding. So first we select the Crop tool and our aspect ratio and go for 16 by nine. And try holding down the Option key. Let's expand. Okay, Let's go little bit wider so we can place the phase of the woman On a thirt. Confirm the cropping. Select this object. Inverted generative fill, generate. There you go. That picture is X band. And it does a really good job. Firstly, it expands those hills in the background and those trees. But also look at this toggle right here. It expands it perfectly and also an expense, the shadow of the woman. So pretty much this is a perfect result. Let's look at some variations. Okay, they're pretty minor. I would say the first is the best one. Now on our second image, Let's try to expand this one. And this is a classic example where this photo was taken for social media in a vertical aspect ratio. And if we would like to use this image on, for example, a website or for printing, we would have to shoot it horizontally. And after fact, we could do nothing to recreate it. Now if there's generative fill tool, Let's try expanding this picture to a 16 by nine. First, we unlock the layer, just like we've said, select 16 by nine. And on this we'll go really wide. Just like that. Selecting image, Burnet, generate a fill. Let's go. There you go. The picture is expanded. It does a really good job. Here's the before, here's the after. And it recreates the city perfectly and expands the mountain. Nobody could tell the difference. In our last example. Let us take a look at this food picture. And I would guess that AI has a hard time recreating cake, the chicken, the whole table. So let us try expanding it with the crop tool. This one, we go pretty wide so we can free, can see if the eye is able to do that. Okay, now I'm impressed. Look at this picture. It perfect, perfectly recreates the chicken, the cake, knives. And it even creates a kitchen or some kind of living room in the background. Let's look at some variations. It's also pretty good this morning. But I like the first one the most. If you look closely, we can see some kind of glitches right here in this area, but we could fix that pretty easily. But overall, I would say, here's the before, here's the after. It is pretty flawless. In conclusion, we can say that it is a really powerful tool. Out of all the use cases, I find this the most reliable one. The results were astounding and pretty flawless. Sometimes there are some glitches, but they are pretty minor and easily fixed with some touch-up. So that is all on expanding photos. In the next video, we will explore how to add elements to a photo. Thanks for watching, and I'll see you in the next one. 5. Adding elements: Hello and welcome to this video on how to use Generative fill to add elements to a photo. We will discuss how to use it, discovered and limitations and find some real-world use cases. Alright, How does it work? First, you need to select the area per you want to place the subject. Keep that I mentioned of the elements you want to place in mind. After you have made your selection. Choose their generate the fill option, enter a concrete and detailed prompt. And to remember the general guidelines for prompts that we've discussed earlier in this course. Click Generate. You can choose from three different variations. Feel free to click Generate again for different results. You can also change prompt under Properties and click Generate again to refine your results. So let us look at some examples together. So in this first example, I would like to place a boat in the foreground. So let's start by selecting an area within Lasso Tool. Click Generate the fill, and then we will describe fishing boat. Okay, let's look at some variations. The first one looks pretty solid. But if we zoom in, we can see that it has a hard time recreating the people and faces. And overall doesn't look really real, but the composite is really soft. So in this picture, I would place a rusty old car in the background. So let's set it back. Area. Describe what we want. Old, Rough the yellow car. What could we use to describe it in more? Let's say, placed sideways with some leaves on it. Generate. Let's go. Okay, There we go. Let's look at some variations. What, this looks pretty awesome. Look at it. Okay, if we zoom in, there's this kind of on realism that we already found on with the Firefly texts to image engine. But from a far away, it looks pretty awesome. So let's do one more. And let us try something difficult. Let's try placing some humans right there. We're selecting that area. Then we say young woman hiking close and a yellow jacket. Back to the camera. And I have purposely entered this prompt with her bag. Turn to the camera because I know already that it would screw up the face. It's completely. So this is a little trick that you can use to create some kind of decent results. And let's expand the selection a bit more. So the I has something to work with. Let's click Generate and let's see what it does. Okay? Does what we've told it to do, but it doesn't do. This one looks really good. Let's zoom in. There. You have the problem. The phases are completely distorted, but from a far, it looks pretty solid. Let's try to create a face that is more realistic. Selecting it once again and describe A face. But I don't think that AI is able to do that. We're phases. It has a pretty rough time. They have it. It is able to do it. The variations are more unrealistic. So I think with humans, it has a pretty hard time. Let us try something else. I would like to play. Beer in the back. So let's just enter. Okay. That is somewhat decent. The first one looks pretty good. Resume in, yeah, it doesn't understand that shallow depth of field of this image, but the places it in a bit of fog. So it looks good. But same thing with animals. The photo realism. Isn't that good? So now let's try something different. I would like to replace this top with a dress. So let's select loafing and be careful with the hands. We have removed them from the selection because just like we've said, it has a hard time recreating them. I will click generated fill. Let's describe it. A white dress for a degenerate. Okay, There we go. The clothing looks pretty flawless. And just like I've already mentioned, it has a hard time recreating those hates. Let's look at some variations. Just looks awesome, Pretty good. In this picture of the hands are screwed up even more. Variation tree. It does also were pretty good job. But I think he has the Nicolette is connected to anything and it has some weird glitches right here. And there's always, the hands are pretty distorted. Now let's try something else and change the hairstyle and column. Send back. Now we'll extend it to right here. So DI has something to work with because we're going to replace it with log here. So let's describe it along. Lone pair, generate. And there you have it. She's blonde. Look at some variations. Second one looks pretty solid. Let's domain. Yeah, it does a pretty good job. Even creates new ears, which are a little bit distorted. Overall, this is a really powerful tool. However, there are some limitations. Just like with the text to image engine. Generative field has problems with photo realism and has a hard time creating feet, hands, and faces. It also sometimes places background elements that aren't correct. Composites are good, but nearly all of them need a touch-up. Lastly, it can be difficult to get consistent results. And often it does not pick up on prompts. Use Generative effectively. There are some general guidelines to follow. You should experiment with different kinds of synonyms. For prompts. Some work better than others. Your prompt should be as detailed as possible. Alter your selection and it makes often a huge difference. Plays hair always as the last element. Be careful with big changes and stay away from faces and hands. With dead set. You can use it for background elements such as birds, boats for animals, and people in the distance. Sometimes it can also be useful to change the clothing of your subject and altering the hairstyle in color. That is all on adding elements. I hope you enjoyed it and I will see you in the next one. 6. Possibilities, limitations, policy: Welcome to this video where I would like to summarize my experience if the Generative AI in Photoshop, Let's start with the possibilities. All in all, it is a truly powerful to with a lot of possibilities for brainstorming, concept Art and real-world use cases. They expand feature is truly impressive and trivia tool that I will use a lot. Also, the feature for removing objects looks very promising. But adding objects at the moment is a bit inconsistent and may not be suitable for frequent use. However, keep in mind, it is only a Beta and Adobe is working on it. So we can expect it to get better with time. Despite its strength, there are some limitations that we need to consider. As we've already mentioned, humans, animals, and major changes to the image are difficult to accomplish. Also, the tools behave and consistently at times, which is currently its biggest problem. But we can hope that in future versions, we can tweak the results with more parameters and get even better results. Now, let's talk about the policy of this feature. With other texts to image models. There are concerns about the dataset that is used, and there is a lot of discussion about copyright infringement. However, Adobe uses photos from Adobe Stock and openly licensed content to train their AI models. They are also planning on paying contributors of Adobe Stock for their contribution to the AI model. As of June 2023, generated field is in beta, so commercial use is not possible yet. But as soon as generative fill leaves the beta phase, it will be available for commercial use. All in all, Adobe Photoshop generate the Fill feature is a promising tool for creating stunning images with ease. Despite its limitations, it has the potential to be a game-changer in the world of image editing. I hope you enjoyed this video and I'll see you in the next one. 7. Conclusion: Hello and welcome to the last video of this course. It's been a pleasure to have you. Throughout the course, we've covered a wide range of topics related to the Generative AI features in Photoshop. I hope you've enjoyed it and learned a lot from the videos. And I hope this has opened your eyes to what modern technology is capable of achieving. If you want to dive deeper into the topics we've covered, I highly recommend reading the blog posts from Adobe. They are a great resource to learn more about the latest versions and features. Finally, I would like to thank you for watching this course. I hope it has been a valuable experience for you. If you have any feedback or questions, please don't hesitate to share it. We always appreciate hearing from our viewers. We hope to see you in future courses where we will continue to explore the latest trends in modern technology. Thanks again for watching and have a great day.