Designing Impactful Posters with Adobe Photoshop & Generative AI

1. About the class: Forget starting from scratch. What if you could build powerful visuals and design using AI generated assets without ever giving up creative control? I am Martin Parak an art director with over 20 years of experience teaching Adobe design tools and working as a creative for brands like Disney, BBC, and Unilever. Let me show you a brand new workflow, a completely new design approach where AI generated images form the foundation of your work without ever sacrificing your creative vision or authorship? We will combine the power of Adobe Photoshop with cutting edge generative tools like Adobe Firefly, Journey, Magnific and hGPT to turn a simple sketch into print ready posters in five distinct visual styles, ranging from illustrative to photo realistic. You will learn how to prompt effectively, remix AI outputs and unite everything in Adobe Photoshop. So you can design in any style you can imagine and ten X your productivity. All you need is basic knowledge of Photoshop and a creative cloud subscription, which includes Adobe FireflGenerative credits. The other AI tools mentioned and featured in this course are optional and not required to complete the class project. So are you ready to level up your creative workflow and future proof your skills enroll now and learn how to design without limits? 2. Research & Understanding the Brief: Before I walk you through the process that went into creating these final compositions that you can see on my screen, the different visual directions, all based on my initial pencil sketch. I wanted to spend some time showing you how I prepared everything for this project because every graphic design project should start with some research and understanding the brief. So in this case, here on the board, I have the information or most relevant information for what this poster needs to communicate. And this is a curriculum about gold mining in the Amazon and its impact. So the first lesson is exactly that, the impact of gold mining. Then we have rainforest ecology, biodiversity of the region, climate change, indigenous communities, and so and so forth. And this is from the client. So the client is Amazon Aid. It's charity, and this is a curriculum they are planning to introduce for high school students. And the main point of this project is to create an engaging visual, a poster that can be placed in the classrooms to get students excited and thinking about this problem. So the first way I used AI for my research is to turn these lessons into keywords. For this, I just use chat GPT, and I asked ten main terms that I should be thinking about when it comes to each of these topics. And out of those I highlighted the ones that I felt like can be depicted easily visually. There's a couple of things here that are much harder like nutrient cycle, the ecosystem and toxic runoff and things like that. I felt like it's not necessarily something I can include in the composition. But using CheGPT and asking for lists of key terms that you can then turn into something like a mindmap can be extremely useful. And then additionally, I wanted to get some help again, using hechPT to come up with a couple of motifs that I could incorporate in my design. And it found some really useful ones, and it even references which particular Amazonian tribe or region it is connected so there's the spiral, the serpent pattern, leaf or vinee patterns, zig zag lines, and so on and so forth. Then I also wanted to get a list of the most iconic animals and plants. And this is how I got to know the name of this particular tree, the kapok tree, which has a lot of other names as well. And I will talk a little bit more about this, but that became a very important part of the project. So I wanted to feature that. And also, I wanted to feature a jaguar so after doing some research with AI and gathering some information, I also still like to look for actual real references for my projects. So in this case, I make sure that I'm not using AI. And when you're searching on Google and other sites, you can just include in your prom that you want to find examples of something that was before 2021 or 2020. And usually that's going to help you to avoid seeing AI examples or generated images. Now you can see I have a few examples here of indigenous people and their celebrations, and this is a great reference just to be more familiar with the features of people as well of the region, what they wear. Obviously, it's a mixture of modern and traditional clothes that I'm looking for. And then I also did more research on the patterns and motifs that they would use. And I found this particular interest board extremely useful collected by San who did this research on Amazonian tribal art. And again, there's a lot of good references here that I used as inspiration for the project. Now, you remember I mentioned the kapok tree, which is also called Lupuna in that region or Sabo, and there's a lot of other terms for it, as well. It's even called drunken tree. And I made sure I find enough references for it for both its canopy. So the leaves and the branches on the top, and also its trunk and its roots. Now, the roots are especially unique, and this part is the one that is really recognizable. And the reason I know that is because I actually been to a couple of places around the world, where I came across these type of trees, or I believe, at least it's some type of similar trees to this particular one that I wanted to use for this composition. Now, this picture was taken in Costa Rica, so it's not the Amazon region, per se, but it is quite close to it, and it was just a gigantic tree. The vinees that were hanging from it were already as big as having both of me and my wife being able to sit on it. And yeah, it was just a very impressive tree. I've also seen smaller ones, but still with very impressive roots in Australia. This particular photo was taken in Sydney. So the reason I'm showing these examples is because a personal experience always translates the best to a composition. So if you can recall anything that you personally experienced, something that you could actually touch, smell, and feel, that is always going to inspire you the most. So when you're doing your research, collecting some photos that you took or even notes you took or sketches you made, these can all be extremely useful for your projects. So don't forget to include them. I also knew that for this composition, I will have to depict the deforestation and the mining and all of these dangers that's affecting the region. And for that, I did a bit of research on the various vehicles and machines that are used for cutting down trees, like modern machines and obviously moving ground around. I had a lot more of these examples, just including a few of them. But more importantly, I also wanted to remind myself that there is an amazing opening credits for the show Yellowstone. So it shows really well the issues that will need to be covered visually in this composition. So in this first one, we can see obviously the chimneys and the construction. We can see the mining equipment and digging the earth and then forest fires. So if you haven't seen this opening credit, I highly recommend to check it out because it's very graphic and very cool. That's something that you can easily translate to your projects because it's mainly relying on silhouettes and just have this duotone or monochrome style. Reason why I included it is because, again, it is something that inspired me even before I started working on this project. And it's just a reminder to have that in the back of my mind when I start working on these posters. 3. Initial Sketch: Let me talk a little bit about the sketch that I created for this project. I always like to start with a sketch once I've done my research and once I have a vague idea of what I want to achieve. And that's what you can see here. And I actually drew most of this apart from the Jaguar and the cow. I thought it would be just faster if I quickly edit some sketches for which I actually used AI compared to me drawing them. So I just edit these, even though I probably would be able to draw them, but I just found it much faster to do it. And I already had this sketch of this woman that I created for another project that I borrowed. So that looks to be the most detailed one out of everything here, obviously, apart from the AI generated sketches. But, yeah, this is something that I felt like could work quite nicely here on the right side of the image. And then you can see the rest of the sketch details are much more rough and more vague because I just did it very quickly. I used procreate to put all of this together. It took me probably around 5 minutes, so it was very quick and dirty. But it captures the main idea. And the main idea is that I will have this giant kapok tree in the center and it's going to be used as a divider between the left and the right side of the composition. So on the left side, we will have the outcome of all the negative things that is happening to the region. So the threats will be here, while on the right side, we will see as a big contrast the untouched lush tropical region with its protectors, the indigenous people, and these iconic animals and plants that we already talked about. So having this contrast created with that main element, the tree in the middle, I felt like it's a strong composition, and there's a lot of room to play around with this, but this was essentially what I wanted to achieve. And I always like to leave myself space for exploration, so I don't try to really restrict where I am going to place things. It's more like a visual reminder of the general idea and direction. This is something I would be referring back to throughout the workflow. So it's easy to forget certain ideas that you had in the beginning. That's why it's good to have that always somewhere at hand so you can reference it. It's also important to mention that sketches can actually be used for generating images because we can use it as structural reference, for instance, in Adobe Firefly, we can also use it as image reference in mid journey, which would count as a structural reference. But I'm going to go into this in more detail, so I will show you how this actually works. Most importantly, you just have to remember that creating a rough sketch is extremely valuable for the entire process. So even if you're not good at drawing, I highly recommend to do it, even if it's just very rudimentary, like the way you can see the trees depicted. It's extremely simple. As long as you have things in the right proportions and roughly in the right placement, that's going to help you to realize your ideas much better. 4. Visual Direction - Illustrative Stylised: Let me talk a little bit about the four visual directions that I chose to create. I wanted them to be obviously recognizable that they are the same composition, but I also wanted to make them slightly different and really give the client options to choose from. And personally, I feel like most of them turned out really well, and I'm also excited to see what we will be able to come up during this course as a fifth direction. Taking a closer look at each of these, I can explain more what I wanted to achieve. This first one is the more stylized illustrative design. And once I open this in Photoshop, we can take a closer look at the details. So I really liked the textures in this and also the colors. I think the color palette works really well. It has all the important elements or ingredients that I felt like will be important for the composition. I actually ended up having this sun in the background, that big disc coming up as a happy accident. It's not something that I used in my prompt or even in my original sketch, but I ended up using that throughout all of the designs because this was the first one that I created. And out of all the visual directions, this is probably still the least complicated one or we could say more minimalistic design. And what's unique in it compared to the other directions is that it heavily focuses on a cow. And that's because one of the main threats to the region is the cattle so the fact that they are cutting down the trees and creating more land for these cows to graze and to live on. So that's something I wanted to feature here, but I feel looking back that this is probably a little bit too dominant. It's too intense to be that big. It probably would make sense to switch these two things around, so to have the vehicle much bigger and the cow to be maybe somewhere down here. So if I were to do this again, I would probably swap those two things around. This is one of the things that can happen easily when you are using AI and you end up heavily relying on something that was generated that you might forget that that wasn't your initial concept. So in this case, to have that much prominence on a certain element. So let me show you how this composition was put together. And you will see that I use chat GPT with the following prom that you can see here on my screen, and I was using these two images as input or references or the sketch and this composition that I created in Mid journey. So this is actually something that I was experimenting with at first and I really like the style. I really liked the abstract shapes and the halftone pattern. And I was initially thinking of creating something like this, which actually ended up being another direction. But this was used in Chet GPT, and this is the resulting image that was generated. So let me just open this up. And you can see it's very similar to what ended up being used in the poster with a few changes which I'm going to talk about. Again, back to Photoshop, I can show how this evolved from what you can see on the left side to the final version that you can see on the right side. So the first and obvious change was that I moved the Jaguars head from here to down there, and then obviously, that's where it ended up. Then I also had to improve how this harvester tree harvester machine looks like. So in the original design, it looked like that, but then obviously it ended up looking like this. So that's something that I improved separately, and I moved back. I also wanted to make sure that these birds look more tropical. So I replaced them with more like maca silhouettes that we can see here on the right side. And then one additional thing is that these tribal patterns that were in the initial generated image, I ended up using in more subtle ways so we can see them being translated around the design. So there's one element. Here. There's another one up there, and then this one is down here. So they are still used, but in smaller size. And then one other interesting thing to note is that I also noticed that this shape in the background, if I look at it like that, it would be not like a perfect circle. So it looks more like an ellipse, and that's something that I had to adjust. So here you can see that shape is coming more in and it becomes more like a perfect circle. I had to reduce that shape, and I also got rid of these lines because I didn't feel like it was necessary. You can see, this was probably the easiest and smoothest transition from what I got from a generated image to what ended up being in the final composition. Of course, I still had to upscale and add more details on it because the original one was fairly low resolution. So if we look at the woman, we will see that at this stage, it was still quite low resolution, while in the final version, we have more detail on it. However, whenever you are enhancing images, so it's not actual upscaling. It's more enhancing with AI. You always end up having slightly different details on certain areas. So certain bits can be lost in that transition. And I can actually show you this transition within magnific the tool that I used. So we can see the before and after version as I'm moving my slider over it. So certain details actually improved like the jaguar, I feel like looks more detailed and more interesting. Some textural details like on the face, I feel like the original version had better details, but it stayed quite close to it. And the settings that I used here, I can see was to try to keep it as close as possible to the original. So I increased the resemblance and I reduced the creativity. In magnific, that's a way that you can try to control it and make it as close as possible to the original. And I did four times upscale, so it got much more detail. So if we zoom even closer, we will be able to see that it added a lot of detail. So even on these parts when there's not much going on, it adds a lot of detail, and that's great because we will be able to use this image in a much larger print size. And that's what we can see here on the right side. So if I do actual pixel size, this is what we got while the actual pixel size for this one and this one was this size. So yeah, this format, by the way, is if I look at it in inches, is 24 by 36 " ready for print with 300 PPI. So, yeah, that's that first composition, and this is how I ended up creating it. So the main tools here used were HGPT for the initial generation, then Photoshop for moving things around, like the Jaguars head, which went down here, as I mentioned before, and then finally magnific to upscale or enhance the image. And then, of course, again, photoshop to add all the typography that was needed. So the details on the top and the bottom. 5. Visual Direction - Montage Abstract: Now the next visual direction, which was my initial idea or concept, was this photo montage, which is slightly more abstract and more graphic, which will feature half done patterns and a bit of distress on some of the details and also some geometric shapes and very vibrant colors. This is something I ended up creating in two different versions, the more brighter version that you can sort right and the one with the darker background. There's some subtle changes between the two. But essentially, here, I felt like it would be important again, to give the client the option to choose from either one of these and notice how the colors or the palette that I used is exactly the same. So I utilize that darker blue as the dominant background color on the left, while this more pale sepia color or vanilla color is the one that is used for the highlights on both of these posters. To see the source of this composition, we have to jump into Mid journey. And here you can see this image, which became the hero image for this composition. And it's a simple prompt that you can see here on the right side. So it's simply just a vibrant montage about the Amazon Rainforest featuring the portrait of Indigenous woman and the Jaguar. So that's the main key elements. Instead of trying to compress all the ideas, including the tree and the deforestation on the left side, I really tried to focus on this focal element of the composition, the hero piece within the composition. And that's a very good practice. So instead of trying to do everything all at once in a prompt, try to simplify and maybe just focus on really something that you want to emphasize. And in this case, I also used four reference images. One of them is my own illustration. So I felt like this illustrative style would translate well. And there's a couple of additional ones here. So we have this one, which is montage or collage, and then we have this, which has these really cool textural details in it and also quite cool colors. And then once again, another one that's a bit more illustrative or painterly and slightly surreal, which I felt like could work really well. So when you adding multiple style references within Mid journey, I always comes up with this really cool blend of those styles. So we can see the resulting image has a bit of all of those. So I feel like mid journey is the best at combining multiple visual references, and I always like to use it for that particular reason. So we can see that before I ended up creating or generating this image, my prompt wasn't strong enough or it wasn't detailed enough. So it ended up creating these less effective coms or comps that were not really useful. Again, it was mainly because the prompt itself was a bit too vague. And then once I started including more information and also the additional styles, I started to like where it was going. And yes, so this is where it's getting closer to what we ended up using. And I even tried extending this composition just to see what other elements Mi journey would use or how it would continue the image, which we can see here, and then a few variations on the same theme. Even though these were cool, I ended up using this particular one that we've seen in the beginning and yeah, there's some strange combinations here that obviously is not something that I ended up using. By opening this design in Photoshop, I can show you a little bit in more detail what went into it. So first of all, I'm just going to turn off this vignette on the edges. So I show you from a distance. There's a vignette that I used here just to make the corners a bit darker. It's a texture that I downloaded from Free Peek. I'm just going to turn that off, and I will actually turn off all the text as well just so we can concentrate on the composition itself. So this is the hero piece here that we can see. And if I turn off everything else, we can see exactly how much of that ended up being in the composition. Although there are some other AI generated details like the tree, I would still consider that the fully generated detail in this case, is only this part here. Because I'm going to show you how I created the tree. I feel like that is something that it could have been replaced with a photo or just put together multiple images. It doesn't really have to be AI generated, and everything else I designed putting assets together. So in this case, I would say around 30% or 40% max of the composition is relying on a fully AI generated image. But besides this main element, probably the very important detail here is the tree. So if I zoom a little bit closer on that, I can show you how that was created. So the image on the left side was generated in free peek using mystic 2.5 image model, and this is how that turned out. I really like the shape of this, even though the roots were not exactly how I imagined them to look like, but this was a good enough composition. Then I used this image and placed it into ChEGPT I asked to turn this into something that looks dead, with cracked bark and obviously barren branches. Once I had these two, because they were very similar, I made sure in Chachi PT that it had to maintain the scale and the proportions of the tree. I could easily put them together, and of course, I wanted to have the barren branches on the left side, and I actually ended up flipping the tree around because I had more brighter details on the left side, which I wanted to use on the right side. So that's how it was combined from two generated images. And here you can actually see my original prompts as well. So for both the tree and then the dead version that was created in CheGPT. Now, once the tree was placed into Photoshop, I can show you what I've done with it here. So I'm just going to go a little bit further down maybe somewhere around here. I try to make it look more similar to the image on the right with the woman, so I can just turn off these effects, and there you can see the original image. So that would have been the fully colored and, like, photo realistic version of the tree. But then I used the threshold filter, and because it's a smart object, it ended up being a smart filter that I can easily come back to and make changes to. The threshold filter is very easy to use. Essentially, it turns it into a bitmap image with only black and white colors, so not even gray scale. So it's literally just two colors, and you can choose the proportions or the distribution of light and dark within the image. And this worked quite well. So it's like a very graphic version of the original image. And then I started using gradient map adjustment layers and fine tuning it to render specific tonal values to the colors that we've seen in the image on the right. So I used and sampled colors from this image, and you can see here this one, for instance, is as simple as having the vanilla color or cream color assigned to the highlights or the whites, and then we have the blues on the midtones and shadows. But compared to this, we have this additional gradient map, which is then masked out to make sure that it only appears on the left side. This one is more complex. It utilizes three colors, shifted a little bit around. And of course, this is, again, still completely non destructive, so I can go back and make edits to it. But I also felt like the tree needed a little bit more color. So I ended up adding more density of those nice blue colors that we have greenish blues that we have on the highlights on the hair here, for instance. And yes, so that's how it ended up looking like. It's now fully refined, but this is, like, a good version that shows my vision and idea for this composition. And then on the left side, interestingly, this changed probably the most throughout this composition because I really struggled finding the right setup here. So before I had this mining equipment, I actually had a tree harvester here. And I also had another barren ground design for it, but I don't have that anymore, I believe. Yeah, I moved it around. So that's basically the tree harvester that I initially had here. I ended up changing it because I felt like it's not filling in deep space well enough. And because it's more horizontal design and I don't really have much horizontal space, I changed this to this mining equipment, and then I decided to include that circular shape behind it and the tribal pattern in front of it just to separate it from the tree. So if I don't have that shape there, it feels a little bit too blended into the tree. So I just zoom a little bit back here, maybe you can see it better. So that pattern there is just to divide it up a bit, and then I also just included a few dead trees here on the top. So this is more of an abstract depiction or stylized depiction of the deforestation. The mining obviously is more utilizing a photo in this case. So this is actually a photo that I use without the gradient map on it. That's how it looks. And again, it has a very useful filter that I can highly recommend to use if you want to achieve these more graphic styles like what you can see here. So besides using levels, I was also using poster edges. So that's a really cool filter. You probably need to fine tune it to make sure it works, but that's essentially the effect that it can create. So if I show you without it, is how it looks, looks just too clean and too photorealistic. While with that, it immediately looks a bit more like a drawing with those stronger highlights and a bit more texture on the image as well. So that's again, how I managed to bring all of these different things closer to each other. So once again, the AI generated image is the reference that I use for the visual style, and then I used filters and adjustments in Photoshop to make everything work well together. Now, these other assets like this pattern in the background is actually a vector asset from free Peak, and the Macas that we can see here, again, the silhouettes are once again, assets that I found on freepeek. So even though there is AI that you can use to generate complex details like the tree in this case, you shouldn't forget about simply using existing assets that were not AI generated. It's a healthy mix of these things put together that's going to give you more creative control and also better results. And it's also important to mention that in this composition, I realized I tried to fit too much things in. So it was already quite busy. But I had an idea of having more of the indigenous people, really, like in this heroic pose, like the real protectors of this land, featured somewhere. But I realized that it was just too much, and I decided against it in the end because it's already quite busy. And there was a lot of other things that I was experimenting with, but this is essentially what I ended up with. And as I showed earlier, I also created a darker version of this, for which I only had to change a few little details to make this work. 6. Visual Direction - Photorealistic Composition: This poster was also an interesting workflow where I wanted to achieve a more photo realistic result. Something that is more believable, make it feel like it's actually photographs that are put together, even though this is almost entirely AI generated. Again, here I started off putting things together in a lower resolution format. Then I was testing the composition with the type and also some refinements that we can see between these two versions. And then the final version that we see on the right is the upscaled or enhanced version for which again, I use magic I'm going to show you the original PSD file, which has all the layers that I used to create this. And we can actually see the original image that was used to start out this composition. And I can also show you how I got to that image in the first place. So I actually used this illustrative version as a starting point. And then I asked Che GPT to make this composition look like a colorful photo collage, keep the original composition. And so on and so forth. And it actually came up with this. And I continued this conversation in Che G PT, and I asked the AI to remove the cow, make the woman look more like an Amazonian native, at some polluted river on the left side, and make sure that the left side of the tree is completely barren and with no leaves. And then it came up with this. So it improved a lot, and it actually followed my guidance or prompt really well. So I even patted the AI's back saying, This is really good. And I said, Just move the jaguar now behind the woman because it was placed in a very strange location. And then I also just clarified as if they were standing next to each other. I also asked to make the roots more similar to a giant capox tree roots, like in the photo, which I included as a reference. There it is. And once again, from this, we can see what it ended up creating on the right. So it managed to improve the roots. The tree looks more similar to that specific tree that I was after, and the jaguar was moved nicely or repositioned. Now, whenever you remix an image that was already generated with AI, this applies to almost all of the tools out there. You can expect almost everything to change slightly. The phase, again, in this case, changed quite a lot, even though I didn't say to change anything on that. However, these things are very easy to fix because you can just combine these two results in Photoshop. And you will actually see me doing that live in the final composition or direction for this project later in this course. Once I had this image that you see on the right, I took this into magnific and I asked it to enhance it and also make it more photorealistic. And that is what you can see here. So that was the original, and that is the updated version. So we can see how it kept the composition but obviously made the person look more real and everything else, all the details in the background. The tree itself looks photo realistic as well. And this is what we can see here in Photoshop being the starting point of this final piece that I put together. And by turning on the layers one by one, we will see what I've done to this. So I wasn't happy with the face. It wasn't bad, but it wasn't really the type of person that I had in mind. She looks a bit more Asian, like South Asian, in my opinion. So I wanted to make sure this person, especially for this photo realistic composition, looks the part who could be from the Amazon region. And I can show you what I've done here. So I covered up her head with the remove tool. So I just painted over that detail. I also added a kettle of cows using the generative feature, the generative feel. So that looks also better there a bit because, well, if we look closer, these don't really look like cows while this one looks already better. If I come back here, where the faces, will see this image added. Now, this looks very out of place. It doesn't really look connected to the body. The colors are off. But there is a really cool feature called harmonize in Photoshop, currently only available in the beta version in the contextual task bar. So there is harmonize. When you have a layer selected, you should be able to use that, and it's going to help to get the colors closer to the environment, all the other layers. So before and after, this is a cool AI feature as well, which retains as much as possible of the existing layer, and it just utilizes AI do something similar to what would happen if you are using adjustments and filters to match the tonal values and saturation and colors and hues to the underlying layers or the layers underneath. But obviously, because this is using AI, it can also fill in the gaps, and any detail that was completely black or just clipped colors can also be added. And, of course, it's much faster to use it than to actually manually go through all the adjustments yourself. So I really like this feature sometimes I use it with a lower opacity, so we could have a transition between the two versions if I set it to 50%. So that was before and after. It's not as dramatic or with 100%, of course, that's what we get. And by the way, the portrait that you can see in the composition, the one that I used to replace the original one was created in mid journey. So you can see I was experimenting quite a lot and I really wanted to focus on the face. And this was the prompt that I used, and I actually used that image from the composition as an image prompt, which is similar to composition or reference with image journey. And then we can see that this is the image that I really liked. I love the look of her and also the strong features and then I extended this by using the editor within mid journey. So it's similar to generative feel or generative expand in Photoshop. And then I did a couple of additional versions which all look quite good extending the hair. And essentially, I just dropped this into Photoshop, and I've already showed you I've used the harmonize to get it closer to the background. And then I started refining details. So let's just go through what happened here. I was using the remove tool to try to blend the edges a little bit closer to each other. And then I also did the same thing up here. If I zoom out a bit, you can see, I wanted to extend the feta. This is simply just using the remove tool also here on the right side, just blending details together. This is a very good technique. So the remove tool is essentially just for removing blemishes, but utilizing AI. However, it can also be used for this type of thing to blend things together, get rid of very noticeable edges, and it did a fairly good job on most of these details. And then I actually removed these birds as well and some issues there on the sky because I thought I'm just going to add those later separately. And then we have also another layer here. If I just zoom a little bit closer, that's to get the skin tones. Actually, there's just a little detail there that I fixed. And then the skin tones is also fixed on a separate layer. I actually did this with a harmonized layer, but it wasn't strong enough, so I manually added additional hue saturation and color balance to get that skin tone right here on the chest. So that is looking better. And then I believe this is just an empty layer, but then we moved on to this additional layer. This was generated within Photoshop. You can see it was just a simple prompt saying Jaguar. And you can see I did three sets of generated versions for this out of which I actually ended up using the original one. There were some other good ones amongst there as well, but this was really cool. I really like the expression on the Jaguar's head. And then we also have another generated layer. So this is using generative fill. I extended the logs a bit, and then I believe we have a few more layers here. Actually quite a lot of additional layers. So yeah, I refined the edge of the face with the removed tool, and then I edit some black curly hair on the left with generative fill. Then around the jaguar's head as well, I did some editing with the remove tool. And then let's see what is this one. It's on the sky. Okay? So I just fix something there. Let's see before and after, just a small detail. Then here are the Macas in the background with a lower opacity for them to blend better. And then if I zoom out a bit, we can see I added generative expand, which did a really good job. The only issue was that I wanted the grass to be on the right side, not on the left. So it was like the wrong way around. But besides that, it did a really good job expanding both top and bottom of the image. And I was experimenting with different generative layers, and I ended up building it up. So sometimes if you use generative fill on an area and it doesn't look perfect, but it gets closer to what you need, you just keep adding to it. So you can make another selection and keep adding to it. And that's what I was doing here as well. So you can see with a couple of additional generative fill layers, I ended up having the setup that I needed, having dry ground on the left side, and we have this lush vegetation on the right side. It's looking good. And then I also replaced the logs with another generated image I had before. So this looks much better for this composition. And then I believe I used harmonized layer on that, which didn't really do much difference, but I kept it. And then I used generative field once again to have the river come out a bit further, and even more water, I added this, I think I just use the clone stamp tool, the good old manual way of improving images or retouching images. So once again, from this, we got to this within Photoshop and utilizing obviously a couple of additional generated images. So specifically the portrait, I generated separately, and then the Jaguar was regenerated here within Photoshop. And then once I had this version, I put this image into magnific where I upscaled it to be twice as big. So it added quite a lot of detail for me, but it did a really good job keeping the original features and resemblance. However, once again, the face changed a little bit, lost a little bit of the original appeal that I liked in this image, but it was still quite close to it. I believe I refined this further within Photoshop. But, yeah, we can see the face is slightly different now. Of course, if that person is someone famous, then it would be an issue to have features changing. In this case, we can handle it as a model and if it doesn't have to be anyone particular or anyone famous, these little changes are fine. However, we can see I actually combined again, multiple images here. So once again, starting off with this version, I chose this one from magnific. I started again, replacing the face first, putting back the Macas, and then I used the remove tool to refine a lot of little details here. So we can see before and after, especially around her head. I wanted to make sure that these mountains in the background continue seamlessly from one side to the other. So once you see the updated version, you can see that. So it nicely transitions from the left side to the right side, and it also creates a nice gap between her head and the tree. So the tree is more noticeable. I prefer it that way. And I also wanted to make sure the right side of the bark and trunk looks healthy to create the contrast between the left and the right side. I am not 100% happy with how the tree is split. It's not equally split in the middle, but that's not that big of a deal. However, one thing I wanted to just point out if we zoom closer, you can see that around here, for instance, around the hair, it improves a lot and obviously around the feathers as well. And that's mainly thanks to generative field. So for this, generative field is perfect. So if I were to create a layer here, just an empty layer, and then use the selection brush tool. That's my favorite selection tool that's recently been addit to Photoshop. So I can just make a selection around the edges that need to be better. And without saying anything, I just say generative fill without any prompt, and most of the time, it's going to do a good job. So you will understand what was there originally, but it's going to improve. Of course, you can combine this with remove tool, remove things first that you don't want to see there, and then use this. And yeah, I feel like this one is probably the best out of these, but, I mean, this one is quite realistic as well. So without it, and with it, that's already good. But we could obviously use a couple of additional generations if we wanted to test this out. I'm just going to turn that off and show you once again, the one I created. This looks really nice, in my opinion. And then we can zoom back a bit. Let's see what else we have here. So, this is just the shading for the type because obviously we have the type here to make sure it's legible. I added some darker shading behind the text. And then I feel like there is only one additional generative layer here in this composition on the top. Yeah, I just had to further extend the top because it wasn't tall enough for the format that we need. So that's how I ended up refining, once again, the version that I got from Magnific. So there's quite a lot of little improvements in this. But this is still not the final size because we needed to make this twice as big as this. And for that, once again, I went back into magneific and we can see I did quite a lot of different versions here until we get to this final one. And again, we can zoom in closer. Here, we can see that I didn't use any prompt. I just simply drop this image in. And when you already have a high resolution image to begin with, magnifica will even more likely retain everything. So it will change much less in the image. So that's with the default settings. I already looked quite good. Once again, the face changes ever so slightly, but everything else gets more detail. And I believe I did a couple of additional versions of this just to make sure it works well. But that's essentially what ended up being the final version of this. So if I zoom closer, we can see it's very high resolution, and everything looks really cool, very realistic, and it looks good as a composition, in my opinion. 7. Visual Direction - Painterly: Last but not least, I also created a more painterly version or direction for which I wanted to achieve an oil painting style. And again, I feel like it's quite appealing result that I got. We can see the original version on the left, which again, came from Ched GPT, and then we can see the refined version on the right. So there were actually certain details that I wanted to make less detailed, like the water. We can compare before and after. I felt like it just had a lot of detail there. Also the logs, for instance, I actually reduced the detail on it here, and then there's also some details like the plants. Maybe that one is actually getting a bit more detailed on the right. But if we open up this photoshop file, can show you once again where it started off with. So that was the enhanced version from Magnific just to make sure that we have the high resolution image. And it's actually not even as high resolution as the other ones, but it will still be possible to print this in a fairly large format. So it has enough detail and information. And you can see, I really went for this painterly look, so I wanted the brush strokes to be visible. And in the background, that definitely is noticeable. Some texture of the canvas, I wanted to see. And it's not perfect because there's some details like here. It's a little bit too detailed, which I didn't want. Also these cracks in the ground looks more like a photograph. So it would still need more refinement. But just to walk you through quickly here the layers again, I have combined the faces of the original one from Cher GPT to the one that was created in magnific when I did the enhancing or upscaling. I also changed the vehicle back to the original instead of using MagnificUscale version. And then if we look on the right side, again, I went down to a lower detail level version of this, something more painterly. I remove those leaves there, and then I edit hair with generative fill, then I simplify that right side. I didn't want any distraction there. And then here we have a more upscale version of the same face, some further retouching on the right side. Then the jaguar, I decided to move higher. So I think some additional layers will come in here, which is going to make this work. Yes, so there's the dress. The dress was an important part. That again, was generated separately. So if I just turn this off and back on, that was a big addition and even the hair changed here. So that's that. And then we also have this more painterly or painted version of the face. This is actually with a lower opacity and blend mode set to luminosity. So if I have this image set to 100% and I go back to normal, this is how it actually look. But having the two blended together like this, I feel like I got a good mixture of the two, and then there's just further changes, generative feel again on the missing detail. And then if I zoom out, I believe these are just small little details for the text to make sure the text is going to work, yes. So there you go. There's one final version, and now we can just jump back here and see them all next to each other. And I guess you can clearly see that it is the same composition. Some are more similar to each other than others, like these two on the right are more similar to each other. These two are maybe a little bit more similar to each other. But generally, I would consider these four distinct visual directions. And this is something that can be presented to the client and explain for each what's the strength and what makes them work well for this project. But essentially, we'll come down to their personal preference. And in this case, the charity consists of multiple stakeholders, so they will have to have meetings and decide on which one they go for. And at the time of recording this video, there is still no final decision. So I'm also not sure which one they will go for. But the most important thing here that I wanted to really highlight and focus on is that within the same amount of time that normally would take me to create one version or one direction, I ended up creating four distinct looks, mainly thanks to AI. So if I didn't use AI in my workflow, I probably would have had time for one of these. And even that maybe wouldn't end up being as detailed as these. Because at the first round, when I just test out compositions and also styles for a client, I probably wouldn't end up creating something as detailed as these because I simply just don't have the time for it. But thanks to generative AI techniques and this combination of tools that I am using, I can be much faster. And I love this as a creative because I really get to test and play around with different looks and distinct styles. And this is a huge and clear benefit of introducing EI in your workflow. So without heavily relying on it and without losing control to AI, it can create opportunity to create more varied and distinct looks and more options for the client to choose from. 8. Presenting and the Importance of AI: Now we can just jump back here and see them all next to each other. I guess you can clearly see that it is the same composition. Some are more similar to each other than others, like these two on the right are more similar to each other. These two are maybe a little bit more similar to each other. But generally, I would consider these four distinct visual directions. And this is something that can be presented to the client and explain for each what's the strength and what makes them work well for this project. But essentially, we'll come down to their personal preference. And in this case, the charity consists multiple stakeholders, so they will have to have meetings and decide on which one they go for. And at the time of recording this video, there is still no final decision. So I'm also not sure which one they will go for. But the most important thing here that I wanted to really highlight and focus on is that within the same amount of time that normally would take me to create one version or one direction, I ended up creating four distinct looks, mainly thanks to AI. So if I didn't use AI in my workflow, I probably would have had time for one of these. And even that maybe wouldn't end up being as detailed as these because at the first round, when I just test out compositions and also styles for a client, I probably wouldn't end up creating something as detailed as these because I simply just don't have the time for it. But thanks to generative AI techniques and this combination of tools that I am using, I can be much faster. And I love this as a creative because I really get to test and play around with different looks and distinct styles. And this is a huge and clear benefit of introducing EI in your workflow. So without heavily relying on it and without losing control to AI, it can create opportunity to create more varied and distinct looks and more options for the client to choose from. 9. Final Composition Idea: I wanted to make sure within this course, we have the opportunity to go through direction from start to finish, and that's what we are going to do now. So as a final direction for this project, I thought I'm going to create something a little bit more surreal, but also grounded at the same time. So quite realistic, a bit paintedly but also a bit surreal. And I also thought of the idea of maybe having the tree and the face merge together instead of just being part of the composition. So instead of the tree being so significant and central to the composition, I wanted the face of this woman on the right side to be bigger, so to have a little bit more focus and be more dominant. And I wanted to create this type of transition where the head transitions into the canopy of the tree. So here's an example where we can see this. So this is the concept. 10. Exploration in Adobe Firefly: These are some of the explorations that I've done this time using Adobe Firefly. Now, firefly, I really like because it's very fast. It generates things really quickly. I also like the fact that it has a very easy way to control both the style and the composition. So in this case, for the compositional reference, I used one of the previously created posters, the photo realistic one this time, and I am using the maximum strength for the composition. So I really want these images close to that setup, having the tree in the middle, the person on the right. And I used this prompt that you can see here at the bottom. Amazon Rainforest featuring giant kapok tree. Left side is dead and barren. Right side is lush and healthy, and there's the portrait of the woman and the Jaguar and tree harvester machine on the left side. Now the Jaguar sometimes disappears. It doesn't always show up on these generated images. But generally, it maintains most of these elements quite well. So I am using Adobe Firefly Image Model four, which improved a lot compared to the previous versions. By the time you are watching this, maybe there will be more updated version, which will give you even better results. But more importantly, I increase the visual intensity to the max. I made sure the content type is set to photo, and I also used a style reference, one of the default style references this one with the colorful clouds, which I thought would be quite cool to combine into this composition. But I reduced the strength of that style to the minimum. So I want to use it, but I don't want it to dominate the composition. And that's all that I did. So nothing else is set here. And I just wanted to show you what happens if we change something like, let's say, the visual intensity. If I reduce that and I generate a new set of images, you will see how this is going to affect the results. So we have the same style reference. We have the same composition reference as before. The only thing we changed is that we reduced the visual intensity. And as you can see, the results are really bad. So I'm not sure what's going on here. Yeah, these are completely not usable. So if I go to higher, maybe set it in the middle to a higher value, I feel like it's going to be slightly better, but it's still not going to be the style and the detail that I'm after. But yeah, we can see a big jump in quality and visual appeal. So that's a nice portrait. This is quite cool as well. This actually does a really good job at having the composition set up the way I imagined it. So the silhouette or profile of the head really nicely transitions into the tree and continues onto the left side. Not everything is perfect, of course, but I also really like the jaguar and how her shoulder almost looks like one of the roots of the apoc tree. So that's quite nice. And this is not good. Also not that interesting, not appealing. But then we can maybe increase the visual intensity a little bit more. And I highly recommend to do this. So whenever you use a tool like Mid journey or Firefly or free pig, you should always do a couple of generations at least to test out various settings. And instead of changing all the settings all at once, only change one setting at a time. So now I can check whether a slightly higher intensity is improving the results or not. I feel like it definitely improved certain details like the fire in the foreground is really cool, even though it's on the wrong side of the image, but I really like it. I also like the fire and, like, ashes and the dark colors on the left. Like, the tree is perfect. That's exactly how I wanted the tree to look on the left side, having no leaves, compared to some of these other versions where we actually still saw the leaves, like on this one. Still see some leaves over here. The tree looks actually healthy on the left side. So yeah, I really like this one, but let's just go even higher with visual intensity and generate again. So don't be shy. Burn some credit to get more options to choose from, especially with tools like firefight where the generation time is so quick. It's probably within five to 10 seconds, we get four results, and then we can see them. Well, we can clearly see that by increasing the visual intensity to the maximum, we really get a lot of details on these images, and some are really cool. So are just strange. Like, obviously there will be always strange details like in here, the jaguars look all deformed and just not correct. While this person standing tall as big as the Capocte makes it look like she's actually a giant. It's quite cool. She almost looks like Gaia, the protector of nature, like a goddess, even though she is not actually from that mythology or the South American mythology, as far as I know. But yeah, it reminds me of like a goddess. And again, the tree looks really cool here. So it's definitely worth experimenting and doing several rounds of these. And remember, keep changing a single setting at a time. And incrementally not going all the way left and right, go through several amounts before you can make a decision on which one work best and then move on adjusting additional settings. And one thing that you can do in Firefly, which is extremely useful is to save something as a favorite. Like, for instance, if I like this one, I can just say save to favorites. I can also come up here and I remember I like the left side of this image, which I can save to favorites. And then once I go back to the firefly dashboard, I can go to files and I can choose favorites. So this is where you can see all your saved favorites, and then you can even select multiple images from here and you can do batch processes like downloading them or deleting them if you wanted to. But yeah, so we can see that these worked really nicely. And I'm just going to select these last three that I marked and download those. 11. Comparing and Analysing works: And because I did a lot of rounds of generations on this using the same settings, I ended up having quite a lot to choose from. And the main thing that I would normally do when I have several images that I like is to bring them all into Photoshop and then start to analyze what I like in each of them and try to combine them together. So we are going to create a composite from all of these or the best parts of all of these images. And before we go into Photoshop, I just wanted to show also one other aspect that I love about generative AI, and that is how it really can surprise me or surprise you with ideas. So even though I had this idea of having the tree combined into the silhouette or profile of the head, it surprised me of putting the head even or making the head even bigger and still achieving the same thing. So instead of seeing a side view or profile, we can also do the same composition with the front view of the face, and that way, the face can be even bigger. And this can also be a great way to attract viewers to your composition because having someone looking directly at you is always more effective than someone just looking to the edge of the frame. So the same thing here again, we have a person looking at us of course we could cut off this part of the nose if we wanted to. But that's, again, a very engaging setup. So we can see another one here. Let me just select a few of these. I can even put them next to each other here in Bridge. But yeah, this way, we can see it. I love this face. It's so intense, very dramatic. There's even a bit of blood there, which I most likely would remove. Maybe that's supposed to be paint, but I really, really like this look, a very serious, dramatic look and I also like the fact that there's not only contrast between the vegetation or the lush details and the like dry cut down trees, but also there could be a contrast between night and day. And it's, again, something I didn't think about myself, but having a completely different lighting on one side and the other could further emphasize the contrast between the two sides. So this is something that I notice from this image in particular, having a completely different lighting in it. Now, having the head directly connect into the tree. For this, I have lots of different options. I'm just going to maybe take one of these so we can see them side by side. And yeah, so this is lots of different ways of doing this. We can see really cool examples for that. We also have a few examples where the head is below the tree. And here the tree almost looks like continuation of the head having something like a decoration or like a hat or piece on top of the head, like a crown. That could also work. Here's another really cool combination of the left and the right side. And then obviously we saw this example when we have the full body be also interesting. Another example where the head continues into this very intricate and detailed decoration on her hair, and then that continues into a tree. I really like this as well. I love the composition. And again, we have the day and night time contrast or separation, which I love. Really cool atmosphere in this image. It's actually one of my favorite ones. And then, yeah, so we've seen lots of different variations. Even this one is quite interesting. So here, the tree almost disappears on the right side. We only see the trunks, but then her head replaces the tree in a way. So, again, we see a lot of her and really focuses on her. So we even see the shoulders. This is also cool. So again, something to think about. But I'm going to select all of these now, and I'm going to open them all in Photoshop with a batch process that I like to use. So if we go to Tools, Photoshop, load files into Photoshop layers. So that's what I would normally use. If I want to keep all of these images in the same Photoshop document, 12. Selecting the Main Image: You can see here on the right side, it's like an action that's going to place all of those layers in. It names the layers according to the original file names, so these are the original fine names. You can see them. And yet, the great thing about this is now we can see them on top of each other, so we can really compare them and start deciding on which elements to use from which images. Here's another interesting example where we have the person looking more like a giant because there's these smaller palm trees and she's surrounded by them. Having these smaller trees around her and near this big tree also helps to perceive this as really a giant tree. So this helps to set the scale up better. That's also something to keep in mind. So if we decide to use one of these images or composition, having some trees near the big tree like here, really make it feel like that one is really a giant. So it helps to see the two scales of trees relative to each other and being close to each other. This is also quite nice combination one is a little bit strange, and there is even a stone jaguar in the background. But what I liked in this one, and that's the reason why I favorited it is because of this tree here in the background. I found this to be really unique and really cool. So that looks like another giant couple tree that was cut down or it's just died out because of the deforestation that's happening. So it looks really big, even though it's in the background. That's what I like about it. And then yeah, so I feel like we've seen most of these other ones. So now it's time to think about which one is going to be the main image that we will use. I mean, there's lots of good ones to choose from. But I'm really feeling drawn to this particular one, especially because of the right side, but even the left side is probably the closest to what I imagined for this composition. I like, again, the fact that there's two trees. So there's another tree in the background. I really helps to get the scale right. We can also see some smaller trees around the bottom, which makes this one giant or makes it feel like a giant. The jaguar is also cool how it's lit. So there's, like, darker details around it. But then there is a section here where there's more sunlight hits the jaguar, so it really helps to focus on it. And instead of having it up here, it's quite a cool placement for it to be at the bottom. But of course, we will have to see the head a bit more. Maybe we want to make it look like it's looking up as if it's seeing something or hearing something maybe from the left side. Once again, I like the night and day contrast here. And a couple of things that I don't like is obviously this shape in the background that needs to go. 13. Refining and Borrowing Details: What I will do is, first of all, I will mark this layer, right click near the icon, and maybe add the magenta mark on it. So that's something that we will be using. But I will keep all of these other layers available as well. And then I will start working on this. So I will first create a selection around this part here. And if I just choose generative fill without any prompt, most of the times, it's going to try to remove whatever is contained within the selection. In this case, it might remove the branch as well. But because the branch continues a bit to the left, it might understand that I wanted to keep it. So there you go. It actually did a really good job, removed what I didn't need. And cap the branch. So that's great. It doesn't have to look exactly the same way because that was obviously generated already to begin with, but this works much better. And then what I want to have is maybe a little bit less trees in the background. So I'm going to use the selection brush tool and just paint over the trees here. And I'll try to remove these. The good thing is you only have to draw around the edges, and this tool will automatically fill the inside. Again, I'm not going to use any prompt, just simply try to remove that. So best way to use the generative fill for removing details is not to include any prompt. In my opinion, that works the best. And yes, so we have three options to choose from. I feel like maybe this second one will work the best for us. So that was before, and this is after. Yeah, I think that's good. The machine here works, although it doesn't have a crane on it or any tools that it would be using, but we can ignore that fact for now. I like the fact that there is a road here that's a good leading line leading into the image, so directs our attention to the center of the image. And what I want to see just quickly is maybe some fire in the foreground. So let's see if we can borrow some fire from one of these images. And I remember seeing something that I really liked. I mean, the smoke here looks really cool, too, but let's keep moving. Okay. No this one, not this one. Not that one. Not that one. There's a tree. Maybe we will borrow this detail. So I'm just going to make a selection of that. I will use the Object Selection tool, and let's see if we can make a selection of that tree. Yes, it did a really good job selecting it. So I'm just going to press Command J or Control J and drag it up here. So we will decide what we will do with that later. And let's continue going down. Just turn off that layer for now. Okay, so this fire here looks really good. I like it. Oh, this looks great, too. I'm looking both at the left and the right side. Okay, this is my favorite. Yeah. The only issue is we will have to somehow get the colors right between this and the other one. But what we can do is just to try to make a selection on the entire left side of this image. Something like that. And press Command J, move it to the top. And then we will come back to this, but let's just go down a bit further. I actually really like the fire on the right side here. So I'm just going to make a selection of this section here as well from the same layer. Let's just move that to the top, turn it off. Okay. That mountain looks quite cool as well. I quite like that. So again, I'm just going to make a selection around this and make sure I have the right layer selected. Command J, move it to the top. So this is what I like to do to borrow details from images. And I'm just looking at the roots. Now, I actually prefer the roots on this version, so I'm going to keep that. But maybe let me just see this again. So no, I like the lighting as well better there. This machine looks really cool as well with the smoke and the dead trees or barren ground, especially the background looks good here. So again, maybe I will make a selection of this just so we can test it out how this side works with our composition. So let's do duplication of that. And then let's go back down. Again, we have a nice fire detail here in the foreground. And this is okay and not bad. Okay. So let's keep these layers turned back on, but we want to work from this image. And then let's take a look at the variations that we have here. I'm going to put these below there. And we have our generative layers ddited here already. And I'm just going to put these bottom ones into a group so we don't get confused between all these layers. 14. Combining and Generating Images: So let's see how this works. Combining the left side from another image with the image we have on the right. Yeah, I mean, there is more contrast probably between this more daylight view compared to this dusk that we have here, but it still has a good combination of cooler tones on the left and the warmer tones on the right. So we can argue that it still works quite well. And these two works well together as well. Again, I love the fact that because we are using the same original structure or composition, the trees quite nicely match or line up, like the roots match together. And here as well, it looks like it was originally put together like that. And yes, so this particular image has that also like a giant kapok tree that's half dead in the background. There's the smoke which I love, and the fire as well burning. Yeah, I feel like this really captures the desolation of the area really well. And we could potentially include a mountain in the background if we wanted to, but I don't think it's necessary. It's just good to have that detail there if we wanted to use it. And we can try to use this image. I'm just going to flip it around and put it here just to see how this would work. If we decided to use it, maybe set it to screen mode, we could have even more fire on the tree if we wanted to, something like that, we could even use it with colored Dodge or let's see what else we could use. Probably screen is still the best or Lighten, Lighten is even better. So if we wanted to have more fire, we could add this here. So the actual tree itself could be on fire. If I add a fully filled mask, we can do that by holding down alter option key and click on the mask icon. We can then use the brush tool and have a soft edge brush with white color. We can paint over the fire and essentially paint the fire in wherever we want it. Let's move this up a bit maybe around here or maybe this section up here. And then see if we want that or not. Maybe done there. Yeah, I'm not sure if we need that. I feel like it's already looking quite good. And by the way, we can just put these additional layers in a group. I'm just going to turn them on but hide this group. By the way, we could obviously use a selection on an area that we want to be on fire. Let's just say around this much. And then say fire. And let's see what happens. This is regenerative field. If we already have some fire in the image, it usually does a good job trying to match that lighting. And yes, maybe this one looks promising. Yeah. That looks quite cool. It's like burning in between those two big chunks of roots, which is quite nice. Yeah, it's like a hollow space inside that's burning. I quite like that, but I still like that big chunk of root there. So when you see something that has potential, like, in this case, this one, you could say, I want to see similar ones, so you can right click on there in the property panel and say generate similar. Let's see what we get. So at the moment in Photoshop Beta, we have access to Firefly Image Model three for these generations. Even though Image Model four is already available in the browser, it doesn't immediately come into Photoshop. So there will be probably a bit more weight until it will appear here. So let's see before and after. Yeah, I still like this one the most. But I'm just going to do one more round of generations. So three additional images just to see what we get, and then we can make a decision whether we want to use this layer or not. Okay. Yeah, not bad, not bad, a bit more fire, but I still prefer the structure of the root like this. If I wanted to go with one, it would be this one, for sure. This one was the most promising one. But I'm going to turn this off for now. I feel like it's still better the way it was. Now, there are issues with the vehicle, even though I like it. It just doesn't really make sense the way it is right now. So this is again, something that you can test and see if you can improve or not. So what we can do here is make a selection, gives space for photoshop to work, and then we can specify what this is. So it's a tree harvest machine carrying logs in its claw. Although that bulldozer that we have there is not bad, but I feel like that would be even better to have something that actually extracting the trees. Let's see if we can get something. Now, it didn't do a good job of matching the style. It created something more illustrative. Maybe we can try to say photo of 3 hours from a machine surrounded by fire and smoke. Maybe that will help to ground it a little bit. And the fact that we are saying it's a photo and it's going to hopefully make it more realistic and less illustrated, like in this example, let's see what we get. Okay. That's better. That's better, too. Yeah, I mean, the colors don't look good. But in terms of actual functionality, these look more what I had in mind originally. So I'm just going to do one more round of this, we might get lucky. It's definitely worth doing a few rounds once you see potential in generative feel layers or generative layers in Photoshop. Again, similar to in general, working with AI, you shouldn't just settle for the first round. You can see this one actually looks quite good. That's actually a functional machine, and I like the claw. I like the big logs there that it's extracting. While the other one is more like just moving ground around. This one actually is cutting down the trees and taking them. So I feel like this is probably better. But what I will do is try to change the color of it a bit. So I'm not happy with this color here. I'm just going to make a selection. I'll try to make an object selection around. Like that and then add a hue saturation adjustment that's command you or control. And then I'm just going to probably add colorize to it. No, that's too much less colorize it. Actually, it would be better to use color balance. Command or Control B, and we can probably make it work better that way. Something like these colors work better in my opinion than the original one, but we will use also levels on this, and I'm going to change the intensity, make it a bit darker, and also drag this down without making it too harsh contrast. Yeah, something like that. So before and after, it's a very subtle change. From a distance, it's not even going to make that much difference, but I feel like it helped to make it more realistic in that environment. Yeah. One other thing you can also try is to duplicate these layers. I'm going to turn off what I created on it, and maybe just make a selection of just the vehicle itself. Not even the logs around it, something like that. And then press Command J, and then this one selected, I can just choose harmonize from the contextual task bar. You can't use harmonize at the moment on generative layers. That's why I had to do the duplicate. And let's see if it can harmonize it into the environment and what it would choose to do with it. I think it's not going to be that much difference because it's already, it's pretty much the same thing. Yeah, it hasn't changed anything on it. Okay, so in this case, harmonize, actually, we have three options to choose from, but I don't think any of them makes any difference. So yeah, we can just remove that and then maybe we can turn back the changes that we did. Yeah, I prefer that one a bit more. I would refine this further, but for now, I'm happy with the way this looks. 15. Refinements and Sky Replacement: Now I want to focus a bit on the sky because although I like it, I want to make it look more bleak. So first, I'm just going to use the remove tool and paint over a few details. Like this tree here can be removed. And then there's also a little bit of detail here that doesn't make sense. So I will remove that too. When you use the remove tool, it's good to have the sample or layers turned on. And also, I like to use the remove after each stroke. That way you get a result after drawing over anything. So I can just draw this here. Maybe a bit more there, maybe a bit more here. Yeah, let's just cut this down a bit more. I'd like to separate that. If the remove tool doesn't work, you can always switch over to the clone stamp, and then maybe with a hard edge, we can just cut into this that I want to paint over these details a bit, create a cool looking crooked tree here, something like that. Yeah, it looks quite dramatic and then use the remove tool over those details that be painted over with the clone stamp. That should blend things together better, maybe do a little bit here as well. Yeah, that looks much better now. Okay. Maybe we can also remove a bit more here. And maybe a bit more here, this looks quite air generated these branches that looks better, much better. Maybe this one as well, can be removed. Okay, that looks strange, too sharp. Yeah. That looks good. Again, a generated. That's why I'm going to just paint over it. Yeah, it's much, much better. Now, if you wanted to add branches, you feel like you need some branches here, maybe on this side, we could just paint over that section and then just use generative fill. I'm using the selection brush by the way there, and then I'll say dead branches. Let's see if that does any improvement on it. And we get, again, three options to choose from. Yeah, I think this one is quite nice. Instead of that, this looks much better. Even that I quite like. Yeah. It's a good way to lead our eyes into the right side. So it's almost like a hand pointing to the right side. I really like that. Now, what I like to do after I use the remove tool is to select all these layers that we created and pressing Commando Control E, I would merge them together. So it's a single layer. So even though it's good to have them on separate layers, just so you have access to them, at the end, it's good to merge them together, and that is looking better. Now, I am going to make a selection of the sky. So I will go to the select menu, and I'll choose Sky. And yes, actually, this feature would only make a selection of the currently selected layer. So what we will do is, without having anything selected, I will just press Command Option Shift E or Control Alt Shift E on PC. And that's going to create a new composite layer. And on this layer, I will actually delete the right side. So I don't want to make any changes on the right side. That can be deleted. So if I have this layer on only, that's how it's going to look like. And having that layer selected, we could go to select and choose sky or even better, we could go to Edit menu and choose sky replacement. So this should give us a fairly good selection of the sky. And, yeah, I can see that it did a good job maybe apart from a little bit of detail here that didn't work perfectly, but we can probably fix that later. More importantly, we can now decide what type of sky we want. So instead of spectacular sky, I wanted something that feels like there's burning fire and it's not a happy sky. So let's see if it's a bit darker like that. That looks quite cool. Mm hmm. Yeah, that's quite menacing and works with the image. I'm just looking at whether it has enough contrast or not on the tree, but I feel like it does. And of course, we can adjust its temperature. So if we wanted to make it even cooler, we could. I'm just going to keep it in the middle. The brightness of it can be affected as well. We can make it slightly brighter. I think that helps. Make it more realistic, and then we can also shift the edge of the selection. I feel like maybe it can be shifted this way a little bit, just around the branches. I feel like that helps to make it better, maybe even more. Yeah, blends details better together. So this was before and this is after. Even though that sky is beautiful, I think this new version looks better. However, I like the bright details here near the horizon. And what we can do is to drag this sky a little bit higher and maybe even move it this way. So we still have that bright detail there. Yeah, I think that's really helping to draw our attention on that tree, but it's not so busy as before. Before that, it was just a lot of clouds and details, and now it really feels like a sad sky. So the cool thing when you use this feature is that it adds a lot of adjustment layers in a separate group. So this is the sky replacement group. And we can actually even delete this layer underneath it. And now that group is going to be there for us where we can go back and make changes to everything. So the selection is not perfect, but I'm not sure if we could do any better than this. Perhaps using Selec subject, we could improve it or we could just use tools like the burn tool and Dodge to paint over this. But I'm not going to make that much changes here. It's just going to using the Dodge tool. Maybe just brighten up those details there a bit within the mask. I'm using the mask view. That's Alter option click on the mask icon. You can access the mask view, and I can just paint over this using the Dutch tool set to highlights. So the range is highlights, and that's how it's going to behave. So that did a fairly good job. I feel like it improved it a bit. There's less of a halo around it. Yeah, I think that looks good now. 16. Generative AI refinements on the Jaguar: Now, the main thing that we need to refine is the jaguar in the foreground. So once again for this, we can try to use the generative fiel. I'm going to make a selection just around here. I try to control the size of the head as well, and I'm just going to say Jaguar. I don't want it to replace the body because I feel like the body is great. And also the lighting is really good here. I am just trying to get it fill in this area, make the head pop up a bit more. Yeah, I mean, that Jaguar is very relaxed. This is quite cool, and I just make sure that this layer ends up in the right place. Okay. Yeah, I quite like this one already. Just feels a little bit too relaxed to me. So maybe do another one. We could obviously be more specific with our prompt and say that we want the jaguar to be looking up or ready to pounce, maybe. We could use that. But in that case, we might need to change the a full body. Yeah. So I'm just going to make a selection of the whole body. Out of these, I actually like this one, so I'm going to keep that. I'm going to make another selection and say jaguar lit by the sun ready to pounce. I wanted to make sure that it has the sun on it. I really like that way of highlighting it, again, creating contrast in the composition. Uh Okay. I mean, that's a nice silhouette, but it looks very stock photo, like a stock photo, very static. This one, the lighting is better. Again, not that happy with the style. I mean, the pose. The pose is the problem here. So I'm going to do another generation. But for this, we might need to test out a couple of other tools. So we might need to leave Photoshop. If we don't get good results, maybe after the second of third generation, I will end up trying another tool. I mean, it's not completely bad, like, not horrible. The results, that's why I still try to give it some more chance to prove itself. And this one looks quite cool. Yeah, this one I like maybe a bit too bright, but it is close to what I had in mind. Yeah, it's a little bit too bright, but it sort of makes sense, although the sun should give it more of a back lit setup. Yeah, it feels more like it's lit from the front and it's like a daylight sun. However, we need more of a sun that's setting in the background. Maybe if we rasterize this layer. Let me just do rasterize layer and then use harmonize and see if it does any better. However, I think harmonize won't work on layers that you just generated directly within Photoshop because it's just simply will do the same thing. So I don't think it's going to improve the colors, but I'd like to give it a try. Yeah. I mean, it kept it exactly the same, very subtle changes. I'm going to keep it as a generated layer, but I will probably turn it off because I still prefer the lighting here. This is the lighting I want, but I want it to look like this. So we could potentially do this within Photoshop without leaving Photoshop by using this feature called the reference object. So for this, we could use this image here or even the one behind it and then start a generation. But instead, I'm going to show you what we could do if we take this into another tool. So I'm going to do a screen capture of this. I mean, you can export this detail as well if you wanted to. But for now, I'm just going to do a screen capture. What I'm going to first use is freepik where under the edit option, you can browse a file, and in this case, this is the image we want to work with. And then we can just paint over what we want to change. Yeah, something like that. And then I want to say the same prom that we used in Photoshop. So I'm just going to borrow that and drop it in here. We want this to be replacement, and we want it to be retouching. And then here on the right in the mode, we could change this to pro or classic, but I'm just going to keep it on auto and then let's test it out. So we should get the four results fairly quickly. And I believe the fact that we didn't include photo is the main problem here. So it created illustrations because it looks quite illustrative the background. The pose is really cool, by the way. I really like the poses that it came up with. So I'm just going to this out, and I'm going to say photo of Jaguar. Let's try it again. Okay. I added a sun as well, which was unnecessary. And I'm not sure why, but it's just not showing these two additional generated results, only these two. Maybe if I just say X, maybe we can change the mode to pro. And let's see then. Uses more credits, but I'm just curious to see if it's any better or not. I may have already used a pro version before. It's taking longer than usual. So I'm not sure if it's working or not, but while we're waiting for this, I'm going to add the same photo in Chachi PT, and I'm going to use the same thing and say, change the jaguar in this image to be lit by the sun and have a more interesting pose. As if it's ready to pounce on something on the left side. Okay. And I want this to be an image, so I say create image. I like the conversational way of changing things. So Chat GPT is really good for this. Maybe I forgot to mention to keep everything else image the same. By the way, here we don't see the results. So yeah, I think it just couldn't finish it. But free PEC also introduced the AI assistant, where you could upload images and do the same thing. So you don't have to have hat GPT in case you have free Peck access. But let's see what Chat GPT can do for us. And after waiting for roughly around a minute, we get this result in Chet GPT, and I really like this. It's much bigger than what I thought to have here. I mean, in terms of the scale of the jaguar, but I think it works, and I like how it interacts with the plants as well. So we can just drop this back in here. And if we just resize it down to the size that we needed, I feel like it needs to go a little bit higher somewhere around there, maybe even higher to match up with the roots. Yeah, something like that. That looks quite cool. So that was before and this is after. And now we can test out the harmonize feature, because it's a completely different image, let's see how closely we'll be able to harmonize it to the background. And it did a quite good job generally getting the colors working. I think this maybe second or the third one works best, but I still feel like the colors are not perfect. So if we look at how this detail looked before, I feel like that was still closer to what we needed. And yeah, I like the roots here a lot, so I wouldn't want to change them too much. So maybe what we could do is to add a mask onto this. So I'm going to group this and a new mask on it, and I'm just going to use a soft edge brush and mask out these details like that, and maybe even mask out here on the left side. I try to show as much of the original roots as possible. Yeah. Something like that. I think that actually starting to work quite well. So let's see before and after. Yeah, it's certainly more interesting. And we can also start blending these a little bit more better into each other at the bottom and on the right. Yeah, it's looking quite good. Let's see before and after. I like it. Maybe we can have a smaller brush with less softness and then just go over these details, and I'm not sure how I could blend this well here, but something like that looks good enough? I'm trying to find a good mixture between the two results. That makes sense. Yeah, I think that looks quite good. However, I would still change the colors on this, so I will use color balance, and I will make it more purple for sure, and maybe even a bit more cooler, more bluish, something like that. So that was before, and this is after. I feel like that looks a little bit better. Then we can use levels on it as well. Add a bit more brightness on it because now it got a little bit too dark. And let's see now before and after. Yeah, I feel like that looks quite cool. So it really has that pause that I was after. And without the color adjustments, actually look like this originally. And then with all the color adjustments, it looks like that, and that's how it is masked out. I feel like this is going to work for now. Might have a little bit too much focus on the jaguar now. Maybe we can reduce the levels, the intensity of this, make it a bit even darker or just move this to the right there. So keep it really dark. And then the color balance, we can go back to maybe make it even more cooler. Or more magenta, more magenta in it. Yeah, something like that. Once again, without the adjustments and with the adjustments. Yeah, I think that's quite nice. That works better now. The lighting is still not like 100%, but we could imagine this is backlit. The light is coming from here. And maybe if you use the whole imaging chat GPT, we would get a little bit better result than this. But I am happy with the way this looks. 17. Aspect Ratio, Upscale and Text Layer: And now I think it is time to check the aspec ratio because we need a two to three aspec ratio. So I'm using the crop tool and set up the aspec ratio that we need. And most likely we would want to have more space both on the top and the bottom. And I want to use the generative expend feature from the contextual taskbar. I'm not going to add anything. Let it understand what is needed here. So let's see what is going to be generated for us. We need a little bit more taller format image than what we had originally here. And let's put this up here on top. And then we can put these other unused layers inside this group. So this is going to be the Jaguar. I'm just going to name it. We have a sky replacement. We have the tree harvested there, some fire. Yeah, that's already named fire. I'm not sure what was this layer for. Oh, yes, that was used on the left side. Okay, so generative Expand layer did a good job. I probably will go with first one, I like the water, the way the water is extended. So there's even more water there. Yes. First one will be the winning one. The sky looks good on the first one as well. There is a little bit of strange line here, but we can just remove that easily by making a selection, going all the way to the top, and then coming down maybe on this side, it's going to be easier on a separate empty layer, and then just use the healing tool or even clone Stem tool and just sample from the sky, make it softer a bit, and then go up to remove it. Yeah. It looks better, and maybe with the healing rush, we can heal this part here. Let's see if it does any improvements on it. No healing brush, it's the remove tool. Yeah, that looks great. So now we have something that we need to upscale because the size of this image is not enough for print. It's only around 1,800 by 2,700 pixels that would need to be more closer to seven to 10,000 pixels. So what I'm going to do is to export it. With its current size and save it. And then we jump back to magnefic and we go to magnificUscaler, and we will place in the input image, this image that we just saved. I'm not going to put any prompt in here because it's already high resolution enough for it to create a good result. And I'm going to go straight for four times as big as the original. Now, this is going to burn a lot of credits, and it's going to take long, but I feel quite confident in magnific and I think it's going to do a good job. So I'm not going to even change any settings here, just simply run an upscale on this. Now for the result, I had to wait a couple of minutes. Probably it was five, 6 minutes, so it took a long time. But let's take a closer look at what we got. If I zoom closer, we can pan to the right side, and let's take a look at the face. That's probably the most important detail here on the right side. Yeah, I mean, it's changing the features. Especially the eyes. They look more European or Caucasian, but it could still pass as South American maybe. It's not too bad. Let's go down here. Jewelry looks good. Tree looks good in the background. Let's look at the jaguar. That's looking great. I like the eyes, although it looked a little bit more interesting with yellow eyes. It looked a bit more menacing, but the green eyes are good, too. It looks to have a bit more cat eyes, which is good. However, in dark, it would be more round the pupils. So I guess that's more realistic, but for now, I think it's fine. And then let's look at the fire detail and reflections. They look good, too. The harvester looks perfect, looks very realistic now. So, yeah, it looks like a real machine. These don't really look like trees, though. That's my only concern there. This tree looks great. All those little details are eddied there to make it work, and the trunk of the tree looks really good, as well here. And these branches look much better. There's less halo around them, as well. Although there is a bit of, like, separation in the sky here in the middle, which we will have to fix. But besides that, I think it did a really good job. Yeah. So while they're magnific, I really like the results. So I'm just going to download the PNG and we will open this in Photoshop. So let's take a look. So it's 91 megabytes. The size of this image. Yeah, it's a big one, and the size is, yeah, it's perfect. If I set this up for inches, and I set the resolution to 300. Yeah, we get a really decent print size for this. That's great. Now we just have to bring in the text to see how that works. So I'm going to go back to my designs, and I believe for this one, we will have to use white text. So maybe I'm just going to use this composition and bring in all the elements that I had here. So let's drop these in here. Yeah, as looking good. And I just have to change the title color overlay to white. I think white is going to work well here. I mean, we could maybe pick yellow, just to soften it a bit, or maybe bright purple. I can only think of this maybe pink. It's actually quite good. Maybe even darker version of pink. Yeah, something along the lines of that. I think yellow is good, too, maybe even darker version of yellow or something like this is nice, too. But same way, white could work. I think it's very clean. And, yeah, I quite like the way that works. It doesn't distract too much from the design, and the text is actually legible here at the bottom really well because the background is already dark. It quite nicely worked out how this island is forming a backdrop for the text, and the logo is very legible as well. So there you go. We have one new version, and I'm just going to save this back into where the other files are. We can just take a look at these two side by side. So the version before using magnific and the version we have after using magnific. And finally, we could also just compare this to the other similar compositions. So the more realistic ones, and we can tell that it looks different. It probably looks like a combination of the colors from the other compositions that are more illustrative, but still looks photographic. And yeah, it feels a bit surreal. I would call it. So it's very dramatic, very cinematic. Yeah, maybe best way to call this visual direction. This would be the cinematic or surreal direction. And I hope you found this useful that we went through every step that was needed to create this. And of course, there's still a lot that I could refine both on this version or the other directions. But this is, again, just another cool image that we can share with the client and see what they think. It might be actually something that resonates with them much more than the other images. It's different enough. But still keeps the original idea and composition that came from that initial sketch that you can see here on the left. 18. Conclusion: Thank you so much for joining me in this creative journey. I hope this course has inspired you to explore new tools, expand your visual language, and rethink how you bring ideas to life. Don't forget to submit your work for the class project. I would love to see your take on it. Keep experimenting, stay curious, and have fun designing.