AI-Powered Expressions in After Effects

1. Intro: Hi, I'm Valerie, and welcome to AI for Motion. If you ever feel intimidated by expressions and you want to leverage AI to help you in your animation workflow, then this one is for you. In five lessons, I will show you how to use AI tools like HAGPT to generate expressions that automate your animations, build smart setups and save you hours of manual work. We'll cover everything from creating carouselels, hover effects, bouncing I panels to automating cursor clicks without writing a single line of call yourself. The great news is that you don't need to be an expert in after effects. You know the basics, you're ready. By the end of this course, you will have a reusable setup you can apply to your own projects. And most importantly, you will have a whole new way of thinking about animation with AI. So if you are ready to animate smarter and faster, let's jump into the first lesson. I'll see you there. 2. Building a Smart Carousel Animation: There, and welcome to the first lesson of the course. I hope you've already downloaded the course folder. I called it expressions for SAS because the animations we're going to create in this course are mostly used in SAS explainer videos. All right. So first, let's begin by checking what's inside this folder. It's nothing too complicated. Inside the hat GPT folder, you'll find PDF files of the conversations I had with Chat GPT during this course. That means inside these PDFs, you'll see my exact conversations with the chat but saved as PDF files. I added those because for each one of us, the chat might give slightly different answers. And in case the AI gives you something completely off topic in your chat, you can always go back to these PDF files and copy the exact expressions I used. But don't worry. I'm sure that no matter which AI you're using, we'll all get very similar results because at the end of the day, we're generating very simple lines of code. All right, let's move on to the next one. There's an Assets folder that contains all the images we'll use in this course, like the animal pictures and the software icon logos. You'll also see an AE folder with the After Effects project files. One file includes the final animation we'll be working toward in this course, and the other is the version without animation, which we'll start with right now. So let's open that one and begin the lesson. Before we move on, let's make sure we're all using the same workspace layout so everything stays in sync. To do this, go to Window, then workspace and choose default. If your layout still looks different from mine, it's probably because some panels were moved around. To fix this, we can reset the workspace together. Now we should all be seeing the same thing. Now, at this point, don't do anything, just watch and listen. We'll get hands on very soon. Let me first walk you through the project panel so you can understand what's in there. Inside the Assets folder, we have the animal images and software icons. There's also a solids folder with the solids I used as backgrounds for each lesson. Each lesson folder contains the main comp for that lesson and a sub folder with the pre comps related to it. I already prepared the scenes we're going to practice with in this course. In each lesson, we'll practice something different. Don't worry. I'll also explain how I set up each scene before we start animating. For now, we can close all the precomps and focus on the first lesson in which we'll create a super cool carousel animation with a lot of micro animations like opacity, scaling, and even drop shadow animations, but without actually creating keyframes for these micro animations. We'll do it all using expressions. But don't worry. You don't need to know expressions at all to make it work. We're going to use AI to generate the expressions we need. Before jumping into the animation, let me quickly explain how I set up this scene because there's a lot you can learn from it. As you can see, I've got a few pre comps here, each one with an animal picture. You might ask, why did you precompose them? Why not just use the picture layer as it is? So let me enter this precomp and explain why I did that. First of all, it's because all the pictures are different sizes and I need them to be the same size. Moreover, I wanted to create the rounded edges look. I created a rounded square using a shape layer, then I used the track Matty function to use this shape as an alpha for the picture. This shape is now used as a mask for the picture. It's better than applying a regular mask directly on the picture layer because it lets me freely position the image however I want. Most importantly, it allows me to easily swap pictures later if a client asks for changes. I simply select the image I want to replace, then hold Alt or Option on Mac and drag the new one onto it. After that, I can adjust the new picture to fit the layout. As you can see, the rounded corners stay the same, and I'm ready to keep working. Alright, I'll press Control Z to undo my changes. Oh, and by the way, when I work on projects that include lots of pictures, I use medium sized pre comp dimensions. A size like 1080 by 1080 usually works well. With that said, I hope you see the benefit of setting it up this way and now we can start working together. Get ready to make some magic. So now let's begin creating the Advanced carousel animation. As I mentioned earlier, we're going to use AI to help us with this. In our case, we'll use the free version of hat GPT. If you don't have an account yet, go ahead and create one. But honestly, by now, most people already have one. I'm using the paid version, but here you'll see me using the free version called GPT four oh. It works the same way with older versions too. Though by the time you're watching this course, that version might not exist anymore. Anyway, let's start building. First, I want to share three important tips that will make this process much more efficient for you. One, when asking for help with After Effects, always start by showing GPT a screenshot of your pre comp. That way, it can see the whole setup, which makes your request easier to understand. Two, when writing your first message or task, always end it with a question like, do you understand what I mean? This helps GPT think through your idea more carefully and often explains it back to you even better than you described it. Three, if things don't go as planned and GPT isn't giving you what you want, don't keep trying in the same chat, close it and start a new one. That resets its memory. If you keep asking in the same thread after several failed attempts, it'll stay locked into your original prompt, which might not even be accurate. So it's better to start fresh. All right. With that in mind, let's start creating the karoel animation. According to our first tip, we'll begin by taking a screenshot of our scene to share with GPT. In my case, I'll search for the Snipping tool. Here are the shortcuts for both PC and MAC. PC users shift Windows S. Mac users Command Shift four. You don't need to capture the entire interface, just the important panels. In our case, what matters most is the preview panel, so GPT can see the layout and the Layers panel, so it can read the layer names. We want GPT can see our scene. So make sure the screenshot includes the preview panel and the layers panel, so it can also see the layer names. Now let's copy the screenshot and paste it into Chat GPT. Next, we'll explain what we want it to do. I'm going to write it using simple English with some grammar mistakes just to show you that perfect English isn't necessary to get good results. That's where our second rule helps. So let's try it. I recommend always starting by mentioning what software you're using. Then briefly describe what's in your scene. Our goal is to give GPT a basic task without getting too deep into the details at the start. For example, a specific value like 20% isn't very important at this point. All the values can be fine tuned later, so don't worry about being exact. Just focus on describing the animation idea the way you imagine it. Like I'm doing here, I'm telling you that I want the opacity of the non centered precomps to be lower to make the centered one stand out more. And now, once you've described everything, don't forget our second rule. If you're not sure you explained your idea clearly, finish your message by asking, Do you understand what I mean? Now, you should know that the answer GPT gives will be a little different for each of us. And that's totally fine. The most important thing is to check that it understands your vision in the first part of its answer. If something isn't right, you can correct it before moving forward. For example, I noticed that GPT suggested scaling the centered precomp by 20%, which in its mind, means going from a scale of 100 to 120. But in my case, my default precomp scale is not 100, so I don't want to scale it to 20. I'm going to tell GPT that in a simple way, my default scale is 40. Then I go back to my project, check what looks good, and I see that 60 works well. So I go back to the chat and tell it 60. Before continuing, I'll also look at GPT's other suggestions. In this case, I see it's offering two different motion styles. I want the animation to be smooth, so I'll mention that in the chat as well. With that done, let's run the task. Now, we have a few answers generated with some simple explanations, you don't have to read them all, but if you want to start learning how expressions work, it's a good idea to read them. The nice thing is, even if you don't understand expressions, GPT usually gives clear explanations inside the expression code without breaking anything. It also tells us which property we should apply the expression to, and that's the important part for us. Just remember, before you move forward, if you have any questions, keep reading the rest of GPT's answer. Most of the time, you'll find the explanation you need right there. Alright. So now let's go to the first scale expression and copy the code. Now, open your project and press S to reveal the scale property. To open the expression input, hold Alt or Option on Mac, and click the Scale Stopwatch. Now, press Control or Command V to paste the expression. To complete the action, click anywhere outside the expression panel. And there we have it. The first precomp is now scaled to 60. Let's open the action grid and place the precomp in the center to see if it's working. Looks good. Now let's do the same for the opacity property. Let's check it. Awesome. It looks really nice. If you run into any errors or something doesn't work, don't worry. You can simply copy the expressions I used. They're listed in the chat GPT folder. In the next lessons, I'll show you how to fix expression errors. But for now, just follow along and get comfortable with the process. Alright, now let's select all the precomps and move them, so the one we're currently looking at is placed in the center. I Great. Now, to apply these two expressions to the rest of the layers, first open the opacity property, right click on it, and choose Copy expression only. Then select the rest of the precomps and press Control or Command V to paste it. Let's do the same for the scale property. Awesome. And now we need to control the position of all the layers at once. To do this, let's create a new null layer and parent all the precomps to it. Make sure not to confuse the track Mat Piwip with the parent and link Pick Whip. Those are two different things. Awesome. Now, we can move the null layer left or right, and we'll get that smooth carousel effect. Just imagine how many key frames it would take to create this manually by animating and timing every single precomp. And that's the real power of knowing how to use AI to help with real tasks in our projects, even if you don't know expressions at all. With that done, we've only scratched the surface of what's possible with AI and expressions. Now let's take it to the next level. Let's say I want to change the scaling and opacity values in the expressions. To do that, I need to open each expression, find the correct parameters, and manually adjust them for each layer. As you can imagine, that takes a lot of time. So now let's learn how AI can help automate this process. When you want to control expression parameters dynamically, the setup is a little different. You've just learned how to make a simple setup. Now we'll build a more advanced one. To start, let's select our precomps and double press E to reveal the expressions. Now we'll delete all the expressions, to do that hold down Alt or Option on Mac, and click on the property stopwatch. Let's do this for all the precomps. And don't forget to clear the expression from the centered pre comp, as well. Great. Now let's go back to Chat GPT and ask it to create a new expression setup for us. I'll write the request in a very simple way, and I can even include some of the expression terms like the ones we saw earlier. So GPT understands which parameters we want to control. So I'll just copy these sentences from the expression GPT gave earlier, so it understands exactly which parameters I want to control. Next, I'll add the opacity parameters, as well. I'm simply telling GPT that I want to be able to control both of these properties. Let's see what it suggests. Here we go. Without even knowing this was possible, I just asked GPT if it could be done, and it gave me the full process step by step. It gave me the expressions and also suggested which numbers to use in the controller. Now, we might not fully understand what everything means, but that's okay. Let's just follow the steps and see what we get. First, I'll create a new null layer just like GPT instructed, and now I'll make sure to label the layers exactly as it suggested, so we don't run into any errors later on. Awesome. Now, let's see what the next steps are. We'll keep following GPT's guidance and create four different slider controls. You can find them in the effects panel either by right clicking on the null layer and going to expression controls or by typing the name of the effect into the effects search bar. Great. Now, let's add this effect to the null layer. Use Control or Command D to duplicate it three times, so you end up with four sliders in total. Finally, rename each slider with the exact names GPT gave. Just copy the names from the chat and paste them into the effect name field. Let's do this for all four sliders. Awesome. Before we go further, this is a great time to save a backup copy of your project, just in case after effects crashes. Let's save it as a new file. You can call it something like practice. It's always smart to keep a version of the project before animation starts, especially in more complex scenes so you don't lose your prepared layout if you ever need to start over. All right. Now, let's go back to the chat and copy the scale expression. We'll test it on the centered pre comp. In our case, that's picture six. Okay, so now we don't see the pre comp at all, but that's okay. We're not getting an expression error, so it's probably just a parameter issue in the controller. As you can see, all the controller values are currently set to zero. That's most likely the problem. If you do see any expression errors, here's my tip. Delete the controller name and type it again manually instead of copy pasting it from the chat. For some reason, after effects sometimes doesn't recognize pasted text from other programs. Even if the name looks the same, it won't work. And that's what causes those weird expression errors. So just retype the names manually, and the issue should go away. Anyway, let's continue following GPT steps. Let's go back to the chat and copy the opacity expression. Now paste it onto the opacity property. Then return to the chat and check what it says to do next. And here it is. Step three, he suggests adjusting the control values. So let's do that quickly. All right. Now, let's test it by moving our null layer from side to side and see what happens. Amazing. It works perfectly. Now I can adjust the controllers for the carousel setup and preview how everything looks just by moving the null layer. The great thing about this setup is that you can always come back and tweak the values anytime you want. Moreover, after we finish the lesson, you can save this project with this cool setup to use it as a template for your future projects. The only thing you'll need to do is replace the pictures with ones that are relevant to your project and adjust the controller if you want to make it look a bit different. For now, let's copy the expressions we added to this precomp and paste them onto the rest of the layers. To view both properties, first press T to show the opacity, then hold Shift and press S to also reveal the scale property. Now, let's copy just the expressions and paste them to the other layers. All right. Now we can test the controller and see if everything works. As you can see, the fox layer isn't reacting to the scale controller, so we probably forgot to paste the scale expression onto that one. Not a big deal. Just copy the scale expression from one of the other precomps and paste it onto the fox layer. Great. Now, I want to mention something really important when working with precomps. If you decide to collapse your precomps to maintain the highest possible image quality, be aware that it might change the appearance of your scene. As you can see here, the opacity looks different now. But in our case, this is super easy to fix. Just adjust the opacity controller in our setup, and it will automatically update all the layers. Awesome. Now, let's take it one step further and make the design look even better by adding a simple drop shadow effect. Let's tweak the shadow parameters and also change the shadow color so it blends well with the background. Here's a great tip. When adding a drop shadow, first sample the background color from your scene, then fine tune the color values. That's how you get more natural looking shadows. The shadow looks awesome, but I want it to appear only on the centered precomp. I'm trying to create a subtle elevated look. But as you can imagine, animating and timing the drop shadow for each layer manually would take forever. That's where our AI friend comes in again. To help GPT understand the context better, I'll take a screenshot of my scene. It's important that GPT sees the drop shadow effect in the layers in the timeline. So take a screenshot and paste it into the chat. Now, let's try to explain what we're trying to achieve. Again, using very simple language. GPT is now suggesting some steps. But since I already have a controller that manages the whole setup, I'm wondering if I can add this drop shadow automation to the same controller. So let's ask it. Great. It says we can. Let's follow the steps and see what happens. I'll copy the name now and go back to my project to follow the steps. So first, we need to create another slider control as he suggested, and we can paste the name we copied a second ago. Okay. Let's also set it to 100, as we saw in the suggestion. Back to GPT. Let's see what's next. Okay, I don't really understand all of it. There's a lot of explanation going on. Let's just skip ahead and copy the expressions. The most important thing for me is to check where the AI says to paste the expression. In this case, it says to apply it to the opacity property of the drop shadow effect. So let's do that now. Now I can control the drop shadow from here. I'll set it to 100 for now and check if it disappears automatically when I move the null layer to the sides. That's awesome. It works perfectly. Now I'll adjust the shadow color to better fit my design. And with that done, let's copy this drop shadow effect and paste it onto the rest of the precomps. Et's move the null around and see if everything works. Nice. The shadow only appears when the precomp is in the center. I'm happy with how the automation turned out. Since we now have a drop shadow applied to all the precomps, let's slightly tweak the opacity, too, because the drop shadow changed how it looks a bit. All right. So now we can animate the scene however we want. In my case, I'll just create a simple side motion for the carousel. At the beginning of the animation, I'll start by showing the elephant. Then after 3 seconds, I'll bring the tiger into the center. Finally, 3 seconds later, I'll move the leopard to the center. I just realized I might not have placed the elephant and the tiger perfectly in the center, so I'll go to the first keyframe and adjust the nulls position value. I'll do the same for the tiger. Awesome. Now, as we usually do, let's improve the motion by adjusting the keyframe velocity. But this time, instead of holding Alter option and double clicking on a keyframe, I'll use a shortcut I discovered not long ago to quickly open the keyframe velocity panel. After selecting your keyframes just hit Control Shift K or Command Shift K on Mac, and there you have it. Now, let's change the velocity to something like 75% for both incoming and outgoing keyframes. Let's preview that. Awesome. Looks great. Now let's go to Second nine and copy paste the first keyframe there so we get a nice looping animation. Press in to shorten the work area and go into the Speedgraph editor to check if the velocity is consistent across the full animation. Alright, everything looks good. And with that, we finish the first lesson. The most important thing I want you to take from this lesson is that sometimes it's worth spending a few extra minutes building an automated setup before jumping into animation. It can save you a lot of time later, because when the client comes back and asks for changes, you'll be able to adjust things super quickly instead of dealing with tons of keyframes and redoing everything manually. I recommend practicing this lesson further by adding a cool blur effect to the setup. You can ask Chat GPT to make all the non centered pre comps blurry and keep only the centered pre comp in focus and see what it suggests. The possibilities are endless, and I hope you now understand how easy it is to build stuff like this with the help of AI. I never used complex expressions in my own projects, but now I feel confident that I can build almost any automation I need without actually learning expressions or coding. Alright, now let's open the second lesson and close this one. And before starting to watch the next lesson, don't forget to take a ten minute break to refresh before we continue. See you in the next one. 3. Bouncy UI Animation: Hi, and welcome back. In this lesson, we're going to create a bouncy motion for the UI panel you see here. But before jumping into the animation, let me quickly show you how I set up the scene. So I have a main precomp of my avatar, and if we open the precomps for this lesson, you'll see there's one for the Avatar picture and one for the button. Inside the Avatar picture precomp, I want to give you a quick tip. When working on scenes like this, avoid making your precomp dimensions too large. It's better to work with small dimensions since we don't need to see this picture up close. Also, just like in the previous lesson, I use the track Mat function to mask the picture with a circle I created using the Ellipse tool. Now let's explore the main avatar comp. Here I have the rest of the layers, including the button precomp, which is just a simple rounded rectangle with a stroke and a text layer. All right, I'll close those precomps and let's get into the animation. This scene, I want to create an intro animation for the UI panel, but I want to make it more interesting. I want to add some secondary motion and make it move in a bouncy way. Once again, I don't really know what expressions to use, but I do know how I want the intro to look. In my case, I want the panel to enter the scene from the bottom. So let's create that basic animation real quick. First, let's open the position property. We can zoom in on the timeline to see the time codes more clearly. Since I want the precomp to end at its current position, I'll go to, let's say, second one and create a keyframe there. Next, I'll go to the beginning of the timeline and move the precomp down out of the frame. So now we have a very simple animation. Before we add any easy Es to the keyframes, let's take a quick screenshot of the scene and ask hat GPT to help us make this animation bounce. Make sure the position property is open so GPT can see the keyframes we've created. All right. Now let's go to Chat GPT. And instead of continuing the chat from lesson one, let's start a new conversation by clicking on the new chat icon. Great. Now we can paste the screenshot and start chatting. First, always begin by telling GPT what software you're using. Next, describe what you're trying to create. And before we move on, I've got a question for you. How should we end our message before GPT helps us? That's right. We need to ask, do you understand what I mean? That's the most important step. We want GPT to describe our vision back to us before it gives a solution. Only then can we be sure it's on the right track? So GPT already prepared the expression and explained how to apply it. He also gave us some tips in case we want to adjust the bounce settings. Here's what he thinks we want to achieve. As you can see, he got it really well, and he even mentioned a new more accurate term for what we're trying to create. He knows we want to create an overshoot and bounce animation. He already prepared the expression and explained how to apply it. And he also suggested what to do in case we want to adjust some of the parameters of the overshoot motion. All right, let's copy the expression and apply it to our position animation to see what happens. I Awesome. We already have a nice bouncy effect. But in my opinion, the animation feels a bit slow. Let's bring the last keyframe a bit earlier in the timeline. Place it around frame 20. Now, zoom out in the preview and check if the expression GPT gave you is also triggering an overshoot on the first keyframe. If it does, don't worry. Just go back to the chat and ask GPT to make the overshoot bounce effect apply only to the second keyframe. Alright, back to the lesson. Now, let's easy ease the keyframes and see what we've got. As you can see, the bouncy effect is barely visible now, and that's because the velocity of the keyframes changed after easing. In that case, we can follow GPT's suggestion and adjust the amount of bounce in the expression. Even if you know nothing about expressions, GPT usually writes them with short notes inside. So you'll know which values affect the motion the most. All we need to do now is try different values until we find something we like. First, I'll change the bounce value in small steps. Then I'll try a bigger jump to see what it gives me. I'm showing you this because I want you to feel free to change values and play around with the settings when practicing on your own. Nothing bad can really happen. Worst case, you just press undo or ask GPT again for help. Alright, now that we found a bounce value that works, I want to fix a pause that's happening in the animation. It's important that you understand why that pause is there. Really simple. When we easy ease keyframes, their velocity changes. And as you can see here in the graph editor, the speed at this final keyframe is currently zero. To fix this, we can convert the final keyframe back to linear by holding control or command on Mac and clicking on it once. Now, the speed is around 13 54 pixels per second. Before it was stuck at zero. Now we've got motion again. So now the animation will continue smoothly with the bounce effect. But as you can see, it's a bit too much right now, and it's because we originally adjusted it based on the eased keyframe. But now we have a regular linear keyframe. So now we'll fine tune the bounce settings again to match the new speed. I'm showing you this process because I want you to explore and get hands on when needed. Don't be afraid to test things. It's the best way to learn quickly. Now, let's make the animation start more slowly by pulling the handle of the first key frame to the right. In my opinion, it already looks awesome, and we're ready to move on. Of course, you can always change the values to get a slightly different look if you want. But now I want to make this animation look even more dynamic by adding a simple rotation animation along with the position. I'll also try to add a fun bouncy motion to the rotation, if that's possible. So let's start by creating the rotation animation. Select the layer, press R, and then go to frame 20. Create a keyframe here. This will be the final rotation position. Then go back to the beginning of the timeline and rotate the precomp. Let's rotate it to 25 degrees and adjust the position at that point so the UI panel is hidden off screen. All right, it looks nice, but I want to go back to Chat GPT and try to come up with a more interesting animation. Let's just tell it what we did and what we want to achieve. I want the rotation to feel a bit wiggly. And since I'm not sure I explain myself clearly, I'll ask GPT. Do you get what I mean? Now I can see some explanations, and GPT is using terms like jitter or shaking motion. So maybe I'll ask it to make the layer jiggle a little and then stop. Let's see if it understands the task. After reading the explanation, I'm still not sure it's what I meant. So in that case, I'll just try it out and correct GPT if needed. That's the helpful part of exploring AI and after effects. I really want you to follow these same steps and not be afraid when trying things out on your own. Okay, so now I see that the motion is a bit jiggly, but it still doesn't look how I imagined. Before going back to the chat, I'll try adjusting some of the values in the expression to see if that helps. After some testing, I still don't get the exact motion I want, so let's cancel the expression, and now go back to the chat and explain it better. I'll start by saying, This isn't exactly what I meant, and then describe my idea in an even simpler way. As always, I'll finish by asking, Do you understand what I mean? This time, GPT says, I fully understands. So let's see if that's true. Awesome. This is exactly what I was going for. Now I can play around with the values and tweak things until it looks perfect. So let's play around with these three values under the setting section in the expression. All right. I think it looks nice. After that, I'll adjust the keyframes and see how they interact with the rotation expression. Looks great now. I'll also adjust the velocity to better match the position motion. Again, it was really important for me to show you this back and forth process with GPT. Sometimes we don't get what we want right away, but if you explain things clearly, GPT will get it eventually. Alright. Now that we've created the intro for the UI panel, let's enter the precomp and create some cool micro animations inside it. In this case, I want to add some bouncy scale animation to layers like the Avatar picture, the star, and the button. For the other layers, we can reuse the bouncy position animation we created earlier for the main precomp. So let's go back to chat GPT and ask it to create a bounce or pop up scale animation. We can also mention that we want it to be similar to the popping position animation we used earlier. So here we've got all the explanations and suggestions for the pop up scale animation. Let's go back to the project and create the scale animation for the Avatar picture so we can add GPTsEpression to it. Just like before, since I want the final scale to be what it is at the end of the animation, I'll go to frame 20 and create a keyframe with the current value. Then at the beginning of the timeline, I'll set the scale to zero. You can also isolate this layer so we can focus only on it. Now that we've got the keyframes in place, let's try using the expression GPT gave us. Great. It works perfectly. Now, I'll adjust the easing, so it matches the position animation we created for the main UI panel. After that, I'll play around with the expression settings to see what looks best for this specific layer. Alright, let's apply the same expression to the next element we want to pop out. Let's do it for the button precomp. First, create the scale animation, then apply the expression. Just select the layer and paste the expression. Let's see how it looks. Now, adjust the velocity and tweak the bounce effect until it feels just right for this layer. Awesome. This looks great. Let's do the same thing for the star and the big number. Now, to apply the expression, select both layers and hit Control V. Let's see how it looks. Since the number layer is pretty small, I prefer to make the bounce a bit more noticeable so it pops nicely. That looks great, too. Now, let's create a pop up animation for the icon I have here. This file is really big, but I scaled it down to a very small value. So let's see if the expression still works on it. Sometimes GPT might create an expression that's based on the actual scale value of the layer, and that might not work well for every layer, especially if they're all scaled differently. If that happens to you, just go back to GPT and ask it to build the expression using percentages like we're doing now. If you run into any issues, just explain that you're working with layers that have different scale values, and you want a universal pop up animation that works no matter what the layers scale is. All right. Now that we've finished the pop up scale animations, let's create a simple position intro for the text layers. First, go to frame 20 and create a keyframe for their current position. Then go back to the beginning of the timeline and move the text layers down and out of the frame. Now, copy the same bouncy position expression we used for the main UI panel and paste it onto the text layers. Finally, adjust the easing and see how it looks. That looks great, but we do have a small problem. You can see the text layers entering the scene outside the borders of the main blue shape. I've got a question for you. What do we need to do to fix this? We need to make the text layers use the alpha mat of that shape. Then just turn the visibility of the shape layer back on. Now, everything looks perfect. We've got the text visible only within the shape layer. The last thing we need to do is time the micro animations we created inside the avatar pre comp to match the animation in the main comp. Let's start the intro animation for the layers inside this comp from frame 15. At this point, you can arrange it however you like. In my case, I want the text layers to enter the scene first, so I'll start with the name coming in. Then after two frames, the next line of text. After that, I'll make the avatar picture appear followed by the icon. And finally, I'll finish with the star, the number, and the button. Now, let's check the animation and see how it all looks together in the main comp. I think the inside animation should start a bit earlier. I let's try moving it to frame five. Yes, that looks much better. And with that, we finished this lesson. I hope it gave you a little more confidence to go out and start automating your own animations. Also, I hope this lesson helped you see that sometimes it's better not to create a single controller for all the layers like we did in Lesson one. In this case, it was a smarter choice to control each layer separately and adjust the values one by one. With that said, we're now ready to move on to the next lesson where we'll create a very cool hover effect for the icons. It's going to be awesome, so see you there. 4. Advanced Hover Animation Setup: Come back. In this lesson, I want to teach you how to create an advanced hovering animation setup. But before we start, let me explain why I precompose my icons instead of using the original layers. The main reason is that each icon has a different size. If I animated them as is, the scene would be much harder to control. When you work on scenes like this, I suggest doing the same. It also makes it really easy to swap icons in the future if you ever need to. As for the pre comp size, I suggest working with small dimensions, something like 300 pixels for both width and height should be fine. Use bigger dimensions only if you know the icons will appear up close in your seen. Anyway, as you can see, the size of this icon layer is very large, and I scaled it down to 14. Same thing for the Illustrator icon. They came in at the same original size. But the Photoshop icon had a different original size, so I scaled that one to 46. And for the Figma icon, I used a scale of 33. My goal was to make them all appear the same size visually, and that's why I precompose them and use the same dimensions for all the precomps. All right, so I hope you get the idea. Now, let's start building the hovering setup. Let's imagine we don't have a clue what to do first. So we're going to ask Chat GPT to help us. With that in mind, I've got a question for you. What will help us explain the situation to Chat GPT clearly? We need to take a screenshot of our scene. So let's do that real quick. Make sure your screenshot includes the preview panel and the layers panel. Here's a small bonus tip. To improve the clarity of your message to GPT, you can draw a quick sketch or arrow on your screenshot. In this case, I'm drawing an arrow from the cursor to the icon. All right. So now let's copy this screenshot and go back to chat GPT. Let's create a new conversation and paste the screenshot in. And as always, let's begin by telling it what software we're using. Then let's try to explain the animation we're trying to create. Don't worry about making grammar mistakes. Just try to be as specific as you can. This time, I'll already tell GPT that I want to control the different parameters of the hover effect because maybe I'll want to change them later. I'll mention the parameters I already know about, and I'll also ask GPT if there are any other parameters I'm missing that could improve the setup. And as always, let's finish by asking, do you understand what I mean? So here we have all the steps we need, and I can see GPT actually added a really nice parameter I didn't even think about, which is being able to control the default size of the icons. Great. So let's keep going and see what else we've got. I noticed in step three that GPT explains how to make the motion smoother. That gave me the idea to ask if we can add this parameter to the controller so we can adjust it without having to open the expression manual. And here it is. GPT gave us the expression for that parameter, too. So now let's follow the steps from the beginning. First, let's go back to our project and create a controller layer using a null object. Make sure to name this layer exactly as GPT suggested, because if the names don't match, the expression won't work. All right. Next, let's create the three sliders and label them exactly as he recommended. Now, we'll adjust the default values for each controller, just like GPT told us to. And before applying the expressions, let's also add that final controller for smooth motion and set the default value as suggested. Now, let's copy the updated scale expression GPT created and paste it onto the scale property of the first icon. Let me resize this panel a bit. If you're having trouble opening the expression field, no worries. Just cancel the current expression and Alt click on the Stopwatch again. All right. After pasting the expression, I see an error, so I'll click the expression to reveal the error message, then use the shortcut to open the screenshot tool and take a snapshot of it. Make sure your screenshot includes the full error message. That's really important. Now, paste that screenshot into the GPT chat and let it take care of the problem. Here we can see the explanation. Looks like the issue is with a name I used. I might have typed something wrong, but even if you don't see a mistake, I still suggest you delete the name and type it again manually. Like I explained in Lesson one, copying names from outside sometimes causes weird bugs. And there we go. The error is gone, and the expression works perfectly. Great. Now we can start playing around with the parameters to see how they affect the motion and overall look of the scene. Now, before applying this expression to all the other precomps, let's try to improve the hovering effect by making the icon move up slightly when the cursor hovers over it. Let's go back to GPT and tell it that we want the icon or precomp to move up smoothly when hovered. Awesome. GPT suggests adding another controller so we can adjust this parameter whenever we want. Let's do that real quick. Now, apply the expression to the position property, just like GPT tells us. Let's check it out. Try changing the value on the new slider to see what looks best. Once we're happy with how it looks, we can copy the expressions from this layer and paste them onto the rest of the icon precomps. Everything looks great, and the best part. Now we're controlling all the icons from a single controller. You can change all the parameter values anytime you want. How cool is that? You can even flip the effect around. So instead of icons growing on hover, they shrink. So now we can even reverse the effect, make the icons shrink when hovered instead of growing. All right. I'll reset everything back to normal. And now let's start animating the cursor layer. At this point, as always, you can create any kind of animation you want. You don't have to copy exactly what I'm doing. My main goal in this course is to help you learn how to build expression setups using AI, not to focus on detailed animation design. And I hope by now you're starting to realize just how easy it is to generate complex expression systems that used to take a lot of time to learn and create manually all thanks to AI. I Alright, I think my animation looks pretty good, and with that, we've finished this lesson. You can go ahead and open the next lesson and close this one. In the next lesson, we'll learn how to make a cool hovering effect for buttons instead of icons. We'll also talk about what to do in situations where AI refuses to give us the result we want no matter how hard we try. It's going to be awesome, so I'll see you there. But before jumping into the next one, I recommend taking a ten minute break. 5. Hover Color Change Automation: Hi, and welcome back. In this lesson, we're going to create a cool hover color change effect, and we'll also talk about what to do when we run into annoying issues with hat GPT. As always, let me quickly show you how I built this simple scene. What I have here is just a few precomps representing buttons. Inside each precomp, there's a text layer and a rounded rectangle. It's important to know that all the methods we use in this course work just as well with regular layers too. You don't have to use precomps if you don't want to. Alright. Now, let's start building the expression. I want the buttons to change color when I hover over them with the cursor. First things first, let's take a quick screenshot of the scene and the layers. Now, let's copy the screenshot, head to chat GPT, start a new conversation, and paste it in. Then we'll describe our idea in a very simple way. Since we have no clue how to make the color change, we'll just ask GPT what effects we should use and how to do it. And as always, we'll finish by asking, Do you understand what I mean? All right. So here we have the full process, and we can also see the effects GPT suggested. I think the tint effect is better in this case, since it comes from the color correction family of effects. Then I noticed GPT created an expression for the fill effect. But for now, let's ignore that and go back to the project to understand what the tint effect actually does, since that's the one I want to use. Search for the effect in the effect and presets panel and add it to one of the button precomps. Let's now change the colors inside the tint effect to get the look we want when the button has hovered. Once that's done, check out the amount to tint property. This controls how strongly the new colors are applied. But let's say we didn't notice GPT created the expression for the fill effect, and we copied that expression and pasted it into the amount to tint property anyway. As you probably guessed, we'll get an error message. Now, let's say I don't understand the error, but I still want to fix it. First, we can cancel the expression. In situations like this, the best thing to do is take a screenshot of your current setup and make sure GPT can clearly see what effect you're using. Then paste the screenshot into the chat and tell GPT, I'm using the tint effect. Can you tell me where exactly I should paste the expression? And right away, GPT gives us new steps in a working expression. Let's copy that and test it in the project. Nice. It works. But honestly, I don't love how the effect only triggers when the cursor is at the center of the button. And I'm not a fan of the weird midtone color blending that happens when the cursor is getting close to the center. Let's say I don't understand why that happens. I just want the color change to happen instantly when the cursor touches the button. In that case, just go back to GPT, tell it you don't like the current behavior, and describe what you want instead as if you're talking to an assistant. A after explaining, GPT will create a new expression for us. Let's copy and paste it into the expression panel. Awesome. It looks much better now. Let's try collapsing the pre comp to check if anything breaks or needs adjusting in the expression. Everything still looks great, so we're good to move on. So now we've applied expressions to effects and basic layer properties. But let's take it one step further and try using expressions on layer styles. For example, right click on the precomp, go to layer styles and add a stroke effect. Open the stroke settings, set the size to ten, and change the color to white. Now, I want this stroke to appear only when the button is hovered. That way, we'll create a more unique and dynamic hover effect. Take a screenshot of this updated setup and go back to Chat GPT. In the chat, briefly describe the situation and explain what you want to happen. And there it is. GPT tells us to apply the expression to the opacity property of the stroke effect. Let's do that and see what happens. Great. It looks really cool. Now we can copy this layer style from the first precomp and paste it onto the rest of the button precomps. After quickly confirming that it works, let's do the same for the tint effect. Copy and paste it to the others. I Looks awesome. Everything's working perfectly. Now, let's create a quick animation for the cursor to practice a bit. As usual, you can animate it however you like. But don't spend too much time on it. We want to move forward and keep exploring what AI can help us build. For example, I want the cursor itself to change color when it hovers over a button. Why? Because in some scenes like this one, the contrast between the hover color and the cursor isn't strong enough. The cursor becomes, but first let's finish the animation, and then we'll try using AI to help us generate the idea and the solution to make the cursor more noticeable on hover. All right. So once we are happy with the animation, let's start working on the cursor changing color automation. Let's take a screenshot showing the cursor hovering over the button and send it to GPT so it can understand our issue visually. Once pasted, we'll explain what we want to do and how we hope to solve the problem. O. GPT suggests using the fill effect on the cursor and then using the expression on the color property of this effect. So let's follow the steps. First, we need to add the fill effect on one of the button precomps. Next, let's change the color. Let's say to yellow. We can also scale down the cursor shape a bit, set the size to 65. And now let's paste the expression GPT gave us into the property it suggested. I see that the cursor doesn't change the color at all, even if we try to change it manually. But nothing happens. The color doesn't change. We've pasted the expression into the correct property, so now let's tell GPT what happened. We'll ask if maybe it's better to control the opacity of the effect instead. That worked when we used opacity for the stroke layer style earlier. All right, so let's copy the new expression from GPT and go back to the project. Cancel the expression on the color property and instead apply it to the opacity of the fill effect. Still, nothing happens. Now, the cursor is invisible for some reason, even though we don't have any expression errors. Let's keep trying. Take a screenshot of the situation and tell GPT. We applied the expression, but now we can't see the cursor at all. A GPT responds with a longer explanation and a new version of the expression. Let's give it a try. Copy the expression, and let's quickly go over the rest of the explanation to see if there's anything important we might need. Let's say we don't really understand what he's saying. So let's just go back to the project and paste the new expression. Okay, I still can't see the cursor at all. Changing the values does nothing. Honestly, after trying more than twice, I'm starting to feel that maybe this approach just won't work. When I check the animation, nothing happens. So clearly, something's wrong here. We can cancel the expression, and moreover, in situations like this, I prefer to just try a completely different approach. So let's remove the fill effect from the cursor and try it with the tint effect instead because that worked better earlier. So let's add the tint effect to the cursor, then change the color to yellow and now take a new screenshot and go back to the chat, et GPT know that the fill effect didn't work, and now we want to try tint. Let him do his thing. Then copy the new expression. It might be a bit complex and try it in the project. Still not working. No errors, but the color isn't changing. Let's try again and show him the current situation. Take a new screenshot and make sure to include the expression in the picture. After pasting the screenshot, let him know that this expression doesn't work. In the meantime, while he's generating the answer, let's go back to the project and cancel the expression. We can also try changing the second color of the tint effect to yellow. Maybe that'll help. All right. The new expression is ready. I'll quickly scan his explanation and suggestions to see if there's any important clue for this setup. I didn't find anything critical, so let's just copy the new expression and see what happens. Still doesn't work. Now, before starting a new conversation with the chat, I want to try one last thing. Let's go back to the chat and tell him again that it's not working. But this time, I'll also give him some technical information and let him know that the opacity stays at zero the whole time. Sometimes adding just a little technical detail like that helps GPT better understand what's going wrong. So let's just copy the new expression and see what happens. Awesome. This time, it works perfectly. I wanted to show you this entire process so you'll understand that sometimes you need to switch your approach, try a different effect, or add a bit of technical detail to help AI assist you better. Remember that when dealing with tricky setups like this. Let's watch the animation we just created. Awesome. Looks nice. Now, before we wrap up, I want to give you a quick tip. Earlier, we added a stroke effect to all the button precomps. Let's say now you want to change the stroke color, but you don't want to do it manually on each one, especially if you might change the color again later. Here's a trick. Search for the color property using the search bar in the timeline panel. So now we can see the color property of the stroke for all the precomps. Then use the pick whip of the property itself, not the layer parenting pick whip to link all stroke color properties to the first one. Do this for each of the button precomps. Now, you're controlling the stroke color for all buttons from a single place. This works for any effect property. You can use this trick if you don't want or don't need to set up complex controllers like we did in the first lesson. Another smart way to control the properties of an effect from one layer is to copy it with the property links. That way, after you paste it onto the rest of the layers in the scene, you'll be controlling all the properties of that effect from just one layer, and you don't need to parent the properties one by one. And with that, we've finished this lesson. Let's close all the layers and watch the animation. Notice that even though we don't have a lot of layers or effects in the scene, the preview render times are super slow. That's happening because we've got some complex expressions inside the layers. Telling you this because I want you to know using too many expressions in your project can really slow it down. Alright, the animation looks great, and we can move on to the next lesson. So let's open it and close the current one. In the next one, we'll learn how to create a super useful cursor click automation. It's going to be a great one. So see you there. 6. Automating Cursor Clicks with Markers: Come back. In this lesson, we'll learn how to create a super useful cursor click automation focused on making layers react to the click animation of the cursor. To start, let's first create the clicking cursor animation. And what I mean by that is, I want to activate the cursor's click without creating keyframes for every single click. Whenever you need a clicking animation for a cursor, it's best to have the cursor in a separate precomp. So let's start by creating the cursor. And this time, let's make it look beautiful. Select the ellipse tool and set the fill to use a radial gradient. We'll deal with the colors in a second, so don't worry if it doesn't look exactly like mine. Next, turn off the stroke. Quick tip. If you want to switch the fill style quickly, just hover over the fill box, hold Alt or Option on Mac, and click. This will cycle through fill types. Same thing goes for the stroke box. Now, hold Shift and draw a small circle in the scene. Let's set the size to 100 so we're all on the same page. Next, align it to the center and start editing the gradient. From my previous gradient tests, I have a bunch of saved colors. I'll drag them out to delete them. Now, pick a nice color for the first gradient stop. Make sure the opacity is set to 100. Do the same for the second color. I'll choose a dark pink, and again, make sure the opacity is 100. If you want to add more color stops, just click along the line. You can then choose any color for the new point. If you change your mind, drag the color stop out of the line to delete it. Now, to adjust the gradient appearance, switch to the selection tool to reveal the gradient handles. Zoom in and tweak them to get the look you want. If you accidentally click somewhere else and the handles disappear, just reselect the ellipsis shape layer or open the gradient fill property from inside the layer. If that still doesn't work, you can adjust the start point or endpoint values manually. That will make the handles appear again in the viewer. All right. Now, let's add a little highlight to the sphere to make it look even better. Right click on the shape layer, go to Layer Styles and choose inner shadow. Open the effect settings, change the color to something bright. And then change the blending mode to screen so it blends with the base color. Now, you'll see the highlight, tweak the values to get a soft, glowing look. Once you're happy with the cursor's design, we'll move on to creating the actual click animation. Like I mentioned earlier, to automate the clicks later, we should precompose the cursor. Call the new precomp cursor, make sure both checkboxes are selected and click Okay. Now go inside the cursor precomp. Here you can create any click animation you want. We'll start with something simple and then add some micro animation to enhance the effect. So first, let's create a simple scale animation. We'll make it ten frames long. I'll start at 100% scale, then shrink it down to 70% and bring it back up to 100. Now, I'll easy ease the keyframes and adjust the motion in the graph editor to make it feel more natural. Looks good. But I think we can make the click feel even stronger. So let's go back and drop the middle keyframe to 60 instead of 70. Yep, much better now. Now, let's add a little micro animation to make the click look even cooler. Kind of like a ripple effect. So I'll select the Ellipse tool again, but this time, I'll turn off the fill and turn on the stroke. I'll color it white and set the stroke width to four just for now. Now I'll create a slightly bigger circle. Let's make it 400 in size, and I'll align it to the center. For the animation, I'll go to the end of the click animation and set a keyframe for the scale. Then I'll go back to the beginning and set the scale to zero. Now I'll also animate the stroke width. At the end of the animation, I'll set it to zero. And at the beginning, let's set it to 20. Let's try 60. Hmm. Let's try 40 instead. I'll easy ease the keyframes and adjust the velocity, so it starts quickly and eases out. Now, I'll move this ripple shape behind the main sphere. Let's preview that. In my opinion, it's too fast, so I'll select all the key frames. And while holding Alt or Option on Mac, I'll drag one of the last key frames to frame 15 to stretch it out. Let's play that back again. Yeah, looks great. Now that the click animation is finished, let's clean things up. We don't want a giant precomp for no reason. So let's resize the precomp to fit the animation. Set both width and height to 500 pixels. Awesome. Let's head back to the main comp and figure out how to activate this click animation without duplicating keyframes every time we want the cursor to click. So let's ask our new friend Chachi PT to help us automate the click. First, take a screenshot where we can clearly see the preview panel and the Layers panel. Now, start a new conversation, paste in the screenshot and explain the situation. It's important to tell GPT that we already have a click animation ready inside the cursor pre comp. Since this task is fairly simple, and I explained it clearly, I won't ask, Do you understand? I'll just ask how to do. Now, I see GPT gave me two options and both involve the time remap function. So I understand that no matter which method I choose, I'll need to enable time remapping on my cursor comp. I've decided to go with GPT's second suggestion, which uses an expression that activates the animation using markers. Let's see if that works. First, enable time remapping on the cursor layer. Nothing happens yet. That's fine. Let's go back to the chat and copy the expression GPT gave us for the time remap property. Now return to the project. Alt click on the Stopwatch next to Time remap and paste in the expression. Nothing happens yet. That's fine. Now, let's test it out. Go to second one and add a marker to the cursor layer. Check the shortcut for adding a marker on your device if you're unsure. Boom, it works perfectly. Now go forward another second and add another marker. Nice. Let's preview it again. Awesome. Now, let's delete the test markers and create a simple hover animation for the cursor as it moves over each icon. At the moment, the cursor motion is linear, so it doesn't pause when reaching the icons. Instead of manually creating pause keyframes, let's just easy ease all the position keyframes. This will make the speed slow down to zero before and after each one, which gives a pause effect. If you want to fine tune the easing, you can do it manually in the graph editor or just use the shortcut Control Shift K or Command Shift K on Mac to open the keyframe velocity panel and adjust the influence. Now we've got nice smooth motion with pauses between icons. Let's go ahead and add markers at each point where the cursor pauses. That way, the clicking animation will trigger exactly where we want just by using those markers. Let's preview the whole thing. How cool is that? We're triggering the click animation just by placing markers. No need to copy keyframes or ripple layers manually. Imagine how long this would take if we had to duplicate all the keyframes and layers for every single click. Now, let's take this automation one step further and make the icons react to the clicking animation without animating them manually. I'll go to the point where the cursors hovering over one of the icons and take a screenshot of the scene, making sure it includes the markers and the keyframes we just created. Then I'll paste that screenshot into chat GPT and explain what I want to do. This time, since I'm not 100% sure I explained it clearly, I'll ask GPT. Do you understand what I mean? I love asking that because GPT will often reply using the correct professional terms for what I'm trying to build so I can learn how to better communicate these ideas. All right, GPT understood and gave me the steps. Since I added the screenshot, it already knows what's in my scene. So all I need to do now is follow the final step. That step is applying the expression to the scale property of the icon precomps. Let's try it on one icon and see what happens. Perfect. It works great. But I'm not a big fan of how the scale animation feels a bit stiff. So let's go back to the chat and ask if it can be smoother. As you can see, GPT already anticipated that and asked if I want the motion smoother, so I'll say yes and wait for the updated expression. Let's test the new version. Awesome. That looks much better. Now that we're happy with the results, we can apply this expression to the rest of the icons. But before we do that, I have a quick question for you. What's the correct way to copy an expression in after effects? We need to use copy expression only, not just regular copy. Looks great. And with that, we finished this lesson. I hope you learned a lot of new and most importantly, useful techniques during this session. Thanks for watching, and I'll see you in the next one. 7. Outro: Hello, it's me again. I just wanted to say thanks a lot for joining me in this course, and congratulations on finishing it. I hope you learn a lot of new ways to use AI to speed up your animation workflow, and you now feel more confident using expressions. If you enjoy the course, I'd appreciate if you leave a quick review. It helps me a lot. Thanks again for watching and I'll see you in the next one.