How I Make Insane Motion Graphics with Opus 5.5 (Better than Fable or Astra)

0:00 Opus 5.5 can now create mind-blowing motion graphics that looked like a professional designer spend hundreds of hours on them. So what I did was give it a video style that I want to copy and a

0:09 link to my app and it created a product launch video for me. The entire thing with smooth animations and everything is adapted for my app. But not only that, but it was even able to create something

0:19 that's a lot more advanced. A cinematic video with smooth motion graphics layered on top of it. just like those videos that you see from Anthropic or OpenAI when they release a new product

0:29 cuz these are created by a full production crew and a bunch of motion designers, but now I can create them by just talking to Opus 5.5 in plain English. So today I'm going to show you

0:38 exactly how I made all of these videos step by step. And even if you've never created motion graphics before in your life, you're going to be able to create something like this by the end of this

0:46 video. So let's get started. Now, you've probably seen videos like this circulating on X since the launch of Opus 5.5. And I've been fascinated with a lot of these motion graphic videos

0:56 that many people have been able to create. But now the problem is whenever you see videos like this and when you actually try it out yourself, it doesn't come out as good. So I wanted to test

1:04 this myself and see if OP 5.5 can actually deliver this time. And the results actually blew me away. So today I want to split this video into two levels. Basically level one, we're going

1:14 to start with this motion graphic video here which is the product launch video by this company called Ren. And we're going to recreate this video using this style but adapt it to my own product.

1:24 And after that, we're going to switch to level two, which is a lot more advanced, where we combine smooth motion graphics and layer it on top of cinematic video like this with actual people, which is

1:33 how a lot of the product launch videos are done by huge companies like Anthropic and OpenAI. So, let's get started with the first video. Now, before you create any type of video, of

1:41 course, you want to find a video that you can reference. And I think the best place to go is to browse X. If I simply type in open 5.5 motion graphics, you'll see a lot of these type of videos that

1:51 are created by different users here. But who actually has the time to go through all this and find all these videos manually, right? So, what I like to do is go to this website called

2:00 watchships.com. So, this is basically a compilation of all the videos that exist on X consolidated into a searchable database. So, you can basically filter by category. You want design, you want

2:12 motion, productivity, consumer, and here you can find all the product launch videos by all these different brands here like ramp and stripe. We got one from cursor, got one from OpenAI, and

2:21 this is the dots that they just released recently. So, what he's trying to do here is that you want to find something that you can use to give it a clot. So, it doesn't need to think from scratch,

2:29 but instead follow a style that already works. So, I found this video here by this company, Ren. I think that's how you pronounce it. And I think this will work perfectly for our example here. And

2:39 basically this is like a chatbot type of thing. So we've got chat bubbles, we've got moving text, we've got pictures of people, notifications. So I think this is a good example to try out capability

2:49 of OBS 5.5. So what I can now do is grab this URL and go to cloud start new session and just paste in that link along with a simple prompt. So what I'm saying here is that in this link there's

3:00 a product launch video for this app. Ren, can you watch this video frame by frame and let me know if you're able to understand the video. I want to recreate this for my own app. And I'm just going

3:09 to hit go. And while that's working, I also want to give Opus some details of my own product. So for the sake of today's video, I'm going to give it a fictional app here, Quicksand, that I

3:17 just created in my last video. So basically, this is an invoicing app. And within this app, I do have a chatbot that I just created here. It's a fictional one called Qbot. And what it

3:27 does is basically helps people with following up with invoices. And it's able to see what's overdue, sending you reminders, things like this. Right? So, I'm going to give this to Opus as well,

3:36 so you can study how my icons are designed, my color palette, so you can adjust all the motion graphics to fit the style of this app. Now, if I go back to Claude here, it's done analyzing the

3:46 video. It's watch the video frame by frame and it's broken down the video into scenes. So, now I know that it understood everything. So, what I'm going to do now is I'm just going to

3:55 paste in that URL to my app and I'm going to say here's my app called Quicksand. I want to create a product launch video just like Ren's but adapted for my app and focusing on the Cubot

4:06 chatbot feature. And if you're missing any UI elements from my app design, you can simply create your own to complete the video, which simply means that it's going to take all of these elements here

4:15 and if it's missing something, it's just going to create it from scratch. So, I'm just going to send it off and wait for it to finish. So, fast forward to a couple minutes later. This is the video

4:24 ready for us. Now, I did ask it to remove any audio here cuz we're going to add that at the end later. So, I'm going to play this real quick. Yeah. So, you can see that everything is

4:34 pretty smooth so far. It's taking the styling from the brand video and it's adapting to my product, which is Cubot. And it's also taken my colors as well, which is this purple.

4:46 And all the words are adapted for my own product. So, yeah, again, this is really, really good so far. All right, that's my outro. So, I think

5:00 overall it did a pretty good job. I've never gotten close to this level of quality with any other model. And considering that this is the first version, I think this looks really good.

5:08 Now, of course, there are a few scenes here that I think can be improved, which we'll get to in a bit. But one thing to understand is that all of this is not actually a video cuz all the motion

5:17 graphics you see in this video is made from code. And because it's just lines of code, it is very easy and fast for Claw to make changes. Now, speaking of changes, I got a few things that I do

5:27 want to change. For example, if you go to this scene here, you're going to see that all of these circles are supposed to be images of people of these clients, right? But since Claw does not have

5:36 access to an image model, we haven't given it instructions yet. It's just giving us placeholders, which is totally fine. We can fix that later. And towards the end here, I think I prefer this to

5:46 be showing the name of the chatbot, which is Qbot, and then maybe it flips into quicksand, which is the name of the app. And I think it'll also be nice to have some sound effects as well. So when

5:54 the chat bubbles appear, I do want that sound just to make the video come to life. So I'll just prompt it again. Hey, can you use his field to generate images for the profile photos and also for the

6:05 outro? I wanted to say Qbot first with its logo and then it flips to quicksand. And can you also add sound effects so that when the chat bubbles appear, we have sound. And in another few minutes,

6:14 we have the new version ready. So you can see that we now have sound effects. >> [bell] >> Yep. So now we have the images here.

6:46 Okay. It looks really good so far. Okay. All right. Awesome. Exactly what I wanted. So, that's how easy it is just to make some adjustments. But there are other times where you just want to make

7:05 quick corrections. Maybe you want to change the text to something else or you just want to make certain elements bigger and it's annoying to have to take screenshots and do markups and send it

7:14 back to Claude. So, I actually created this video editor here and I designed this like a simple version of Cap Cut or Premiere Pro just so you can make changes easier, right? So, you can

7:23 actually see the timeline here and we've got one, two, three, six different scenes here and you can scrub through the timeline and you can see how the video was actually made. So, for

7:31 example, if I want to make a change, I can simply click on edit here and I can go into the screen and I can now make changes, right? I can simply click on the right side here which gives you the

7:41 attributes of the item that you selected here. So, let's say I want to add something here, uh, the Qbot. And then maybe I want to make this a little bit smaller. So, I can use this to drag it.

7:50 I can move things around. This actually looks funky now, but when you actually send this to Claude later, it's going to fix everything and it's going to understand your request. So, if I go to

7:59 another scene, let's say I want to see how this scene was actually made. I can probably click on this profile photo here, and it actually brings up a menu here of every single profile here. And

8:08 maybe I just want to change the status here from set to something else. It actually gives me an option here. Maybe we can do paid. And then that will save my requests and I can continue to make

8:17 further request. So let's say this one I want to change it to add a exclamation point. So now when I'm ready to send all my requests to claude, I can just click on send to claude here and it will give

8:28 you a full prompt of all those changes that I requested that I can just copy and paste it into claude. So this way you don't have to take screenshots, you don't have to do markups. Claude will

8:38 know which exact duration or time stamp you want those changes on. And I can now go back to Claude and paste in that prompt. Now, if you guys want access to the editor, you can just download this

8:48 skill here that I just created called motion OS. So, simply give Claude this URL, ask it to install it. And whenever you're creating a motion graphic video and you want your video or your output

8:57 rendered on that player, you can simply type in your prompt as usual, but simply ask Claude to render it on this motion OS player. So, if you guys want access to this, it is free. just grab it in the

9:07 description below. Now, for this particular video, I think there's one more thing that I want to add, which is background music. Right now, if you give Opus the freedom to create its own

9:15 music, it's probably not going to be good cuz it does not have the ability to understand what good music or what music is good to match the type of video that we're creating. And unfortunately, as of

9:23 today, this is something that you got to do manually. Now, to solve this issue, you can find any royalty-free music that you can use. But for me, I use Epidemic Sound. This is not sponsored or

9:32 anything, but they just have a huge library of any type of music that you want. And they are royalty-free if you have a paid account with them. Now, I've already done the research and pick a

9:40 music for this. So, I just downloaded that track from Epidemic Sound. Put it in the folder and say, "Hey, can you add a background music to this video? Make sure that it fades in and out smoothly

9:48 and arrange it so that it plays with our animations nicely." And now it's added the background music. And I'm going to play that for you. >> [music]

10:15 [music] [music] [music] >> Yeah, I think that makes a lot of difference. You can see how adding music

10:41 just makes the video comes to life and it even adjusted the music to match the animations. So, it's fading in and out at the right time. All right, so let's take it up one level. And now in this

10:52 second video, I want to create a more advanced product launch video. And this is a video by Claude. And as you can see, this is a very well produced product launch video. We've got the top

11:01 down shot with the actor walking in. And we have some motion graphics overlaid on the video itself. I want to create something like this where we combine motion graphics and overlay that onto a

11:12 cinematic video like this. So what I did was again just give it a link to this video so you can study it and also a link to my app here. And because it's already created that first motion

11:21 graphic video, it's able to take some of those things and add it to this second video here. And after giving Claude all those things, this is the video that I get. And I want to show you this video

11:30 first and then we're going to walk backwards and show you every step that I took to get to this point. So I'm going to play this video. Today we're launching Cubot. It lives

11:47 inside Quicksend [music] and it chases your invoices for you. Chasing invoices is the worst part of running your own business. Just ask Cubot who's paid, who's opened, and

11:58 who's [music] late. When someone's late, it writes the follow-up. You just tap send, and the second they pay, you'll [music] know so you can get back to the actual work.

12:19 So I think that's pretty good. It took that idea from that claude video and it adapted for our own product with our model and also overlaid some of the motion graphics here, the chat bubbles

12:29 and the story line here is also adapted for our own product. Now just to give you a highle understanding of how this video is built, we've got basically two layers here, right? The first layer on

12:39 the left is the cinematic video itself. Obviously, we don't have a real human, so we're going to use an AI actor for this, and we are using Claude's video as a reference for what the scenes are

12:48 going to look like, but of course, we don't want to copy them exactly. So, we come up with our own actor and the office design. Now, the second layer is the motion graphics itself. So, we built

12:57 this in part one. So, we can use some of these elements here and layer it on top of the cinematic video here. And when we combine the two, we have the full video. So, that is the overview of how the

13:08 video is done. Now I'm going to walk you through how I'm doing the first layer here which is the cinematic video. Now to create the cinematic video you've got four simple steps to go through and I'm

13:17 going to tell you later why it's important to do it in this specific order. So number one is to come up with the AI actor and also the image of the location which is the office here. And

13:26 once I'm happy with both of these images here, step number two will be to combine both of these images and basically put the actor in the office. And again once I'm happy with the result I go to step

13:36 number three which is to create all the other scenes and only when I'm happy with the result in this step I move on to the final step which is to animate all the scenes. Now if you want the best

13:46 quality it is crucial that you follow these steps in this particular order cuz I actually tried one-shotting everything and what I get is basically AI slop like this. You can immediately tell that this

13:57 looks like AI. The color and the vibe is totally off. So, yes, I know this will probably take a bit more time and effort, but the results will be 100 times better. Now, I'm going to break

14:06 down all the steps in more detail and show you the exact prompts that I use, but the one thing you need to generate images and videos is something like Hicksfield. And Hicksfield is kindly

14:14 sponsoring this video, but they just launched their API recently. And this is actually a great option for people who don't do a lot of generations. So, you can just top up have a balance and then

14:23 it will pull from that balance whenever you need to generate any images or any videos. And I think the pricing is also pretty competitive. Like if you compare this with its competitor file.AI, it is

14:33 actually very similar. But so if you're using any of these AI models like Cash 2.5 or even Clling, they are 15 to 50% off their listed pricing here. So pricing wise, they're very competitive.

14:44 And they even have their own image model called Hicksfield Soul 2.0 here. And not a lot of people know this because most people they default to GBT 2.5 or Nano Banana Pro. But Soul 2.0 I know here, as

14:54 you can see from the quality that it's able to create, it's pretty mind-blowing. So, if you're doing any type of portraits of real people, I would actually choose this over

15:03 something like Nano Banana or GPT2.5 cuz not only is able to create something this realistic, but it's also much cheaper to produce cuz it takes less credits. In fact, this one was actually

15:12 free cuz right now I can create basically 14 15,000 of this type of images for free with Soul 2.0. So yeah, don't always default to the latest model cuz it's not always the best for certain

15:23 things. Now for this particular video, I actually created a guide here on how I created this video step by step. So if you guys want to download this, uh this is actually a step by step in how I

15:34 picked the characters, how I designed the office and how I created all the scenes and put the video together. Now you can go ahead and grab this down below. Now the first thing that I did

15:43 was to find a character that I want to use in my video. And for that, we want to create something called a character sheet. So, I found this picture on Pinterest that I'm going to use for of

15:53 reference. So, I simply save this in my folder and also have this prompt ready to create our character sheet. And what a character sheet is basically a portrait of this character here from

16:02 various angles. So, when you actually get to the point where you create the video, it's going to be more accurate in terms of the facial features here. So, this is a very important step that you

16:11 need to take, right? So, this is the prompt that I have. So I can go to claude now and I can just paste in this whole prompt and also attach the photo of the actor that I want to reference.

16:21 Now if you notice this is kind of a long prompt, right? And I didn't actually create this prompt. So what I recommend doing is obviously to have a detailed prompt but also to ask claw to write the

16:30 prompt for us. So when it goes to Hicksfield to generate this character sheet is going to be detailed and it's going to generate exactly the thing that you have in mind. Because if you look

16:40 here, I even have the type of camera that I want to use for this type of shot and something that I wouldn't be able to write. And for me, I used Hickfield MCP because I already have a subscription

16:48 there. So, I'm going to make sure that the connector is on here for Hicksfield. And then I can say something like, "Hey, use Cream Pro 5 via Higfield MCP." So, I find that using Cream Pro here is also a

17:00 lot cheaper than using something like Nano Banana or GPT images. So, what I learned is that there's no best AI model. It really depends on your use case. So if you're creating a character

17:11 like this, Cream 5 is really good at this. The reason why I'm not using Soul 2.0 is because it's not able to take in any reference photo. Otherwise, I would have used that one. But Cedream Pro 5

17:22 allows you to upload a reference photo. So if you don't care about having a reference photo here, you wanted to come up with a random character, I think Soul 2.0 could be the right option. So with

17:31 this prompt here, I'm going to hit go. So in just a couple minutes, the character sheet is ready. And as you can see here, it's use Cream 5 Pro and I think the quality is really good. So my

17:40 next step would be to create the office. Now I also have a prompt ready here. Again I asked lot to come up with this prop as well and this is actually the result which is pretty good in my

17:50 opinion. Now the AI model that I use for this is actually called Soul Cinema and this is also Hicksfield's own image model and this is actually good for creating location style images and again

18:02 it's a lot cheaper in terms of credits as well. And the thing is, if you go directly on the HVE platform itself and you just paste in your prompt right here, you actually get 15,000 free

18:12 generation just like as you would with Soul 2.0. So if you see this in your account, you might want to just take advantage of this cuz your images are going to be free. Otherwise, if you go

18:20 through the MCP, it's going to cost you about 12 cents or 0.12 credits per image. Now, once that's done, I'm going to combine those two images, the character sheet and the office as well.

18:31 And I've got this prompt here that I used. And once I'm happy with this, I'm ready to create all the other scenes. And again, here is the prompt that I used to create the other scenes. So you

18:40 can just simply copy this and then it will create the other scenes for you. And I didn't even come up with the angles here. This is what Claude was able to capture from this video. It

18:51 understood that there's a top- down shot here and there's a model coming in and here is the zoomedin version of it. So it understood all these angles here and created basically the same one with our

19:01 model in it. Now, so far I have not created any videos yet cuz so far we're just making sure that all the shots, all the angles are correct. Now, once you're happy with all these angles here, you

19:12 can then ask claw to animate every shot, right? And for this, of course, I use Cance 2.5. It is an expensive model, but it is still the best one as long as your images are really good. So, here is the

19:23 prompt that you can use. Now, one trick that I use for Cance, if you are creating any type of video, I do recommend creating the video in 480p draft first because you will take a lot

19:33 less credits because most of the time you could run into errors, right? You don't want to waste credits here. But only after you're satisfied with a result, then you're going to ask it to

19:42 recreate the whole thing in 1080p. And here is the final video again. Today we're launching Cubot. [music] It lives inside Quicksend and it chases your invoices for you. Chasing invoices

20:01 is the worst part of running your own business. Just ask Cubot [music] who's paid, who's opened, and who's late. When someone's late, it writes the follow-up. You just tap send, and the second they

20:13 pay, you'll know so you can get back to the actual work. >> [music] [music] >> All right. So, I think I really like

20:30 this scene in particular cuz I think the angle just looks good. But things like this one, I think this scene can be improved. Um, I personally feel that this still looks or sound like AI.

20:41 Unfortunately, that's just the limitation of AI at the moment. I mean, I'm pretty sure it would get better, but whenever you see AI talking, it just looks like AI. But the other scenes like

20:50 this when there's just narration or a voice over, it looks pretty good, right? And I'm very happy with how the motion graphics appeared, especially on the video here. It looks very natural and

21:00 they all align well with the script and the story line. Yeah. So, let me know what you guys think of this. Let me know in the comments if you guys will be creating something like this. And if you

21:08 do, don't forget to tag me. And that is all for me today. If you learned something, I would appreciate a like and subscribe. I appreciate you guys. I will see you in the next video.