Claude Opus 5 + Higgsfield AI = EASY AI Motion Graphics

0:00Making motion graphics the normal way means hours in After Effects setting up layers, key frames, and endless little tweaks. But with Claude Opus 5, you can make results like these straight out of the chat. There's one technique you need first though, because every video model out there messes up text the second it tries to animate a word. So in this video, I'll show you exactly how to get around that. And by the end, you'll see how easy it is to make one without opening [music] a design tool. The first thing to do is connect Higgsfield to Claude. So go ahead and open Higgsfield through the link down in the description

0:27and head to the MCP page. Keep the Claude tab selected in the MCP toggle on. Then copy the connector URL set under them. Head over to Claude, go into customize, then connectors. Add a new one, name it Higgsfield, paste the URL in, then connect and sign in. Every chat after this already has the models ready, and it runs on your own Higgsfield credits. So just make sure your plan covers what you're using. Make sure you're on Opus 5 as well, because that's

0:49the model doing all the thinking here. So, I'll start by explaining to Claude that we'll be working on motion graphics across four levels of complexity, ranging from the most basic to the most advanced. It returns the complete workflow follow to handle text effectively. We start by generating stills using an image model, then feed those stills into the video generation model to lock in the visuals based on

1:08the screens we've already approved. Then, I'll ask for some ideas to get started. Nothing fancy, just something simple to warm up with. And it comes back with the simplest yet functional motion graphic we can create. A stat card with one big number and a line underneath it. And it brings four directions we can follow, with text, colors, and mood already sorted for each one. I'll take the 2.4 million one and the 48 times one. And then it walks me through the settings. And that bit's way simpler than it looks. It starts with the image model, and there's GPT [music] Image 2, both Nano Bananas, Seedream, Soul, and Cinema Studio Image. I'll take

1:38GPT Image 2, and it's got resolution at 1K, 2K, or 4K, and quality at low, medium, or high. Quality defaults to low, and this is all text that people are going to be reading off the screen. So I'll go 4K and high and get it looking really clean. Then it moves on to the video model, and [music] it's SeaDance, Cling, Vio, Gemini Omni Flash, or Cinema Studio. They can all do this, so it's really just about the kind of movement you want. I'll take Sea Dance,

2:00[music] and it comes in three versions. 2.5 runs from 4 to 30 seconds at up to 1080p. 2.0 stops at 15 seconds, but pushes to 4K. And Mini is the cheap one that caps at 720p. Nothing here needs 30 seconds or 4K, so any of the three would do the job, and I'll take 2.5 at 1080p because it's the newest. Then it asks what shape I want, and I'll go 16 by 9, though 9 by 16, square, [music] and 4 by 3 are all there, too. And it has to match on both, because if you build the picture one shape and animate it another, it comes back cropped or with

2:29black bars down the side. [music] And it sets that as its own setting, rather than writing it into the prompt, because models get confused when the prompt describes one shape and the setting says another. And that's everything locked now, apart from the length and whether there's sound, and those two change with each one. Now, every prompt from here opens the same way, and this is the bit that makes the whole thing hold together. It starts by listing everything that has to stay exactly as it is. So, the color, the lighting, the grain, and only then does it name the one thing that's changing. [music] That's what stops the background

2:58shifting between two pictures, because a shift there turns into a flicker the second it moves. So, the first card starts on an empty background, just a flat warm paper color with nothing on it at all. And every prompt and template I use today is in one free kit linked below. So, you can build these along with me instead of pausing to copy anything down. Everything after this takes its color and its lighting from that one. So, it's worth getting exact before anything goes on top. Then that same background goes back in as a reference, and the words get added onto

3:24it. And this is the finished card. Honestly, it looks like something a really top-tier designer made. The text is unbelievably sharp, and that's the point, because it was made as a picture, rather than by a video model. So, now I'll animate between those two pictures, and it runs 5 seconds with no sound.

3:44And every bit [music] of it fades into place, and then stays exactly where it lands. Then I'll do the 48 times card, and it runs exactly the [music] same way, just with a slight twist. So, I'll start it on an empty mint background. Then the words go on top of it off that same background as the reference with everything else held the same. And where the first card faded in, this one springs with a slight overshoot before it settles. It's 5 seconds again with no

4:04sound on it, either.

4:11And that feels completely different off exactly the same process. So, [music] that's two finished motion graphics and neither of them needed more than two pictures and one sentence. But in both of them, everything just appears where it's meant to be. And that stops working the moment something has to move across the screen. And the fix to that is a third picture in the middle. So, this one is a logo animation and I'll get it to invent the brand as well. And I'll ask for something editorial because those logos are usually designed really well. It gives me Pharaoh, a company that makes fonts and there's a proper

4:36idea in the logo. The bottom of the R is cut off flat at an angle and the cut is filled in a dark red like the stained edge of a book. And I like [music] that it picks the look before it generates anything because every picture after this copies it. So, I'll need three pictures this time starting with empty black and that needs to be completely flat because any grain or shadow in there starts moving under the letters the second it animates. Then I'll get the R on its own sitting dead center at about half the height of the frame. And that's exactly the shape I wanted. So,

5:01the full word gets built straight off that picture [music] and the lettering carries across instead of me describing it all again. The cut leg tucking under the O is what gets me and that's the sort of detail you'd normally be paying someone for. So, now I'll set up the animation and the black goes in as the start with the full word as the end. And the R goes in between them as a rough

5:18target to head towards on the way.

5:25And honestly, that came out great. The R fades in on its own first, then it shrinks down into place and the rest of the letters fill in around it all in 6 seconds. But that's still just one thing moving on a black screen and most of what you'd actually need has way more going on than that. So, this time I'm

5:38chaining the pictures together instead. This one's a settings panel that looks like something off an iPhone. You flip one master toggle, everything underneath turns on with it, then you hit save and it confirms. By the way, the Higgs field link is down in the description if you want to follow along and have your own

5:52first motion graphic by the end of this. And I'll do these one at a time so I can check each one before it moves on. So the first is the panel just sitting there, everything switched off and save grayed out. Then I turn the master toggle on and save wakes up blue. Then the other three rows follow it on and the cursor starts moving down towards save. And the last one is straight after the click where the save button collapses into a green tick. And this turned out really impressive because you genuinely can't tell it's four different pictures. And there's one thing I always add when it's an interface. So the

6:19cursor has to land on the toggle at the right edge, not on the text because that's the bit that looks wrong if it's off. And it throws in one I wouldn't have thought of. So the toggle circles stay perfectly round or they stretch as they slide. Then all four go in as one eight second take with the third in the middle so the cursor travels down to save. And this is the first one with sound and it's interface noise only. So there's a whoosh on the cursor, a click as the toggles flip, a deeper click on

6:42the press, and a pop on the check mark. And sound does way more for a motion graphic than people give it credit for. A click landing at the exact moment the toggle flips is what sells it as an actual interface instead of a just a picture that moves.

7:01And honestly, that came out really clean. It runs itself from start to finish and every click lands exactly where it should. But every single one of these has the camera bolted to the floor and that's not what a proper ad looks like. So the last one I'm going to make is a product ad and this time the camera is the thing doing the work. I'll ask for hyper motion, fast camera, 3D render, and say it should pick the product too, something with materials worth shooting. It goes with a patisserie candy bar called Halo. And it picks that because it gives four materials that all behave differently on

7:28camera. You've got matte paper, foil, stamped gold, glossy chocolate, and molten caramel. And that's a better reason than I'd have come up with because when the camera's moving that fast, the only thing you really see is light sliding across the surfaces. And the product sheet came out gorgeous. That caramel pull between the two halves is the shot, and it's what decides how

7:45the bar looks for everything after it. Then I'll have it lay out the storyboard, and I'll put in one rule of my own, to not add numbers in the storyboard, as they usually end up in the video and mess everything up. And the board came out exactly right, and every panel looks like it came off the same shoot. Then I'll have it animate the whole thing into 18 seconds of video. Both the sheet and the board go in as references, and neither one is the start frame, because if you hand it a grid as the start frame, it animates the grid, and you end up with six panels

8:10sliding around instead of one video. So, the sheet is what tells it what the bar looks like, and the board sets the order the shots come in. Everything else is down to the camera, and it only stops twice in the whole thing. And that's on purpose, because those two stops are the only moments the word Halo can sit still, and text falls apart the second a

8:25fast camera drags it around.

8:45And this last one is genuinely crazy. It runs the full 18 seconds without a single cut in it, and honestly, it looks really expensive. I love that the caramel actually behaves like caramel when the camera spirals through it, because that's the part I was sure it'd get wrong. And once you've got this set up, you can make title cards, logo animations, UI demos for your own product, explainer graphics, anything you'd normally be opening After Effects for. And the whole thing comes down to one idea. So, make the pictures first, and let it animate between them. Go ahead and set Higgs Field up through the link down in the description, and try it

9:14today. Thank you for watching, and I'll see you in the next one.