Motionly - AI-native motion graphics editor

byโ€ข
Motionly is an open-source motion graphics editor built for the AI era. Instead of generating videos directly, AI generates editable .motion files that can be refined visually or in code. It combines a visual editor with a structured animation format, making motion graphics versionable, reusable, and easier for developers, designers, and AI agents to collaborate on.

Add a comment

Replies

Best
Hey Product Hunt! ๐Ÿ‘‹ We're excited to relaunch Motionly! Over the past week, we've shipped a lot of improvements across the editor, rendering engine, and overall experience. We've also significantly improved our AI capabilities, making it easier to go from an idea to an editable animation while keeping you in control. This is still just the beginning. We have big plans ahead, including AI-powered animation generation from voice, websites, educational content, and much more. As always, we'd love to hear what you think. Your feedback helps shape what we build next. Thanks for checking out Motionly! ๐Ÿš€

Motion graphics is one of the harder creative domains to make genuinely AI-native โ€” most tools just bolt AI-generated assets onto a traditional timeline editor. Curious what "AI-native" means concretely here: is the AI driving keyframe/timing decisions directly, or is it more focused on asset generation that you still manually place and animate? And since it's open source, curious what's driving that choice โ€” community/extensibility, or more about trust given how much of a black box a lot of AI creative tools are.

ย Great question! For us, AI-native isn't about using AI to generate a few assets and then dropping them onto a traditional timeline.

The core idea is that AI generates the entire motion projectโ€”a .motion file that describes the scene, timeline, asset placement, animations, easing, transitions, and effects. Think of it like how Photoshop has a PSD or Figma has a .fig file. Instead of outputting a rendered video, AI outputs an editable project that you can open, inspect, and refine visually or through code.

We're less focused on AI creating the assets themselves. In most real workflows, people already have their own logos, branding, illustrations, or media they'd like to animate. Of course, if you're using an AI provider that can generate assets (like SVGs), Motionly can incorporate those too. A workflow we imagine is asking AI to generate a specific animated SVG, then using that as part of a larger .motion project.

As for open source, it's both about extensibility and trust. We want developers to build custom renderers, plugins, AI integrations, and new workflows on top of Motionly. And because the project format is open, you're never locked into a black boxโ€”you own the project file and can edit or generate it however you like.

ย That's a genuinely different model โ€” outputting an editable project file rather than a rendered result means the AI's decisions stay inspectable and revisable, not just a black box you accept or regenerate from scratch. The PSD/Figma comparison makes it click immediately. And the "you own the project file, never locked in" trust angle resonates with me specifically โ€” we've built ClipStudios around similar thinking on the video side (content provenance, not black-box outputs). Nice architecture, good luck with the launch!

ย Thank you so much!

the pitch that's usually missing from AI motion tools is exactly what you're doing here, keeping it as an editable timeline instead of just a rendered clip you have to regenerate from scratch every time you want a small change. how deep does the editability go though, can you swap an easing curve on one layer without the AI touching anything else around it?

ย Yes you can! You can change the keyframe, animation duration.. Without affecting the styling or its positions. Update color, change the text rotation or scale without messing up the current animation. But right now Weโ€™re working on the keyframe and making better, with graph visuals and more keyframe options and also custom keyframes. Hopefully We can rapidly make these all better than what they are soon!

The ".motion file as an intermediate format" idea is the real differentiator most AI motion tools skip straight to rendered video, so any edit means regenerating from scratch. Structuring it as an editable, versionable format is the part that actually makes it collaborative rather than one shot. Curious how conflict resolution works in practice, though if a designer tweaks something in the visual editor while an AI agent is also generating changes to the same .motion file, is there a merge/diff model like Git, or does one side just overwrite the other? That's usually where "developers, designers, and AI agents collaborating on one file" gets messy in real tools.
Thanks for the question! We havenโ€™t explored this workflow deeply yet since I havenโ€™t personally edited a .motion file while an AI agent is modifying the same file at the same time. One direction we might explore is a review-based workflow where AI changes are shown before being applied, but weโ€™re still figuring out the best approach. Would love to hear any ideas or suggestions on how you think this should work!

The editable-.motion-file approach is the right call โ€” AI generating a structured, diffable format instead of pixels is what makes motion actually maintainable at scale. Two genuine questions: (1) how does .motion relate to existing runtime formats like Lottie โ€” can a project export to something that renders on web/mobile via an existing runtime, or is playback tied to your own player? (2) once AI generates a project and someone hand-tweaks timing/layout on the canvas, does the underlying file stay clean enough that versioning/diffs stay meaningful, or does regeneration tend to churn the whole file?

ย Thanks for the thoughtful questions! Also, we love LottieFiles. ๐Ÿ˜Š

  1. Right now, playback is handled by Motionly's own renderer. The idea with .motion is to provide a structured, editable source format for motion graphics that both AI and humans can work with. Export targets and runtime integrations are something we're exploring as the ecosystem grows.

  2. This is actually a really interesting challenge we're working through. When you make visual edits, they update the underlying .motion file, and currently users can choose whether to overwrite the same file or save a new version. We're exploring better approaches for diffs and versioning, especially around cases where a human is editing visually while an AI agent is also making changes. There are a lot of interesting questions around conflict resolution and keeping changes meaningful.

Also, we're exploring ways to work with existing animation ecosystems like LottieFiles, for example using Lottie animations as reusable components inside larger .motion compositions.

Thanks again for the thoughtful questions!

Giving you something you can still edit instead of a finished video is exactly right. Nobody wants to start from scratch just to change one small thing. Great thinking here. Congrats on shipping!

ย Yep we are trying to make it easier to refine. Thank you!

What could you build with GPT-5.6 that wasnโ€™t practical before?
GPT-5.6 makes it possible to build Motionly into a true AI animation partner. Previously, generating motion graphics required manually designing scenes, timing, layouts, and animations. With GPT-5.6, AI can better understand creative intent, plan complex animation structures, and generate editable .motion projects instead of just producing static outputs. This enables workflows where users can describe an idea, provide a reference, or share content, and GPT-5.6 can help create a complete animation with better composition, timing, and storytelling โ€” while keeping everything editable and customizable.

Hello everyone!

Motionly is open source! ๐Ÿš€

If youโ€™re interested in how it works or want to contribute, check out the GitHub repository:

Weโ€™d love your feedback and ideas as we continue building it. โค๏ธ

๐Ÿš€ We just updated the Motionly website!

Find more information about the project, along with links to the demo, documentation, and GitHub repository here:



Would love to hear your thoughts and feedback!

Thank you