Create maths animations with AI
Turn a mathematical idea into editable Manim scenes, with optional local narration, subtitle timing and a chaptered viewer for focused revisions.
Create maths animations with AI by giving an idea a sequence of visible changes. The manim-skill workflow writes Manim scenes, renders them and joins the result into an MP4 while keeping the editable Python source.
Each scene has its own class, so feedback can target a specific part of the film. Optional Kokoro narration runs locally after its dependencies and model are available; measured clip durations guide subtitles and animation timing. A silent route remains available. The local viewer adds chapters and timestamped feedback, but it cannot establish that a proof is correct or a formula is readable. Review the actual frames, explanation and audio before using the film in a lesson.
What it does
- Animate a formula across separate scenes.
- Explain a concept or algorithm visually.
- Time scenes against optional local narration.
- Review subtitles and chapter boundaries.
- Revise the scene identified in feedback.
Install and use
- Install the skill folder and prepare Manim, the formula-rendering dependencies and FFmpeg. Local narration additionally needs Kokoro and its model; the default voice is English.
- Ask for a three-scene explanation of a changing line slope, stating duration, audience, aspect ratio and whether narration is wanted.
- Inspect the Python source and rendered MP4. Use a timestamp to identify an overlapping label or rushed step, then check the revised scene, joined sequence and subtitle timing again.
FAQ
Source
Made by Yusuke Miyashita. Original skill: manim-skill, published on GitHub under the MIT licence.
View on GitHub