figma-ai/motion-graphics-bench
Dataset · https://huggingface.co/datasets/figma-ai/motion-graphics-bench
MotionGraphicsBench: Evaluating Motion Graphics Generation with Programmatic Verification Jiaju Ma1,2,* Ashvini Kumar Jindal2,* Chuhua Wang2 Sumithra Bhakthavatsalam2 1Stanford University 2Figma *Equal contribution MotionGraphicsBench is an expert-curated benchmark for evaluating how well large language models (LLMs) can animate static vector graphics. It consists of 100… See the full description on the dataset page: https://huggingface.co/datasets/figma-ai/motion-graphics-bench.
First seen by Oriel on .
AI summary
MotionGraphicsBench is an expert-curated benchmark using deterministic verification programs to evaluate how well LLMs animate static vector graphics from motion edit prompts.
- The benchmark contains 100 tasks, each providing a static SVG, PNG rendering, motion prompt, source Lottie data, SVG frames, and Python verification program.
- Verification measures geometry, appearance, and motion across ordered SVG frames, producing individual check results and an overall decision without requiring generated animations to match references exactly.
- The release includes 53,180 model outputs: 51,200 attempts evaluating 32 LLMs across 47 settings, plus 1,980 outputs collected for human review.
- Its pipeline can generate GSAP programs through LiteLLM, render them into SVG frames using MoVer, and verify those frames in configurable stages.
- Verification programs execute with file and network access despite subprocess timeouts, so untrusted programs or dataset files require an isolated container or virtual machine without secrets.
#llm evaluation #motion graphics #svg #programmatic verification
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Likes
Total likes reported by the source.
- Hugging Face Datasets: 11, latest observation ; +2 over observed period since
Downloads (30d)
Rolling 30-day downloads. A change in this value is not new downloads during the observed period.
- Hugging Face Datasets: 948, latest observation ; +416 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hugging Face Datasets | figma-ai/motion-graphics-bench | https://huggingface.co/datasets/figma-ai/motion-graphics-bench Read discussion |