EditHero: A Benchmark for Long-Horizon Part-Level 3D Editing and Vibe Modeling
EditHero benchmarks long-horizon part-level 3D editing, finding LLM/VLM agents follow instructions and preserve untouched regions better but edit far slower.
EditHero is presented as the first benchmark for long-horizon, part-level 3D editing, using natural-language instructions and target images for both geometry and texture. A deterministic assembly engine produces the exact target after every edit, and all sequences are hand-reviewed. Non-agentic top-down methods often miss the requested change and disturb regions that should stay fixed, while LLM/VLM agents follow instructions more closely and preserve unedited parts but take minutes per edit. The engine and edit sequences will be publicly released.
- First benchmark for long-horizon, part-level 3D editing with natural-language instructions
- Deterministic assembly engine produces exact targets; every sequence hand-reviewed
- Non-agentic methods miss requested changes and disturb regions meant to stay fixed
- LLM/VLM agents follow instructions better but each edit takes minutes
- Organizations
- EditHero
From the original article
1 min read at huggingface.co
3D editing methods are usually tested on a single edit, yet an asset is built through a long sequence of revisions, each of which must implement the requested change while leaving everything else unchanged. We introduce EditHero, to our knowledge the first benchmark for long-horizon, part-level 3D editing, with natural-language instructions and target images for both geometry and texture.
By Ruihan Yu, Yu-Ju Tsai, Muyao Niu · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.