Artisan Benchmark
Towards Physically Grounded Articulated 3D Assets: A Multi-Agent Pipeline and Benchmark
A benchmark evaluating spatial fidelity, physical plausibility, downstream task performance, and overall quality of articulated 3D assets using 100 reference assets, 50 stability prompts, and 50 ABO objects.
How to run
$ npm i -g lettokenburn-cli$ ltb bench info artisan-benchmark
External harness. No Harbor adapter has been generated for this benchmark yet. `ltb bench info` prints the known links.
Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs
Scores
4 sourced entriesEvery number links to where it was published. Low-confidence rows are muted and need per-cell verification.
Reproduced by the community
All runsAbout
The benchmark comprises 100 reference assets from ArtVIP and LightWheel, 50 stability prompts, and 50 ABO objects with reference masses. It evaluates generated assets across four key aspects: spatial fidelity, physical plausibility, downstream task performance, and overall quality.
Scoring involves multiple metrics for each aspect, including collision rates, stability rates, and human judgments. The Artisan pipeline achieves the lowest collision rates and highest stability rates among evaluated methods.