This benchmark ranks image-to-3D models on how faithfully they rebuild an object as a textured 3D asset from a single image. We curated 357 input images, each showing an object to reconstruct, held fixed to ensure reproducibility of the results. Every asset is shown to annotators as a turntable video rendered identically for every model, next to the input image. The rankings come from 61,354 pairwise matches, decided by 306,670 responses. Quality is evaluated in two dimensions: Geometry and Faithfulness. All data and judgments are public; see the details for more and the methodology for evaluation and scoring.
We use crowd intelligence to deliver real, diverse human feedback at scale for model evaluation and post-training.