{"paper":{"title":"ViPS: Video-informed Pose Spaces for Auto-Rigged Meshes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Video diffusion priors can be distilled into a universal, controllable pose space for arbitrary auto-rigged meshes that matches models trained on synthetic 4D data and generalizes zero-shot to new species and skeletal topologies.","cross_cats":["cs.GR"],"primary_cat":"cs.CV","authors_text":"Ayush Tewari, Changxi Zheng, Honglin Chen, Karran Pandey, Matheus Gadelha, Niloy J. Mitra, Paul Guerrero, Rundi Wu, Yannick Hold-Geoffroy","submitted_at":"2026-04-19T21:21:11Z","abstract_excerpt":"Kinematic rigs provide a structured interface for articulating 3D meshes but lack any associated pose space, i.e., an explicit representation of the plausible manifold of joint configurations for a given mesh. Without such a pose space, stochastic sampling or manual manipulation of raw rig parameters easily results in semantic and/or geometric violations, such as anatomical hyperextension and non-physical self-intersections. We propose Video-informed Pose Spaces (ViPS), a feedforward framework that discovers the latent distribution of valid articulations for auto-rigged meshes by distilling mo"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"ViPS, trained solely on video priors, matches the performance of state-of-the-art methods trained on synthetic artist-created 4D data in both plausibility and diversity. Most importantly, as a universal model, ViPS demonstrates robust zero-shot generalization to out-of-distribution species and unseen skeletal topologies.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That motion priors encoded in a pretrained 2D video diffusion model can be reliably transferred into a universal distribution over arbitrary rig parameters, and that differentiable geometric validators applied to the skinned mesh are sufficient to enforce asset-specific validity without manual regularizers.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"ViPS distills a compact, controllable distribution of valid joint configurations for any auto-rigged mesh from video diffusion priors, matching 4D-trained methods in plausibility while generalizing zero-shot to unseen species and skeletal topologies.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Video diffusion priors can be distilled into a universal, controllable pose space for arbitrary auto-rigged meshes that matches models trained on synthetic 4D data and generalizes zero-shot to new species and skeletal topologies.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e95e0cfdd772dcbdea8da4d4e2d3867a5275a52bc8a625c0154c769793c71473"},"source":{"id":"2604.17623","kind":"arxiv","version":3},"verdict":{"id":"9f0ff1dc-483b-45b8-817b-a37f1d3dd961","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T05:35:57.017466Z","strongest_claim":"ViPS, trained solely on video priors, matches the performance of state-of-the-art methods trained on synthetic artist-created 4D data in both plausibility and diversity. Most importantly, as a universal model, ViPS demonstrates robust zero-shot generalization to out-of-distribution species and unseen skeletal topologies.","one_line_summary":"ViPS distills a compact, controllable distribution of valid joint configurations for any auto-rigged mesh from video diffusion priors, matching 4D-trained methods in plausibility while generalizing zero-shot to unseen species and skeletal topologies.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That motion priors encoded in a pretrained 2D video diffusion model can be reliably transferred into a universal distribution over arbitrary rig parameters, and that differentiable geometric validators applied to the skinned mesh are sufficient to enforce asset-specific validity without manual regularizers.","pith_extraction_headline":"Video diffusion priors can be distilled into a universal, controllable pose space for arbitrary auto-rigged meshes that matches models trained on synthetic 4D data and generalizes zero-shot to new species and skeletal topologies."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.17623/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}