{"paper":{"title":"TOPOS: High-Fidelity and Efficient Industry-Grade 3D Head Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"TOPOS generates single-image 3D heads locked to one fixed studio topology so every output shares identical vertices for rigging and animation.","cross_cats":["cs.GR"],"primary_cat":"cs.CV","authors_text":"Bojun Xiong, Bowen Cai, Huan Fu, Jing Li, Junchen Deng, Jun Liang, Xinghui Peng, Yunmu Wang, Zoubin Bi","submitted_at":"2026-05-14T09:02:32Z","abstract_excerpt":"High-fidelity 3D head generation plays a crucial role in the film, animation and video game industries. In industrial pipelines, studios typically enforce a fixed reference topology across all head assets, as such a clean and uniform topology is a prerequisite for production-level rigging, skinning and animation. In this paper, we present TOPOS, a framework tailored for single image conditioned 3D head generation that jointly recovers geometry and appearance under such an industry-standard topology. In contrast to general 3D generative models which produce triangle meshes with inconsistent top"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"TOPOS achieves state-of-the-art performance on 3D head generation, surpassing both classical face reconstruction methods and general 3D object generative models, highlighting its effectiveness for digital human creation.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the Perceiver Resampler can reliably convert point clouds sampled from head meshes of diverse topologies into the fixed reference topology while preserving geometric fidelity and semantic correspondence needed for downstream rigging and animation.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"TOPOS creates high-fidelity 3D heads with fixed industry topology from single images via a specialized VAE with Perceiver Resampler and a rectified flow transformer.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"TOPOS generates single-image 3D heads locked to one fixed studio topology so every output shares identical vertices for rigging and animation.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"8ff0197d3dd6ccb8fd64ac6bebe4638957d24430c48fb6fe6f1929c0845f1fd4"},"source":{"id":"2605.14594","kind":"arxiv","version":1},"verdict":{"id":"31c7012e-e966-4af9-b3d3-7e61613d8bde","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T01:28:49.983886Z","strongest_claim":"TOPOS achieves state-of-the-art performance on 3D head generation, surpassing both classical face reconstruction methods and general 3D object generative models, highlighting its effectiveness for digital human creation.","one_line_summary":"TOPOS creates high-fidelity 3D heads with fixed industry topology from single images via a specialized VAE with Perceiver Resampler and a rectified flow transformer.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the Perceiver Resampler can reliably convert point clouds sampled from head meshes of diverse topologies into the fixed reference topology while preserving geometric fidelity and semantic correspondence needed for downstream rigging and animation.","pith_extraction_headline":"TOPOS generates single-image 3D heads locked to one fixed studio topology so every output shares identical vertices for rigging and animation."},"references":{"count":145,"sample":[{"doi":"","year":null,"title":"Advances in neural information processing systems , volume=","work_id":"48172a5a-0dfc-45cf-9fdc-988f99c16450","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2010,"title":"Denoising Diffusion Implicit Models","work_id":"8fa2128b-d18c-405c-ac92-0e669cf89ac0","ref_index":2,"cited_arxiv_id":"2010.02502","is_internal_anchor":true},{"doi":"","year":2021,"title":"International conference on machine learning , pages=","work_id":"727406c9-82e7-46d8-98d5-4cc67c806271","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Flow Matching for Generative Modeling","work_id":"6edb71c4-5d64-40af-a394-9757ea051a36","ref_index":4,"cited_arxiv_id":"2210.02747","is_internal_anchor":true},{"doi":"","year":null,"title":"Flow Matching Guide and Code","work_id":"2be93143-ab6f-48d6-96d5-3e85d7246f07","ref_index":5,"cited_arxiv_id":"2412.06264","is_internal_anchor":true}],"resolved_work":145,"snapshot_sha256":"2fd06bd2dc331a5bd247f677c0e3037ca8424e1a147ec3540e1b291e49f8cae7","internal_anchors":14},"formal_canon":{"evidence_count":1,"snapshot_sha256":"f4a57a7d9208c486d0d9eeaa6328c91f68df03103c3b6e54fccbba57c5e278ca"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}