{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LG6XRG7DALQ6CQZT6JR2VC4X5K","short_pith_number":"pith:LG6XRG7D","schema_version":"1.0","canonical_sha256":"59bd789be302e1e14333f263aa8b97eab554a7fea072cc8d62dc0236713469b6","source":{"kind":"arxiv","id":"2502.06608","version":3},"attestation_state":"computed","paper":{"title":"TripoSG: High-Fidelity 3D Shape Synthesis using Large-Scale Rectified Flow Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"TripoSG generates high-fidelity 3D meshes from images via a large-scale rectified flow transformer trained on two million samples.","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Dehu Wang, Ding Liang, Wanli Ouyang, Xingchao Liu, Yangguang Li, Yan-Pei Cao, Yuan-Chen Guo, Yuan Liang, Zexiang Liu, Zhipeng Yu, Zi-Xin Zou","submitted_at":"2025-02-10T16:07:54Z","abstract_excerpt":"Recent advancements in diffusion techniques have propelled image and video generation to unprecedented levels of quality, significantly accelerating the deployment and application of generative AI. However, 3D shape generation technology has so far lagged behind, constrained by limitations in 3D data scale, complexity of 3D data processing, and insufficient exploration of advanced techniques in the 3D domain. Current approaches to 3D shape generation face substantial challenges in terms of output quality, generalization capability, and alignment with input conditions. We present TripoSG, a new"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2502.06608","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-02-10T16:07:54Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"5af621890d5e288d22d2defea32177771807c7d11f9e35f1eec1d4ab317e14d3","abstract_canon_sha256":"b5b4e235d2ba8825bfb771cabdf662fbd289b46f55f20e3714d6c1382c6898c1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:46.470761Z","signature_b64":"I2SlGvCN4JWwrXHHieDGY6Bb0Uf880sgTdrg8p6dnhnMwM7Y0fypPdmHSfRDSHhxa6mZQten9VD66cz09tOYBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"59bd789be302e1e14333f263aa8b97eab554a7fea072cc8d62dc0236713469b6","last_reissued_at":"2026-05-17T23:38:46.470101Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:46.470101Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TripoSG: High-Fidelity 3D Shape Synthesis using Large-Scale Rectified Flow Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"TripoSG generates high-fidelity 3D meshes from images via a large-scale rectified flow transformer trained on two million samples.","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Dehu Wang, Ding Liang, Wanli Ouyang, Xingchao Liu, Yangguang Li, Yan-Pei Cao, Yuan-Chen Guo, Yuan Liang, Zexiang Liu, Zhipeng Yu, Zi-Xin Zou","submitted_at":"2025-02-10T16:07:54Z","abstract_excerpt":"Recent advancements in diffusion techniques have propelled image and video generation to unprecedented levels of quality, significantly accelerating the deployment and application of generative AI. However, 3D shape generation technology has so far lagged behind, constrained by limitations in 3D data scale, complexity of 3D data processing, and insufficient exploration of advanced techniques in the 3D domain. Current approaches to 3D shape generation face substantial challenges in terms of output quality, generalization capability, and alignment with input conditions. We present TripoSG, a new"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"TripoSG achieves state-of-the-art performance in 3D shape generation. The resulting 3D shapes exhibit enhanced detail due to high-resolution capabilities and demonstrate exceptional fidelity to input images with strong generalization across diverse image styles and contents.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that the custom data processing pipeline produces sufficiently high-quality, diverse, and representative 3D samples at the claimed scale of 2 million without introducing biases or artifacts that would limit generalization or fidelity in real-world use.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"TripoSG generates high-fidelity 3D meshes from input images via a large-scale rectified flow transformer and hybrid-trained 3D VAE on a custom 2-million-sample dataset, claiming state-of-the-art fidelity and generalization.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"TripoSG generates high-fidelity 3D meshes from images via a large-scale rectified flow transformer trained on two million samples.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"4a2dc2f8178abfde708fa28c77b0a0caaeecac4726b8b33917e1bf6e417d6851"},"source":{"id":"2502.06608","kind":"arxiv","version":3},"verdict":{"id":"e183c5ad-ca0f-45bb-922e-077c01f8c92a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T21:45:37.289109Z","strongest_claim":"TripoSG achieves state-of-the-art performance in 3D shape generation. The resulting 3D shapes exhibit enhanced detail due to high-resolution capabilities and demonstrate exceptional fidelity to input images with strong generalization across diverse image styles and contents.","one_line_summary":"TripoSG generates high-fidelity 3D meshes from input images via a large-scale rectified flow transformer and hybrid-trained 3D VAE on a custom 2-million-sample dataset, claiming state-of-the-art fidelity and generalization.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that the custom data processing pipeline produces sufficiently high-quality, diverse, and representative 3D samples at the claimed scale of 2 million without introducing biases or artifacts that would limit generalization or fidelity in real-world use.","pith_extraction_headline":"TripoSG generates high-fidelity 3D meshes from images via a large-scale rectified flow transformer trained on two million samples."},"references":{"count":187,"sample":[{"doi":"","year":2021,"title":"Frozen in time: A joint video and image encoder for end-to-end retrieval","work_id":"62783d40-abfc-4447-87fd-a95adb06ec97","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"All are worth words: A vit backbone for diffusion models","work_id":"4b93ec35-06cf-40f1-8cbe-c6b896c36f19","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"blackforestlabs. Flux. https://github.com/black-forest-labs/flux, 2024","work_id":"88376550-0f76-486a-a242-24b59f42de82","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Video generation models as world simulators","work_id":"2f310660-a305-4eef-af77-31702966d6b4","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"R., Nagano, K., Chan, M","work_id":"8ac2b3a2-4416-4cb6-8c5e-ca196b57bc17","ref_index":6,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":187,"snapshot_sha256":"5015d930db6a9f5002856bd66b62fee3d2b44a675b06dad5088d09f3c343ebf0","internal_anchors":13},"formal_canon":{"evidence_count":2,"snapshot_sha256":"c3a962feed591baef6f88e53edd07370afbf7cab731e9d818d1906ff9eb9788b"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2502.06608","created_at":"2026-05-17T23:38:46.470221+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.06608v3","created_at":"2026-05-17T23:38:46.470221+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.06608","created_at":"2026-05-17T23:38:46.470221+00:00"},{"alias_kind":"pith_short_12","alias_value":"LG6XRG7DALQ6","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"LG6XRG7DALQ6CQZT","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"LG6XRG7D","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":30,"internal_anchor_count":30,"sample":[{"citing_arxiv_id":"2605.23381","citing_title":"VDE: Training-Free Accelerating Rectified Flow Model via Velocity Decomposition and Estimation","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2510.17991","citing_title":"Demystifying Transition Matching: When and Why It Can Beat Flow Matching","ref_index":4,"is_internal_anchor":true},{"citing_arxiv_id":"2512.14692","citing_title":"Native and Compact Structured Latents for 3D Generation","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21121","citing_title":"ROAR-3D: Routing Arbitrary Views for High-Fidelity 3D Generation","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20290","citing_title":"TelePhysics: Physics-Grounded Multi-Object Scene Generation from a Single Image with Real-Time Interaction","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2604.28134","citing_title":"MeshReGen: A Unified 3D Geometry Regeneration Framework","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"2605.16355","citing_title":"Generative 3D Gaussians with Learned Density Control","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2605.19786","citing_title":"Fast 4D Mesh Generation by Spatio-Temporal Attention Chains","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17853","citing_title":"CelloCut: Constructive Watertight Remeshing via Tetrahedral Cell Cuts","ref_index":31,"is_internal_anchor":true},{"citing_arxiv_id":"2509.07435","citing_title":"DreamLifting: A Plug-in Module Lifting MV Diffusion Models for 3D Asset Generation","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2506.15442","citing_title":"Hunyuan3D 2.1: From Images to High-Fidelity 3D Assets with Production-Ready PBR Material","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2512.16767","citing_title":"Make-It-Poseable: Feed-forward Latent Posing Model for 3D Characters","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2603.29585","citing_title":"Learn2Fold: Structured Origami Generation with World Model Planning","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2603.11633","citing_title":"MV-SAM3D: Adaptive Multi-View Fusion for Layout-Aware 3D Generation","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2603.16869","citing_title":"SegviGen: Repurposing 3D Generative Model for Part Segmentation","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2605.13862","citing_title":"Seed3D 2.0: Advancing High-Fidelity Simulation-Ready 3D Content Generation","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2604.01479","citing_title":"UniRecGen: Unifying Multi-View 3D Reconstruction and Generation","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2604.28134","citing_title":"MeshReGen: A Unified 3D Geometry Regeneration Framework","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"2604.26917","citing_title":"AnimateAnyMesh++: A Flexible 4D Foundation Model for High-Fidelity Text-Driven Mesh Animation","ref_index":49,"is_internal_anchor":true},{"citing_arxiv_id":"2605.09606","citing_title":"On the Generation and Mitigation of Harmful Geometry in Image-to-3D Models","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10922","citing_title":"Pixal3D: Pixel-Aligned 3D Generation from Images","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2604.23629","citing_title":"From Visual Synthesis to Interactive Worlds: Toward Production-Ready 3D Asset Generation","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2604.23629","citing_title":"From Visual Synthesis to Interactive Worlds: Toward Production-Ready 3D Asset Generation","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05163","citing_title":"PhysForge: Generating Physics-Grounded 3D Assets for Interactive Virtual World","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2605.04527","citing_title":"Velox: Learning Representations of 4D Geometry and Appearance","ref_index":47,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LG6XRG7DALQ6CQZT6JR2VC4X5K","json":"https://pith.science/pith/LG6XRG7DALQ6CQZT6JR2VC4X5K.json","graph_json":"https://pith.science/api/pith-number/LG6XRG7DALQ6CQZT6JR2VC4X5K/graph.json","events_json":"https://pith.science/api/pith-number/LG6XRG7DALQ6CQZT6JR2VC4X5K/events.json","paper":"https://pith.science/paper/LG6XRG7D"},"agent_actions":{"view_html":"https://pith.science/pith/LG6XRG7DALQ6CQZT6JR2VC4X5K","download_json":"https://pith.science/pith/LG6XRG7DALQ6CQZT6JR2VC4X5K.json","view_paper":"https://pith.science/paper/LG6XRG7D","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.06608&json=true","fetch_graph":"https://pith.science/api/pith-number/LG6XRG7DALQ6CQZT6JR2VC4X5K/graph.json","fetch_events":"https://pith.science/api/pith-number/LG6XRG7DALQ6CQZT6JR2VC4X5K/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LG6XRG7DALQ6CQZT6JR2VC4X5K/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LG6XRG7DALQ6CQZT6JR2VC4X5K/action/storage_attestation","attest_author":"https://pith.science/pith/LG6XRG7DALQ6CQZT6JR2VC4X5K/action/author_attestation","sign_citation":"https://pith.science/pith/LG6XRG7DALQ6CQZT6JR2VC4X5K/action/citation_signature","submit_replication":"https://pith.science/pith/LG6XRG7DALQ6CQZT6JR2VC4X5K/action/replication_record"}},"created_at":"2026-05-17T23:38:46.470221+00:00","updated_at":"2026-05-17T23:38:46.470221+00:00"}