{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:UYYGECLB6OFQVINVQRDVX5ECYW","short_pith_number":"pith:UYYGECLB","schema_version":"1.0","canonical_sha256":"a630620961f38b0aa1b584475bf482c5ae1cb62454adfbe8b54217204c015bb8","source":{"kind":"arxiv","id":"2508.05254","version":3},"attestation_state":"computed","paper":{"title":"CF3: Compact and Fast 3D Feature Fields","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hyunjoon Lee, Jaesik Park, Joonkyu Min","submitted_at":"2025-08-07T10:45:08Z","abstract_excerpt":"3D Gaussian Splatting (3DGS) has begun incorporating rich information from 2D foundation models. However, most approaches rely on a bottom-up optimization process that treats raw 2D features as ground truth, incurring increased computational costs. We propose a top-down pipeline for constructing compact and fast 3D Gaussian feature fields, namely, CF3. We first perform a fast weighted fusion of multi-view 2D features with pre-trained Gaussians. This approach enables training a per-Gaussian autoencoder directly on the lifted features, instead of training autoencoders in the 2D domain. As a resu"},"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":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2508.05254","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-08-07T10:45:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8281ee3f05ba43731c2efb3c8f9cb6e9f99f439d3f3d24b5df7e8c88fd2a9467","abstract_canon_sha256":"d9fb32c41e173e8df169ad83d345a6199f99c90446d76d5ad4f601ce1f2e2eee"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:02:49.648245Z","signature_b64":"bswzjNIGiUDNd5VdgtqhKlwcU9TVrzzs6DzqIBDbZ+AkhrQbiuRc+wsKhErEnH3l6bq5rxezLNR1G+zWiJqCBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a630620961f38b0aa1b584475bf482c5ae1cb62454adfbe8b54217204c015bb8","last_reissued_at":"2026-07-05T12:02:49.647687Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:02:49.647687Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CF3: Compact and Fast 3D Feature Fields","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hyunjoon Lee, Jaesik Park, Joonkyu Min","submitted_at":"2025-08-07T10:45:08Z","abstract_excerpt":"3D Gaussian Splatting (3DGS) has begun incorporating rich information from 2D foundation models. However, most approaches rely on a bottom-up optimization process that treats raw 2D features as ground truth, incurring increased computational costs. We propose a top-down pipeline for constructing compact and fast 3D Gaussian feature fields, namely, CF3. We first perform a fast weighted fusion of multi-view 2D features with pre-trained Gaussians. This approach enables training a per-Gaussian autoencoder directly on the lifted features, instead of training autoencoders in the 2D domain. As a resu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.05254","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2508.05254/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2508.05254","created_at":"2026-07-05T12:02:49.647762+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.05254v3","created_at":"2026-07-05T12:02:49.647762+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.05254","created_at":"2026-07-05T12:02:49.647762+00:00"},{"alias_kind":"pith_short_12","alias_value":"UYYGECLB6OFQ","created_at":"2026-07-05T12:02:49.647762+00:00"},{"alias_kind":"pith_short_16","alias_value":"UYYGECLB6OFQVINV","created_at":"2026-07-05T12:02:49.647762+00:00"},{"alias_kind":"pith_short_8","alias_value":"UYYGECLB","created_at":"2026-07-05T12:02:49.647762+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2512.04021","citing_title":"C3G: Learning Compact 3D Representations with 2K Gaussians","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22439","citing_title":"NRGS: Neural Regularization for Robust 3D Semantic Gaussian Splatting","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UYYGECLB6OFQVINVQRDVX5ECYW","json":"https://pith.science/pith/UYYGECLB6OFQVINVQRDVX5ECYW.json","graph_json":"https://pith.science/api/pith-number/UYYGECLB6OFQVINVQRDVX5ECYW/graph.json","events_json":"https://pith.science/api/pith-number/UYYGECLB6OFQVINVQRDVX5ECYW/events.json","paper":"https://pith.science/paper/UYYGECLB"},"agent_actions":{"view_html":"https://pith.science/pith/UYYGECLB6OFQVINVQRDVX5ECYW","download_json":"https://pith.science/pith/UYYGECLB6OFQVINVQRDVX5ECYW.json","view_paper":"https://pith.science/paper/UYYGECLB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.05254&json=true","fetch_graph":"https://pith.science/api/pith-number/UYYGECLB6OFQVINVQRDVX5ECYW/graph.json","fetch_events":"https://pith.science/api/pith-number/UYYGECLB6OFQVINVQRDVX5ECYW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UYYGECLB6OFQVINVQRDVX5ECYW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UYYGECLB6OFQVINVQRDVX5ECYW/action/storage_attestation","attest_author":"https://pith.science/pith/UYYGECLB6OFQVINVQRDVX5ECYW/action/author_attestation","sign_citation":"https://pith.science/pith/UYYGECLB6OFQVINVQRDVX5ECYW/action/citation_signature","submit_replication":"https://pith.science/pith/UYYGECLB6OFQVINVQRDVX5ECYW/action/replication_record"}},"created_at":"2026-07-05T12:02:49.647762+00:00","updated_at":"2026-07-05T12:02:49.647762+00:00"}