{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:JF2CHEPWQTAWNURNM4KBEEOHG7","short_pith_number":"pith:JF2CHEPW","schema_version":"1.0","canonical_sha256":"49742391f684c166d22d67141211c737f72cc381b799a1a9fda0ee8c875838d0","source":{"kind":"arxiv","id":"2406.01970","version":1},"attestation_state":"computed","paper":{"title":"The Crystal Ball Hypothesis in diffusion models: Anticipating object positions from initial noise","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Boqing Gong, Cho-Jui Hsieh, Minhao Cheng, Ruochen Wang, Tianyi Zhou, Yuanhao Ban","submitted_at":"2024-06-04T05:06:00Z","abstract_excerpt":"Diffusion models have achieved remarkable success in text-to-image generation tasks; however, the role of initial noise has been rarely explored. In this study, we identify specific regions within the initial noise image, termed trigger patches, that play a key role for object generation in the resulting images. Notably, these patches are ``universal'' and can be generalized across various positions, seeds, and prompts. To be specific, extracting these patches from one noise and injecting them into another noise leads to object generation in targeted areas. We identify these patches by analyzi"},"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":"2406.01970","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-06-04T05:06:00Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"67cc30d5433831113dbba9c893c1ca6fe1ec830da6bd6a9a8bec0075cecf152c","abstract_canon_sha256":"369d534ad1a1251d09d95da668ce0132f082d75fb6279d09b0c061e254aa37df"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:27:12.389721Z","signature_b64":"0pIWbNxao8iwvA/Tp+qeUaAqp3ok9mW8+gm2G7FO1dYrryN/0a8IflfJeEt5eKzPLtZ2uUve3DZH967iuoZaCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"49742391f684c166d22d67141211c737f72cc381b799a1a9fda0ee8c875838d0","last_reissued_at":"2026-07-05T08:27:12.389243Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:27:12.389243Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Crystal Ball Hypothesis in diffusion models: Anticipating object positions from initial noise","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Boqing Gong, Cho-Jui Hsieh, Minhao Cheng, Ruochen Wang, Tianyi Zhou, Yuanhao Ban","submitted_at":"2024-06-04T05:06:00Z","abstract_excerpt":"Diffusion models have achieved remarkable success in text-to-image generation tasks; however, the role of initial noise has been rarely explored. In this study, we identify specific regions within the initial noise image, termed trigger patches, that play a key role for object generation in the resulting images. Notably, these patches are ``universal'' and can be generalized across various positions, seeds, and prompts. To be specific, extracting these patches from one noise and injecting them into another noise leads to object generation in targeted areas. We identify these patches by analyzi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.01970","kind":"arxiv","version":1},"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/2406.01970/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":"2406.01970","created_at":"2026-07-05T08:27:12.389301+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.01970v1","created_at":"2026-07-05T08:27:12.389301+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.01970","created_at":"2026-07-05T08:27:12.389301+00:00"},{"alias_kind":"pith_short_12","alias_value":"JF2CHEPWQTAW","created_at":"2026-07-05T08:27:12.389301+00:00"},{"alias_kind":"pith_short_16","alias_value":"JF2CHEPWQTAWNURN","created_at":"2026-07-05T08:27:12.389301+00:00"},{"alias_kind":"pith_short_8","alias_value":"JF2CHEPW","created_at":"2026-07-05T08:27:12.389301+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2411.15115","citing_title":"Self-Correcting Text-to-Video Generation with Misalignment Detection and Localized Refinement","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2601.00090","citing_title":"It's Never Too Late: Noise Optimization for Collapse Recovery in Trained Diffusion Models","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09850","citing_title":"Training-Free Object-Background Compositional T2I via Dynamic Spatial Guidance and Multi-Path Pruning","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07253","citing_title":"LENS: Low-Frequency Eigen Noise Shaping for Efficient Diffusion Sampling","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19730","citing_title":"FASTER: Value-Guided Sampling for Fast RL","ref_index":20,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JF2CHEPWQTAWNURNM4KBEEOHG7","json":"https://pith.science/pith/JF2CHEPWQTAWNURNM4KBEEOHG7.json","graph_json":"https://pith.science/api/pith-number/JF2CHEPWQTAWNURNM4KBEEOHG7/graph.json","events_json":"https://pith.science/api/pith-number/JF2CHEPWQTAWNURNM4KBEEOHG7/events.json","paper":"https://pith.science/paper/JF2CHEPW"},"agent_actions":{"view_html":"https://pith.science/pith/JF2CHEPWQTAWNURNM4KBEEOHG7","download_json":"https://pith.science/pith/JF2CHEPWQTAWNURNM4KBEEOHG7.json","view_paper":"https://pith.science/paper/JF2CHEPW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.01970&json=true","fetch_graph":"https://pith.science/api/pith-number/JF2CHEPWQTAWNURNM4KBEEOHG7/graph.json","fetch_events":"https://pith.science/api/pith-number/JF2CHEPWQTAWNURNM4KBEEOHG7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JF2CHEPWQTAWNURNM4KBEEOHG7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JF2CHEPWQTAWNURNM4KBEEOHG7/action/storage_attestation","attest_author":"https://pith.science/pith/JF2CHEPWQTAWNURNM4KBEEOHG7/action/author_attestation","sign_citation":"https://pith.science/pith/JF2CHEPWQTAWNURNM4KBEEOHG7/action/citation_signature","submit_replication":"https://pith.science/pith/JF2CHEPWQTAWNURNM4KBEEOHG7/action/replication_record"}},"created_at":"2026-07-05T08:27:12.389301+00:00","updated_at":"2026-07-05T08:27:12.389301+00:00"}