{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KJSD3RHTGVC4FXS2IRU7EFDO6Q","short_pith_number":"pith:KJSD3RHT","schema_version":"1.0","canonical_sha256":"52643dc4f33545c2de5a4469f2146ef430be236bd3085d71c7a91e71bd20192e","source":{"kind":"arxiv","id":"2605.16810","version":1},"attestation_state":"computed","paper":{"title":"Training-Free Occluded Text Rendering via Glyph Priors and Attention-Guided Semantic Blending","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A restarted dual-stream framework enables training-free occluded text rendering by preserving typography via glyph priors and attention-guided mask replacement.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hongtian Wang, Jingqi Hou","submitted_at":"2026-05-16T04:58:07Z","abstract_excerpt":"We present a training-free framework for occluded text rendering with a pretrained FLUX.1-dev backbone. The task requires a model to render recognizable typography and place an occluding object over the intended text region. This setting remains difficult for existing text-to-image generators: the occluder often drifts away from the text, while the text may be distorted or appear to float on top of the occluding object. To address this problem, we propose a restarted dual-stream inference framework that decouples text-layout preservation from occluder insertion. A Base Stream provides a clean "},"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":"2605.16810","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-16T04:58:07Z","cross_cats_sorted":[],"title_canon_sha256":"87a83de8127fedb229b677efd3c67af0a9b7fd81b090b459232112f89678eeb7","abstract_canon_sha256":"6b222bd4162af6ddb633eda396d138458888a4bc99c8f85bed745bb145d2ce52"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:03:23.507687Z","signature_b64":"Q0+6wYUDLOSZ9mYNSA1pm7hdStFcX/sAq7eFVUOO97Z5ZIf477QcdXVYazvaw8HNL7wI+RAoleooISLcFx9UCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"52643dc4f33545c2de5a4469f2146ef430be236bd3085d71c7a91e71bd20192e","last_reissued_at":"2026-05-20T00:03:23.506591Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:03:23.506591Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Training-Free Occluded Text Rendering via Glyph Priors and Attention-Guided Semantic Blending","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A restarted dual-stream framework enables training-free occluded text rendering by preserving typography via glyph priors and attention-guided mask replacement.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hongtian Wang, Jingqi Hou","submitted_at":"2026-05-16T04:58:07Z","abstract_excerpt":"We present a training-free framework for occluded text rendering with a pretrained FLUX.1-dev backbone. The task requires a model to render recognizable typography and place an occluding object over the intended text region. This setting remains difficult for existing text-to-image generators: the occluder often drifts away from the text, while the text may be distorted or appear to float on top of the occluding object. To address this problem, we propose a restarted dual-stream inference framework that decouples text-layout preservation from occluder insertion. A Base Stream provides a clean "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experiments on representative occluded text scenarios demonstrate substantially improved text readability and competitive occlusion alignment, yielding more stable object-on-text compositions without any model fine-tuning.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the spectral glyph-prior from FreeText combined with token-conditioned attention can reliably localize the target text region and produce an anchor-aware hard fusion mask that allows clean K/V replacement without distorting typography or causing the occluder to drift.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A restarted dual-stream inference approach with glyph priors and attention-guided masks improves occluded text rendering in pretrained diffusion models without fine-tuning.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A restarted dual-stream framework enables training-free occluded text rendering by preserving typography via glyph priors and attention-guided mask replacement.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"5647b5a76b893ee0a8edf8990a18bc51900a18800e76dc71d0fab8aa5dd61fe4"},"source":{"id":"2605.16810","kind":"arxiv","version":1},"verdict":{"id":"4fd93a9c-4d0e-4bca-b73f-9c73f3e580fc","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T21:00:46.995818Z","strongest_claim":"Experiments on representative occluded text scenarios demonstrate substantially improved text readability and competitive occlusion alignment, yielding more stable object-on-text compositions without any model fine-tuning.","one_line_summary":"A restarted dual-stream inference approach with glyph priors and attention-guided masks improves occluded text rendering in pretrained diffusion models without fine-tuning.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the spectral glyph-prior from FreeText combined with token-conditioned attention can reliably localize the target text region and produce an anchor-aware hard fusion mask that allows clean K/V replacement without distorting typography or causing the occluder to drift.","pith_extraction_headline":"A restarted dual-stream framework enables training-free occluded text rendering by preserving typography via glyph priors and attention-guided mask replacement."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.16810/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T21:31:19.264544Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T21:11:43.436972Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T19:01:56.278867Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T18:33:26.417367Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"66faf80359618712032300813a935e53ebdef6e47dc2e0f05a52381be81aca37"},"references":{"count":20,"sample":[{"doi":"","year":2025,"title":"arXiv preprint arXiv:2503.23461 (2025)","work_id":"4d349026-15b1-433b-ace5-5104109b2bef","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2026,"title":"FreeText: Training-Free Text Rendering in Diffusion Transformers via Attention Localization and Spectral Glyph Injection","work_id":"a3717c37-b25c-411d-935c-fa126aec7dd4","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"arXiv preprint arXiv:2305.10855 (2023) 7","work_id":"f43372f1-5688-4262-a5f2-2e8f7da8d54c","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Anytext: Multilingual visual text generation and editing","work_id":"6ca52f26-db50-422e-86c6-89c334703122","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"TextDiffuser-2: Unleashing the Power of Language Models for Text Rendering","work_id":"8cc07be4-b59c-466e-b152-0a92bc8ca0d6","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":20,"snapshot_sha256":"a659294889b05cba20740b9907961382afb6978b07a75375ea672944d1d67d77","internal_anchors":2},"formal_canon":{"evidence_count":2,"snapshot_sha256":"cf7b3d4d03917a5bc5463bf35a5d7fbbacccb88453cde739eb153d258d5af115"},"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":"2605.16810","created_at":"2026-05-20T00:03:23.506776+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.16810v1","created_at":"2026-05-20T00:03:23.506776+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.16810","created_at":"2026-05-20T00:03:23.506776+00:00"},{"alias_kind":"pith_short_12","alias_value":"KJSD3RHTGVC4","created_at":"2026-05-20T00:03:23.506776+00:00"},{"alias_kind":"pith_short_16","alias_value":"KJSD3RHTGVC4FXS2","created_at":"2026-05-20T00:03:23.506776+00:00"},{"alias_kind":"pith_short_8","alias_value":"KJSD3RHT","created_at":"2026-05-20T00:03:23.506776+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KJSD3RHTGVC4FXS2IRU7EFDO6Q","json":"https://pith.science/pith/KJSD3RHTGVC4FXS2IRU7EFDO6Q.json","graph_json":"https://pith.science/api/pith-number/KJSD3RHTGVC4FXS2IRU7EFDO6Q/graph.json","events_json":"https://pith.science/api/pith-number/KJSD3RHTGVC4FXS2IRU7EFDO6Q/events.json","paper":"https://pith.science/paper/KJSD3RHT"},"agent_actions":{"view_html":"https://pith.science/pith/KJSD3RHTGVC4FXS2IRU7EFDO6Q","download_json":"https://pith.science/pith/KJSD3RHTGVC4FXS2IRU7EFDO6Q.json","view_paper":"https://pith.science/paper/KJSD3RHT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.16810&json=true","fetch_graph":"https://pith.science/api/pith-number/KJSD3RHTGVC4FXS2IRU7EFDO6Q/graph.json","fetch_events":"https://pith.science/api/pith-number/KJSD3RHTGVC4FXS2IRU7EFDO6Q/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KJSD3RHTGVC4FXS2IRU7EFDO6Q/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KJSD3RHTGVC4FXS2IRU7EFDO6Q/action/storage_attestation","attest_author":"https://pith.science/pith/KJSD3RHTGVC4FXS2IRU7EFDO6Q/action/author_attestation","sign_citation":"https://pith.science/pith/KJSD3RHTGVC4FXS2IRU7EFDO6Q/action/citation_signature","submit_replication":"https://pith.science/pith/KJSD3RHTGVC4FXS2IRU7EFDO6Q/action/replication_record"}},"created_at":"2026-05-20T00:03:23.506776+00:00","updated_at":"2026-05-20T00:03:23.506776+00:00"}