{"paper":{"title":"Beyond Text Prompts: Visual-to-Visual Generation as A Unified Paradigm","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Visual specification pages replace text prompts in frozen generators without any retraining.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Haoxuan Che, Jean-Michel Morel, Kangning Cui, Meng Chu, Raymond H. Chan, Rui Liu, Suiyun Zhang, Xiaodong Cun, Yaofang Liu, Zhaoqing Li","submitted_at":"2026-05-12T15:35:34Z","abstract_excerpt":"Humans often specify and create through visual artifacts: typography sheets, sketches, reference images, and annotated scenes. Yet modern visual generators still ask users to serialize this intent into text, a bottleneck that compresses signals like spatial structure, exact appearance, and glyph shape. We propose \\textbf{\\emph{visual-to-visual} (V2V)} generation, in which the user conditions a generative model with a visual specification page rather than a text prompt. The page is not an edit target, but a visual document that specifies the desired output. We introduce \\textbf{V2V-Zero}, a tra"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"V2V-Zero reaches 0.85 on GenEval with a frozen Qwen-Image backbone, closely matching its optimized text-to-image performance without fine-tuning, and scores 32.7/100 on Simple-V2V Bench while a video extension scores 20.2/100.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The frozen VLM already maps both text and images into the generator's conditioning space so that final-layer hidden states from visual pages can replace text conditioning without any fine-tuning or architectural changes.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"V2V-Zero adapts frozen VLMs for visual conditioning via hidden states from specification pages, scoring 0.85 on GenEval and 32.7 on a new seven-task benchmark while revealing capability hierarchies in attribute binding and structural control.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Visual specification pages replace text prompts in frozen generators without any retraining.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"8274e763bce1d3161b4a9cacccf7d804a20e8f7e95d6e45c2976c625ad3f3f6e"},"source":{"id":"2605.12271","kind":"arxiv","version":2},"verdict":{"id":"a5ab15ab-d024-4dc1-944f-22d1a68cae8e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T05:56:50.607649Z","strongest_claim":"V2V-Zero reaches 0.85 on GenEval with a frozen Qwen-Image backbone, closely matching its optimized text-to-image performance without fine-tuning, and scores 32.7/100 on Simple-V2V Bench while a video extension scores 20.2/100.","one_line_summary":"V2V-Zero adapts frozen VLMs for visual conditioning via hidden states from specification pages, scoring 0.85 on GenEval and 32.7 on a new seven-task benchmark while revealing capability hierarchies in attribute binding and structural control.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The frozen VLM already maps both text and images into the generator's conditioning space so that final-layer hidden states from visual pages can replace text conditioning without any fine-tuning or architectural changes.","pith_extraction_headline":"Visual specification pages replace text prompts in frozen generators without any retraining."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.12271/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-26T14:45:24.437485Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-20T14:01:25.467010Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-20T10:17:13.945599Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T22:41:58.321139Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"06ac31b9b5c5d3c1c8e7e01e86603ebd62e45f52bf211ba126be209800a564cf"},"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"}