{"paper":{"title":"PRISM: Prior Rectification and Uncertainty-Aware Structure Modeling for Diffusion-Based Text Image Super-Resolution","license":"http://creativecommons.org/licenses/by/4.0/","headline":"PRISM uses flow matching on paired latents and uncertainty-aware residuals to correct unreliable text priors and refine stroke boundaries inside a single diffusion pass.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Xiaokang Yang, Xiaoyang Liu, Yulun Zhang, Zheng Chen, Zihang Xu","submitted_at":"2026-05-13T05:31:06Z","abstract_excerpt":"Text image super-resolution (Text-SR) requires more than visually plausible detail synthesis: slight errors in stroke topology may alter character identity and break readability. Existing methods improve text fidelity with stronger recognition-based or generative priors, yet they still face two unresolved challenges under severe degradation: the text condition extracted from low-quality inputs can itself be unreliable, and a plausible global prior does not fully determine fine-grained stroke boundaries. We present PRISM, a single-step diffusion-based Text-SR framework that addresses these two "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"PRISM achieves state-of-the-art performance with millisecond-level inference on both synthetic and real-world benchmarks.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the privileged training-time prior constructed from paired low-quality/high-quality latents combined with uncertainty-aware structural residuals will reliably correct stroke boundaries under severe real-world degradation without introducing new identity-altering errors.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"PRISM improves text image super-resolution by rectifying global priors with flow-matching and modeling local structural uncertainty in a single diffusion pass, achieving SOTA results at millisecond inference.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"PRISM uses flow matching on paired latents and uncertainty-aware residuals to correct unreliable text priors and refine stroke boundaries inside a single diffusion pass.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"393d3e4ae5d98dd3ebd5f664f2fa5a1622136eeb16ab5427b0742ffa36b567c8"},"source":{"id":"2605.13027","kind":"arxiv","version":1},"verdict":{"id":"4fc74ba2-f8b1-4d00-a6bb-e4adb0706c67","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:20:35.209545Z","strongest_claim":"PRISM achieves state-of-the-art performance with millisecond-level inference on both synthetic and real-world benchmarks.","one_line_summary":"PRISM improves text image super-resolution by rectifying global priors with flow-matching and modeling local structural uncertainty in a single diffusion pass, achieving SOTA results at millisecond inference.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the privileged training-time prior constructed from paired low-quality/high-quality latents combined with uncertainty-aware structural residuals will reliably correct stroke boundaries under severe real-world degradation without introducing new identity-altering errors.","pith_extraction_headline":"PRISM uses flow matching on paired latents and uncertainty-aware residuals to correct unreliable text priors and refine stroke boundaries inside a single diffusion pass."},"references":{"count":58,"sample":[{"doi":"","year":2024,"title":"One-step effective diffusion network for real-world image super-resolution","work_id":"04042b14-25a5-4433-ae0b-0b31b8f909be","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Tsd-sr: One-step diffusion with target score distillation for real-world image super- resolution","work_id":"f8e5530a-2017-466b-9723-5e3dd288a485","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"One diffusion step to real-world super-resolution via flow trajectory distillation","work_id":"38ba7d9d-5886-40c3-9797-2261d65fd84e","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Learning generative structure prior for blind text image super-resolution","work_id":"ba98adde-5c3a-4d0b-83d4-69241593212e","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Scene text image super-resolution in the wild","work_id":"48132753-2a6b-4d75-b80c-40e0e536fcd7","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":58,"snapshot_sha256":"83758de1edb7ceeb3395dfd84852ed47c6ebdc93cacf747fff1a04b57b9f1571","internal_anchors":2},"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"}