{"paper":{"title":"Fast Image Super-Resolution via Consistency Rectified Flow","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Rectified flow from low-resolution to high-resolution images enables single-step super-resolution when trained with consistency constraints and dual scheduling.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Fan Li, Haoran Yang, Haoze Sun, Jiaqi Xu, Jingjing Ren, Long Peng, Pheng-Ann Heng, Renjing Pei, Wenbo Li, Xiaowei Hu, Zhixin Wang","submitted_at":"2026-05-12T16:42:38Z","abstract_excerpt":"Diffusion models (DMs) have demonstrated remarkable success in real-world image super-resolution (SR), yet their reliance on time-consuming multi-step sampling largely hinders their practical applications. While recent efforts have introduced few- or single-step solutions, existing methods either inefficiently model the process from noisy input or fail to fully exploit iterative generative priors, compromising the fidelity and quality of the reconstructed images. To address this issue, we propose FlowSR, a novel approach that reformulates the SR problem as a rectified flow from low-resolution "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"FlowSR achieves outstanding performance in both efficiency and image quality by reformulating the SR problem as a rectified flow from LR to HR images with an improved consistency learning strategy incorporating HR regularization and fast-slow scheduling.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the HR regularization term forces precise convergence to ground-truth HR targets and that the fast-slow scheduling captures fine-grained textures without introducing artifacts or fidelity loss in the single-step inference.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"FlowSR enables single-step image super-resolution by learning a rectified flow from LR to HR with consistency distillation, HR regularization, and dual fast-slow timestep scheduling.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Rectified flow from low-resolution to high-resolution images enables single-step super-resolution when trained with consistency constraints and dual scheduling.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"caa78e891cc1712f7414b791087a7a3d794b3a7d6d79eb26d8625c9264230859"},"source":{"id":"2605.12377","kind":"arxiv","version":2},"verdict":{"id":"62612083-7262-4f37-ad6c-d46fdd92b451","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T07:11:08.781019Z","strongest_claim":"FlowSR achieves outstanding performance in both efficiency and image quality by reformulating the SR problem as a rectified flow from LR to HR images with an improved consistency learning strategy incorporating HR regularization and fast-slow scheduling.","one_line_summary":"FlowSR enables single-step image super-resolution by learning a rectified flow from LR to HR with consistency distillation, HR regularization, and dual fast-slow timestep scheduling.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the HR regularization term forces precise convergence to ground-truth HR targets and that the fast-slow scheduling captures fine-grained textures without introducing artifacts or fidelity loss in the single-step inference.","pith_extraction_headline":"Rectified flow from low-resolution to high-resolution images enables single-step super-resolution when trained with consistency constraints and dual scheduling."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.12377/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-26T14:40:29.634334Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-20T13:31:25.107504Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-20T09:46:56.814660Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T22:41:58.230120Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"8b14572a19b6731a076790ba7ebb1001f9e4e0a704c81b7ef48ce14edf02c6a3"},"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"}