mlsanityData to 14 Oct 2026

Efficient Scene-Text Refinement via Pixel-Space Grouped Image Generation

arXiv 2610.18284 ↗First posted 13 Oct 2026cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Generated images often have coherent layouts and visual content, yet local errors in small text and complex characters still limit their readability and practical use. Scene-text refinement requires accuracy, preservation, and efficiency. Existing methods based on variational autoencoders (VAEs) can lose fine strokes during reconstruction, while full-image editing can shift background colors. In this work, we introduce PixText, a model that formulates local text refinement as grouped image generation in pixel space. Conditioned on the original image, PixText generates only the target regions, preserves unedited content, and uses MeanFlow for four-step generation. We further introduce Text Refinement Bench to evaluate text correction and image preservation in generated images. On this benchmark, PixText achieves 83.06% glyph pass and 94.66% exact target-character OCR accuracy, versus 36.77% and 52.31% for the strongest baseline. Its four-step MeanFlow variant retains an 81.84% pass rate with less than one GPU-second of amortized generation cost per image.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 77 of its 85 references that are in the map. First posted on arXiv 13 Oct 2026.

Builds on

77 of 85 references.

References · 77

  1. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
  2. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,177
  3. 2022Scalable Diffusion Models with Transformerscited by 6,054
  4. 2022Classifier-Free Diffusion Guidancecited by 5,847
  5. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,703
  6. 2025Qwen-Image Technical Reportcited by 1,044
  7. 2025FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Spacecited by 1,006
  8. 2025Mean Flows for One-step Generative Modelingcited by 561
  9. 2023Simple diffusion: End-to-end diffusion for high resolution imagescited by 422
  10. 2019Character Region Awareness for Text Detectioncited by 278
  11. 2023TextDiffuser: Diffusion Models as Text Painterscited by 229
  12. 2024Deep Compression Autoencoder for Efficient High-Resolution Diffusion Modelscited by 217
38 more
  1. 2022Scalable Adaptive Computation for Iterative Generationcited by 195
  2. 2024Flex Attention: A Programming Model for Generating Optimized Attention Kernelscited by 185
  3. 2025PaddleOCR 3.0 Technical Reportcited by 143
  4. 2023TextDiffuser-2: Unleashing the Power of Language Models for Text Renderingcited by 140
  5. 2023GlyphControl: Glyph Conditional Control for Visual Text Generationcited by 131
  6. 2024Scalable High-Resolution Pixel-Space Image Synthesis with Hourglass Diffusion Transformerscited by 118
  7. 2025Improved Mean Flows: On the Challenges of Fastforward Generative Modelscited by 110
  8. 2025LongCat-Image Technical Reportcited by 100
  9. 2025PixelFlow: Pixel-Space Generative Models with Flowcited by 97
  10. 2025PixelDiT: Pixel Diffusion Transformers for Image Generationcited by 83
  11. 2019Editing Text in the Wildcited by 82
  12. 2025PixNerd: Pixel Neural Field Diffusioncited by 82
  13. 2024Glyph-ByT5: A Customized Text Encoder for Accurate Visual Text Renderingcited by 81
  14. 2025DeCo: Frequency-Decoupled Pixel Diffusion for End-to-End Image Generationcited by 60
  15. 2023UDiffText: A Unified Framework for High-quality Text Synthesis in Arbitrary Images via Character-aware Diffusion Modelscited by 59
  16. 2024AnyText2: Visual Text Generation and Editing With Customizable Attributescited by 53
  17. 2026One-step Latent-free Image Generation with Pixel Mean Flowscited by 53
  18. 2020SwapText: Image Based Texts Transfer in Scenescited by 50
  19. 2021TextStyleBrush: Transfer of Text Aesthetics from a Single Examplecited by 49
  20. 2024Glyph-ByT5-v2: A Strong Aesthetic Baseline for Accurate Multilingual Visual Text Renderingcited by 48
  21. 2025DiP: Taming Diffusion Models in Pixel Spacecited by 48
  22. 2019STEFANN: Scene Text Editor using Font Adaptive Neural Networkcited by 47
  23. 2025FLUX-Text: A Simple and Advanced Diffusion Transformer Baseline for Scene Text Editingcited by 46
  24. 2023Improving Diffusion Models for Scene Text Editing with Dual Encoderscited by 45
  25. 2023DiffUTE: Universal Text Editing Diffusion Modelcited by 45
  26. 2024TextCtrl: Diffusion-based Scene Text Editing with Prior Guidance Controlcited by 45
  27. 2022Exploring Stroke-Level Modifications for Scene Text Editingcited by 37
  28. 2026Latent Forcing: Reordering the Diffusion Trajectory for Pixel-Space Image Generationcited by 36
  29. 2026FireRed-Image-Edit-1.0 Technical Reportcited by 36
  30. 2026HiDream-O1-Image: A Natively Unified Image Generative Foundation Model with Pixel-level Unified Transformercited by 32
  31. 2024GlyphDraw2: Automatic Generation of Complex Glyph Posters with Diffusion Models and Large Language Modelscited by 29
  32. 2026InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editingcited by 28
  33. 2026AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillationcited by 26
  34. 2024Edify Image: High-Quality Image Generation with Pixel Space Laplacian Diffusion Modelscited by 23
  35. 2026Qwen-Image-Bench: From Generation to Creation in Text-to-Image Evaluationcited by 22
  36. 2024DreamText: High Fidelity Scene Text Synthesiscited by 18
  37. 2026PixelGen: Improving Pixel Diffusion with Perceptual Supervisioncited by 14
  38. 2025TextFlux: An OCR-Free DiT Model for High-Fidelity Multilingual Scene Text Synthesiscited by 13

50 most cited of 77 All 77 (JSON)

Built on by

Not cited yet.