TextAlign: Preference Alignment for Text Rendering with Hierarchical Rewards

• URL: http://arxiv.org/abs/2605.19320v1
• Date: Tue, 19 May 2026 03:55:59 GMT
• Title: TextAlign: Preference Alignment for Text Rendering with Hierarchical Rewards
• Abstract summary: We study text rendering as a post-training preference-alignment problem.<n>Key component is a hierarchical vision-language model (VLM)-based reward that decomposes rendering errors into global, word, and glyph levels.<n>Experiments on FLUX.1-dev and Z-Image-Turbo show consistent gains in OCR-based text accuracy without degrading general generation quality.
• Score: 25.768329293709176
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Faithful text rendering remains a persistent weakness of large text-to-image generative models, as it requires both semantic instruction following and fine-grained glyph-level structure. Prior methods often improve this ability through architecture-specific modules or encoder modifications, which complicate deployment across foundation models. We study text rendering as a post-training preference-alignment problem and propose TextAlign, a non-invasive framework that keeps the generator architecture unchanged. The key component is a hierarchical vision-language model (VLM)-based reward that decomposes rendering errors into global, word, and glyph levels, then converts binary defect judgments into a scalar preference signal. The resulting signal supports both Group Relative Policy Optimization (GRPO) and Direct Preference Optimization (DPO). Experiments on FLUX.1-dev and Z-Image-Turbo show consistent gains in OCR-based text accuracy without degrading general generation quality. Compared with strong foundation and text-rendering baselines, including SD3.5, Qwen-Image, AnyText, and TextDiffuser, these results indicate that reward design offers a scalable alternative to model redesign for improving text rendering.

Related papers

• GlyphAnchor: Enhancing Visual Text Rendering via Position-Anchored Glyph Priors [44.606371845449424]
  We introduce GlyphAnchor, a novel text-rendering enhancement method for both text-to-image and image-editing diffusion transformer models.<n> Experiments across multiple backbones and benchmarks, including long, complex, and densely arranged text and rare character scenarios, show that GlyphAnchor consistently improves text fidelity while preserving overall image quality.
  arXiv  Detail & Related papers  (2026-09-02T09:23:23Z)
• Pixel-TTS: Image based Text Rendering for Robust Text-to-Speech [3.62994537177299]
  Grounding text in its visual form allows structurally similar characters with different Unicode encodings to produce similar embeddings.<n>We propose Pixel-TTS, the first framework for visually grounded speech synthesis.<n>It renders text as images and projects them through a 2D convolutional layer to generate embeddings.
  arXiv  Detail & Related papers  (2026-06-05T11:46:30Z)
• TextSculptor: Training and Benchmarking Scene Text Editing [88.11688559021628]
  We present TextSculptor, a comprehensive framework for data construction and evaluation of scene text editing.<n>TextSculptor improves open-source text editing performance and narrows the gap to proprietary models.
  arXiv  Detail & Related papers  (2026-05-20T12:22:26Z)
• CycleCap: Improving VLMs Captioning Performance via Self-Supervised Cycle Consistency Fine-Tuning [65.10059440725041]
  Visual-Language Models (VLMs) have achieved remarkable progress in image captioning, visual question answering, and visual reasoning.<n>They remain prone to vision-language misalignment, often producing overly generic or hallucinated descriptions.<n>Existing approaches address this via instruction tuning-requiring costly, large-scale annotated datasets or via complex test-time frameworks for caption refinement.<n>In this work, we revisit image-text alignment through the lens of cycle consistency.
  arXiv  Detail & Related papers  (2026-03-18T20:57:31Z)
• GlyphPrinter: Region-Grouped Direct Preference Optimization for Glyph-Accurate Visual Text Rendering [83.65452884616624]
  Existing methods typically enhance text rendering by training on a large amount of high-quality scene text images.<n>The limited coverage of glyph variations and excessive stylization often compromise glyph accuracy, especially for complex or out-of-domain characters.<n>We propose GlyphPrinter, a preference-based text rendering method that eliminates reliance on explicit reward models.
  arXiv  Detail & Related papers  (2026-03-16T17:59:31Z)
• TextPecker: Rewarding Structural Anomaly Quantification for Enhancing Visual Text Rendering [76.53315206999231]
  TextPecker is a plug-and-play structural anomaly perceptive RL strategy.<n>It mitigates noisy reward signals and works with any textto-image generators.<n>It significantly yields average gains of 4% in structural fidelity and 8.7% in semantic alignment for Chinese text rendering.
  arXiv  Detail & Related papers  (2026-02-24T13:40:23Z)
• BideDPO: Conditional Image Generation with Simultaneous Text and Condition Alignment [53.94214054918876]
  Conditional image generation enhances text-to-image synthesis with structural, spatial, or stylistic priors.<n>Current methods face challenges in handling conflicts between sources.<n>We propose a bidirectionally decoupled DPO framework (BideDPO)
  arXiv  Detail & Related papers  (2025-11-24T16:20:11Z)
• UniGlyph: Unified Segmentation-Conditioned Diffusion for Precise Visual Text Synthesis [38.658170067715965]
  We propose a segmentation-guided framework that uses pixel-level visual text masks as unified conditional inputs.<n>Our approach achieves state-of-the-art performance on the AnyText benchmark.<n>We also introduce two new benchmarks: GlyphMM-benchmark for testing layout and glyph consistency in complex, and MiniText-benchmark for assessing generation quality in small-scale text regions.
  arXiv  Detail & Related papers  (2025-07-01T17:42:19Z)
• TextDiff: Mask-Guided Residual Diffusion Models for Scene Text Image Super-Resolution [17.95994419104427]
  TextDiff is a diffusion-based framework tailored for scene text image super-resolution.<n>It achieves state-of-the-art (SOTA) performance on public benchmark datasets.<n>Our proposed MRD module is plug-and-play that effectively sharpens the text edges produced by SOTA methods.
  arXiv  Detail & Related papers  (2023-08-13T11:02:16Z)
• Towards Robust Scene Text Image Super-resolution via Explicit Location Enhancement [59.66539728681453]
  Scene text image super-resolution (STISR) aims to improve image quality while boosting downstream scene text recognition accuracy. Most existing methods treat the foreground (character regions) and background (non-character regions) equally in the forward process. We propose a novel method LEMMA that explicitly models character regions to produce high-level text-specific guidance for super-resolution.
  arXiv  Detail & Related papers  (2023-07-19T05:08:47Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.