Pith. sign in

REVIEW 4 major objections 5 minor 55 references

Type-R: Automatically Retouching Typos for Text-to-Image Generation

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper claims that a post-processing pipeline called Type-R can locate, erase, and rewrite misspelled words in images from any text-to-image model, improving text accuracy without sacrificing image quality.

desk verdict A solid post-processing pipeline for typo correction that deserves refereeing, but the unnamed OCR evaluator leaves the headline accuracy claim underdetermined. read the letter →

arxiv 2411.18159 v2 pith:453INXJZ submitted 2024-11-27 cs.CV

Wataru Shimoda , Naoto Inoue , Daichi Haraguchi , Hayato Mitani , Seiichi Uchida , Kota Yamaguchi This is my paper · ORCID
classification cs.CV
keywords text-to-imagegenerationtypographycorrectionpost-processingOCRtextrenderingimageinpaintingeditinglayoutregeneration
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Type-R is a post-processing approach to the persistent problem of misspelled or missing words in text-to-image generation. Instead of retraining or conditioning the image generator, it runs four automatic stages on the already generated image: detect words with OCR, erase unintended text, plan boxes for missing words, and iteratively rewrite misspelled words. The paper argues this corrects typos while preserving visual design, and that pairing Type-R with current generators such as Stable Diffusion or Flux delivers the best balance of text accuracy and image quality among the methods tested. A sympathetic reader would take away that typo-free design images can be obtained by retouching rather than by building better text renderers.

What carries the argument

The mechanism that carries the argument is a four-stage pipeline whose pivot is the word-matching step. Detection and recognition turn the image into a set of OCR words; an optimal matching with Levenshtein distance pairs them against the prompt words, with padding tokens for missing or extra words. This match decides which regions to erase, where to insert missing words, and which words to send to a text editor. The text editor is applied iteratively, and only successfully corrected regions are composited back, so the image changes as little as possible.

What would settle it

Take a generated image in which the requested word is rendered correctly but in a font the OCR model reads incorrectly; if Type-R then erases or rewrites that correct word, it shows the method's accuracy is bounded by its internal recognizer. A cleaner test is to evaluate Type-R's output with an independent OCR model from a different family; if the reported accuracy gain shrinks or disappears, the gain is partly tuning the image to the pipeline's own recognizer.

Watch

Extended reading notes

Core claim

The paper's central claim is that typographic errors in text-to-image outputs are better fixed after generation than by changing the generator. Type-R formulates the task as a matching problem between the words requested in the prompt and the words an OCR model reads in the image: an optimal assignment with Levenshtein edit distances decides which rendered words are correct, which are unwanted, which are missing, and which are misspelled. It then erases surplus text, asks a vision-language model to place missing words, and repeatedly applies a text editor until the OCR no longer reports an error. The reported experiments show that Type-R paired with the Flux generator reaches the highest OCR accuracy among the compared methods while keeping graphic design quality near that of the raw generator, and similar gains appear when it is attached to other base generators.

Load-bearing premise

The pipeline treats the OCR model's reading as the truth about which words in the image are wrong, so a misread can make it erase or rewrite a correct word or leave a real typo untouched.

Editorial extensions

If this is right

  • Any current or future text-to-image generator can be paired with Type-R without fine-tuning, since the pipeline treats the generator as a black box.
  • The main comparison reports that Type-R with Flux beats text-focused generators TextDiffuser and TextDiffuser-2 on OCR accuracy and on the quality-accuracy trade-off, implying dedicated layout-conditioned generators are not required for typo-free output.
  • The pipeline's gains stack: erasing surplus words, regenerating missing layouts, and correcting typos each add accuracy, with layout correction contributing most for Flux.
  • Iterative typo correction concentrates its benefit early, reaching roughly 80 percent of total improvement within four attempts across base models.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If OCR accuracy is the bottleneck, Type-R effectively converts 'render the text correctly' into 'render the text legibly to a chosen recognizer'; the paper's finding that the text editor works best with the recognition model it was trained with suggests the reported metric can be sensitive to model pairing.
  • The same retouching pipeline could be repurposed as an automatic data-cleaning tool: instead of discarding images with typos from training sets, one could retouch them and keep the image, which may be cheaper than regenerating.
  • Because Type-R depends on OCR, a text editor, and a vision-language model, its ceiling is set by those components, so improvements in any one should transfer directly to the pipeline without changing its structure.
  • The pipeline decomposes text by spaces, so it needs explicit handling for scripts without spaces or for reading-order-sensitive captions before it generalizes beyond the English-like cases tested; the paper notes that reading order is not guaranteed.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. Type-R is a training-free post-processing pipeline that detects and repairs typographic errors in images generated by arbitrary text-to-image models. It operates in four stages: word-level error detection via scene-text detection/recognition matched to prompt words by optimal transport (Eq. 1), erasure of unintended words with inpainting, layout regeneration of missing words using GPT-4o, and iterative typo correction via a text-editing model (Algorithm 1). The paper evaluates Type-R combined with SD3, Flux, SD3.5, and DALL-E 3 on the MARIO-Eval benchmark, reporting OCR accuracy, GPT-based graphic quality/content alignment scores, FID, CLIP score, and a human user study. The central claim is that Type-R with Flux achieves the best trade-off between text rendering accuracy and image quality.

Significance. If the claims hold, Type-R is a practically useful contribution: it offers a plug-and-play improvement over existing text-to-image generators without fine-tuning, and the ablation studies (Tables 3, 4, 5, 9, 11, 12) give a fairly complete picture of how each module and each backend choice matters. The paper is also transparent about limitations, including OCR failures and the need for better backend models. The main weak point is that the headline 'text rendering accuracy' rests entirely on an OCR evaluator that is never named in the main text, while the pipeline itself is driven by an OCR model from the same family (Paddle). This creates a real risk that the reported gains partly measure Paddle-readability rather than human-legible correctness. The human study (Table 2) asks about graphic quality and content matching, not word-level legibility, so it does not resolve this concern.

major comments (4)
  1. [§4.2 and Fig. 8 caption] The evaluation OCR model used to compute the headline OCR accuracy in Tables 1, 3, 4, 5, 9, 11, and 12 is never specified in the main text. The Fig. 8 caption only says that OCR* is computed by the OCR models in Type-R and is distinct from the OCR model for evaluation, but the main text does not name the evaluator. Given that the pipeline internally uses Paddle (Sec. 4.3) and Sec. 4.7 (Table 4) explicitly shows that AnyText's compatibility with Paddle affects the measured OCR accuracy, the reported gains could partly reflect the pipeline's ability to produce Paddle-readable text rather than generally legible text. The authors should specify the exact evaluation OCR model and rerun the main comparisons with at least one recognizer that is not used anywhere in Type-R, plus a human word-level legibility check.
  2. [§3.1, Algorithm 1, and Appendix F] The pipeline treats OCR output as ground truth for what text is actually in the image. In Eq. (1) and Sec. 3.1, detected words are matched to prompt words using Levenshtein distance; in Algorithm 1, the same OCR readings decide when to stop retyping. Appendix F explicitly calls the internal error counts 'pseudo-errors.' This makes the method vulnerable to OCR misreads: a correct word misread as wrong can be erased or retyped, and a wrong word misread as correct will be left in place. The paper acknowledges in the Limitation section that OCR models sometimes fail, but it does not quantify how often this happens in the benchmark or how it affects final image correctness. A sensitivity analysis with a second, independent recognizer would help establish that the correction decisions are not systematically biased by the internal OCR's blind spots.
  3. [Supplementary A and Table 2] The claim of 'highest text rendering accuracy' is supported only by OCR accuracy. The GPT-based graphic quality scores show substantial disagreement with human ratings (Supplementary A reports, e.g., 56% of GPT scores above 8.0 versus 3% for humans for Type-R w/ Flux), and the user study in Table 2 asks participants to rate graphic design quality and text-image matching, not word-level legibility. Thus the evaluation does not directly measure whether a human reader can correctly read the retouched words. The authors should add a small human reading test (e.g., forced-choice word identification) on a sample of images, or at least report per-word human legibility agreement for the main comparison.
  4. [Table 1 and Sec. 4.5] The main comparison reports single OCR accuracy numbers and GPT scores without error bars, confidence intervals, or per-prompt statistical tests. The user study provides p-values for the matching score only, not for OCR accuracy or graphic quality. Because the benchmark has 5,414 prompts, bootstrapping or per-prompt mean differences with paired tests would be straightforward to add and would make the headline 'significantly outperforms' claim more robust.
minor comments (5)
  1. [§4.3] The detector is called 'Deepsolo' in the implementation details but 'DeepSolo' in Table 4 and the references; please use consistent capitalization.
  2. [Fig. 8] The symbol OCR* is used in the caption and main text without a formal definition; define it where it first appears.
  3. [Appendix B] The citation '[15?]' contains a stray question mark; replace it with the proper reference.
  4. [§4.2] The phrase 'compute the similarity between each image and its prompt-based on the CLIP' should read 'based on CLIP' or 'using CLIP.'
  5. [Table 7] The grid search selects θ on a validation set, but the main results then use the selected value on the benchmark; a sentence noting that this selection did not use the benchmark would help clarify the protocol.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: Type-R is a self-contained post-processing pipeline; the OCR-in-the-loop correction is not the same as the evaluation metric by the paper's own statement, though the evaluation OCR is unnamed.

full rationale

Type-R's derivation chain is an engineering pipeline, not a mathematical derivation. Error detection (Sec. 3.1) uses prompt words as the specification and an off-the-shelf OCR (Paddle) to read the generated image; typo correction (Sec. 3.4, Algorithm 1) iteratively re-edits until the internal recognizer reads the target words, erases surplus text, and regenerates missing layouts. The headline claim of 'highest text rendering accuracy' is measured by OCR accuracy in the MARIO-Eval setup (Sec. 4.2), a standard metric from prior work [9,10]. Crucially, Fig. 8 states that the OCR* shown in the iteration analysis is 'computed by OCR models in Type-R which is distinct from the OCR model for evaluation,' so the reported accuracy is not literally the same function that the loop optimizes. Section 4.7 notes that AnyText is trained with a Paddle-based loss and that Paddle gives the best final OCR, which would be a real circularity risk if the evaluation OCR were Paddle or a close relative; however, the paper does not state that, and the human user study (Table 2) and FID/CLIP provide additional independent axes, even though the user study does not test word-level legibility. The paper's own Limitation section concedes OCR failures, and Appendix F honestly labels internal counts as 'pseudo-errors.' Self-citations (e.g., ref [29] by two co-authors) appear only in related work and are not load-bearing. No prediction in the paper reduces to its input by construction; the only substantive weakness is that the evaluation OCR is not named, which is a reproducibility gap rather than a demonstrated circularity.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central contribution depends on a chain of external models and benchmark assumptions; these are enumerated rather than hidden. No new physical or conceptual entities are introduced; the system is a composition of existing components, so invented_entities is empty.

free parameters (3)
  • theta (small-text height filter) = 4% of image height
    Grid-searched on 500 MARIO-LAION validation prompts (Supplementary Table 7); filters out detected text regions below this height before word matching.
  • padding distance constant in optimal transport matching = not reported
    In Section 3.1, a constant distance is assigned to word-padding matches but its value is not stated; this affects whether missing or surplus words are flagged as errors.
  • t_max (maximum typo correction iterations) = 10
    Set by hand in Section 4.3; controls how many AnyText rewrite attempts are allowed before the loop gives up.
assumptions (5)
  • domain assumption OCR recognition is a valid ground-truth for text rendering accuracy.
    Pipeline error detection in Section 3.1 and Algorithm 1 treat OCR readings as ground truth to decide what to erase and when to stop correcting; Section 5 concedes OCR can fail.
  • domain assumption Prompt text W is the complete and only desired text set.
    The matching in Section 3.1 treats any detected word not matching a prompt word as an error and erases it. This holds by construction on MARIO-Eval, but real posters often include unrequested text.
  • domain assumption Words are separated by spaces and reading order is irrelevant.
    Type-R decomposes text into words by spaces and does not preserve reading order; acknowledged in Section 5. Multi-word or vertically stacked text layouts can be broken.
  • domain assumption External models (GPT-4o, AnyText, LaMa) perform their subtasks well enough.
    Layout regeneration in Section 3.3 relies on GPT-4o JSON output; typo correction in Section 3.4 relies on AnyText. Supplementary F and Section 5 note VLM corruption and unstable outlines.
  • domain assumption Inpainting and text editing can erase or rewrite a localized word region without harming the rest of the image.
    Text erasing (Section 3.2) and typo correction (Section 3.4) composite edited regions back into the original; the claim about maintaining image quality depends on this compositing being invisible.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Type-R: Automatically Retouching Typos for Text-to-Image Generation." pith.science (2026). https://pith.science/paper/453INXJZ

@misc{pith2026241118159,
  author       = {Pith},
  title        = {Pith review of: Type-R: Automatically Retouching Typos for Text-to-Image Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/453INXJZ}},
  note         = {Machine review of arXiv:2411.18159}
}
read the original abstract

While recent text-to-image models can generate photorealistic images from text prompts that reflect detailed instructions, they still face significant challenges in accurately rendering words in the image. In this paper, we propose to retouch erroneous text renderings in the post-processing pipeline. Our approach, called Type-R, identifies typographical errors in the generated image, erases the erroneous text, regenerates text boxes for missing words, and finally corrects typos in the rendered words. Through extensive experiments, we show that Type-R, in combination with the latest text-to-image models such as Stable Diffusion or Flux, achieves the highest text rendering accuracy while maintaining image quality and also outperforms text-focused generation baselines in terms of balancing text accuracy and image quality.

Figures

Figures reproduced from arXiv: 2411.18159 by the authors.

Figure 1
Figure 1. Type-R automatically corrects typographic errors in an image generated from any text-to-image model without sacrificing their [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the Type-R pipeline. Type-R automatically detects errors, erases unintended texts, inserts missing words, and [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Comparisons of generated images by Flux, Flux w/ Type-R, and TextDiffuser models. The left column shows prompts, and the [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (11 more)
Figure 5
Figure 5. Figure 5: Plot of the relation between OCR accuracy and graphic [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Examples of generated images through Type-R. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Generated images from long prompts by Flux, Flux w/ Type-R and TextDiffuser-2. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Relationship in the typo correction iteration and OCR* [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: presents images generated by Type-R along with their graphic design scores from both GPT and human eval￾uations. We observe that GPT and human evaluations share common trends and unique tendencies. The upper row highlights common trends. GPT and human evaluators as￾sig…
Figure 10
Figure 10. Figure 10: Plots for graphic design scores by GPT and Human. [PITH_FULL_IMAGE:figures/full_fig_p010_10.png]
Figure 11
Figure 11. Figure 11: Examples of typo correction across text editing methods. [PITH_FULL_IMAGE:figures/full_fig_p011_11.png]
Figure 12
Figure 12. Figure 12: Typo retouched images by Type-R across text-to-image models. [PITH_FULL_IMAGE:figures/full_fig_p012_12.png]
Figure 13
Figure 13. Figure 13: The plots of capacity for handling texts through Type [PITH_FULL_IMAGE:figures/full_fig_p013_13.png]
Figure 14
Figure 14. Figure 14: Examples of results from Type-R using rough layout [PITH_FULL_IMAGE:figures/full_fig_p013_14.png]
Figure 15
Figure 15. Figure 15: Examples of generated images through Type-R. [PITH_FULL_IMAGE:figures/full_fig_p017_15.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 44 canonical work pages

  1. [1]

    https://openai.com/dall-e-3

    DALL-E 3. https://openai.com/dall-e-3 . Ac- cessed: 2024-03-15. 2, 6, 11

    work page 2024

  2. [2]

    https://github.com/deep-floyd/ IF

    DeepFloyd IF. https://github.com/deep-floyd/ IF. Accessed: 2024-11-14. 2

    work page 2024

  3. [3]

    https://blackforestlabs.ai/ announcing - black - forest - labs/

    FLUX.1 [dev]. https://blackforestlabs.ai/ announcing - black - forest - labs/. Accessed: 2024-11-14. 1, 2, 4, 5, 6, 7, 11

    work page 2024

  4. [4]

    https://openai.com/index/gpt- 4o- system-card/

    GPT-4o. https://openai.com/index/gpt- 4o- system-card/. Accessed: 2024-11-14. 3, 4, 13

    work page 2024

  5. [5]

    https : / / github

    PaddlePaddle. https : / / github . com / PaddlePaddle / PaddleOCR / blob / release / 2.7/doc/doc_ch/PP-OCRv4_introduction.md . 4, 7

    work page

  6. [6]

    https://huggingface.co/ stabilityai/stable- diffusion- 3.5- large

    Stable Diffusion 3.5. https://huggingface.co/ stabilityai/stable- diffusion- 3.5- large . Accessed: 2024-11-8. 1, 6, 12

    work page 2024

  7. [7]

    What is wrong with scene text recognition model comparisons? dataset and model analysis

    Jeonghun Baek, Geewook Kim, Junyeop Lee, Sungrae Park, Dongyoon Han, Sangdoo Yun, Seong Joon Oh, and Hwal- suk Lee. What is wrong with scene text recognition model comparisons? dataset and model analysis. In ICCV, 2019. 7

    work page 2019

  8. [8]

    Character region awareness for text de- tection

    Youngmin Baek, Bado Lee, Dongyoon Han, Sangdoo Yun, and Hwalsuk Lee. Character region awareness for text de- tection. In CVPR, 2019. 7

    work page 2019

Show all 55 references
  1. [9]

    Textdiffuser: Diffusion models as text painters

    Jingye Chen, Yupan Huang, Tengchao Lv, Lei Cui, Qifeng 14 Chen, and Furu Wei. Textdiffuser: Diffusion models as text painters. In NeurIPS, 2023. 2, 4, 5, 6

    2023

  2. [10]

    Textdiffuser-2: Unleashing the power of language models for text rendering

    Jingye Chen, Yupan Huang, Tengchao Lv, Lei Cui, Qifeng Chen, and Furu Wei. Textdiffuser-2: Unleashing the power of language models for text rendering. In ECCV, 2024. 2, 4, 5, 6, 9

    2024

  3. [11]

    Scaling recti- fied flow transformers for high-resolution image synthesis

    Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas M ¨uller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, et al. Scaling recti- fied flow transformers for high-resolution image synthesis. In ICML, 2024. 2, 4, 5, 6, 7, 12

    2024

  4. [12]

    Layoutgpt: Compositional visual plan- ning and generation with large language models

    Weixi Feng, Wanrong Zhu, Tsu-jui Fu, Varun Jampani, Ar- jun Akula, Xuehai He, Sugato Basu, Xin Eric Wang, and William Yang Wang. Layoutgpt: Compositional visual plan- ning and generation with large language models. InNeurIPS,

  5. [13]

    Gans trained by a two time-scale update rule converge to a local nash equilib- rium

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule converge to a local nash equilib- rium. arXiv preprint arXiv:1706.08500, 2017. 4

    2017 arXiv

  6. [14]

    Improving diffusion models for scene text editing with dual encoders

    Jiabao Ji, Guanhua Zhang, Zhaowen Wang, Bairu Hou, Zhifei Zhang, Brian Price, and Shiyu Chang. Improving diffusion models for scene text editing with dual encoders. arXiv preprint arXiv:2304.05568, 2023. 2

    2023 arXiv

  7. [15]

    Cole: A hierarchical generation frame- work for graphic design

    Peidong Jia, Chenxuan Li, Zeyu Liu, Yichao Shen, Xingru Chen, Yuhui Yuan, Yinglin Zheng, Dong Chen, Ji Li, Xi- aodong Xie, et al. Cole: A hierarchical generation frame- work for graphic design. arXiv preprint arXiv:2311.16974,

    arXiv

  8. [16]

    Refining text-to-image generation: Towards accurate training-free glyph-enhanced image gen- eration

    Sanyam Lakhanpal, Shivang Chopra, Vinija Jain, Aman Chadha, and Man Luo. Refining text-to-image generation: Towards accurate training-free glyph-enhanced image gen- eration. arXiv preprint arXiv:2403.16422, 2024. 2

    2024 arXiv

  9. [17]

    The surprisingly straight- forward scene text removal method with gated attention and region of interest generation: A comprehensive prominent model analysis

    Hyeonsu Lee and Chankyu Choi. The surprisingly straight- forward scene text removal method with gated attention and region of interest generation: A comprehensive prominent model analysis. In ECCV, 2022. 8

    2022

  10. [18]

    Joytype: A robust design for multilingual visual text creation

    Chao Li, Chen Jiang, Xiaolong Liu, Jun Zhao, and Guoxin Wang. Joytype: A robust design for multilingual visual text creation. arXiv preprint arXiv:2409.17524, 2024. 2

    2024 arXiv

  11. [19]

    Trocr: Transformer-based optical character recognition with pre-trained models

    Minghao Li, Tengchao Lv, Jingye Chen, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, and Furu Wei. Trocr: Transformer-based optical character recognition with pre-trained models. In AAAI, 2023. 7

    2023

  12. [20]

    Empowering backbone models for visual text generation with input gran- ularity control and glyph-aware training

    Wenbo Li, Guohao Li, Zhibin Lan, Xue Xu, Wanru Zhuang, Jiachen Liu, Xinyan Xiao, and Jinsong Su. Empowering backbone models for visual text generation with input gran- ularity control and glyph-aware training. arXiv preprint arXiv:2410.04439, 2024. 2

    arXiv

  13. [21]

    First creating backgrounds then rendering texts: A new paradigm for visual text blending

    Zhenhang Li, Yan Shu, Weichao Zeng, Dongbao Yang, and Yu Zhou. First creating backgrounds then rendering texts: A new paradigm for visual text blending. In ECAI, 2024. 2, 9, 10

    2024

  14. [22]

    Mask TextSpotter v3: Segmentation proposal net- work for robust scene text spotting

    Minghui Liao, Guan Pang, Jing Huang, Tal Hassner, and Xi- ang Bai. Mask TextSpotter v3: Segmentation proposal net- work for robust scene text spotting. In ECCV, 2020. 7

    2020

  15. [23]

    Layoutprompter: Awaken the design ability of large language models

    Jiawei Lin, Jiaqi Guo, Shizhao Sun, Zijiang Yang, Jian- Guang Lou, and Dongmei Zhang. Layoutprompter: Awaken the design ability of large language models. In NeurIPS,

  16. [24]

    Parrot captions teach clip to spot text

    Yiqi Lin, Conghui He, Alex Jinpeng Wang, Bin Wang, Wei- jia Li, and Mike Zheng Shou. Parrot captions teach clip to spot text. In ECCV, 2024. 6

    2024

  17. [25]

    Character-aware models improve visual text rendering

    Rosanne Liu, Dan Garrette, Chitwan Saharia, William Chan, Adam Roberts, Sharan Narang, Irina Blok, RJ Mical, Mo- hammad Norouzi, and Noah Constant. Character-aware models improve visual text rendering. In ACL, 2023. 1, 2

    2023

  18. [26]

    Glyph-byt5: A customized text encoder for accurate visual text rendering

    Zeyu Liu, Weicong Liang, Zhanhao Liang, Chong Luo, Ji Li, Gao Huang, and Yuhui Yuan. Glyph-byt5: A customized text encoder for accurate visual text rendering. In ECCV, 2024. 2

    2024

  19. [27]

    Glyph-byt5-v2: A strong aesthetic base- line for accurate multilingual visual text rendering

    Zeyu Liu, Weicong Liang, Yiming Zhao, Bohan Chen, Ji Li, and Yuhui Yuan. Glyph-byt5-v2: A strong aesthetic base- line for accurate multilingual visual text rendering. arXiv preprint arXiv:2406.10208, 2024. 2

    2024 arXiv

  20. [28]

    Glyphdraw: Seamlessly ren- dering text with intricate spatial structures in text-to-image generation

    Jian Ma, Mingjun Zhao, Chen Chen, Ruichen Wang, Di Niu, Haonan Lu, and Xiaodong Lin. Glyphdraw: Seamlessly ren- dering text with intricate spatial structures in text-to-image generation. arXiv preprint arXiv:2303.17870, 2023. 2

    2023 arXiv

  21. [29]

    Ty- pographic text generation with off-the-shelf diffusion model

    KhayTze Peong, Seiichi Uchida, and Daichi Haraguchi. Ty- pographic text generation with off-the-shelf diffusion model. In ICDAR, 2024. 2

    2024

  22. [30]

    Sdxl: Improving latent diffusion models for high-resolution image synthesis

    Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas M ¨uller, Joe Penna, and Robin Rombach. Sdxl: Improving latent diffusion models for high-resolution image synthesis. In ICLR, 2023. 1, 2

    2023

  23. [31]

    Exploring stroke-level modifi- cations for scene text editing

    Yadong Qu, Qingfeng Tan, Hongtao Xie, Jianjun Xu, Yuxin Wang, and Yongdong Zhang. Exploring stroke-level modifi- cations for scene text editing. In AAAI, 2023. 2, 10

    2023

  24. [32]

    Learn- ing transferable visual models from natural language super- vision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learn- ing transferable visual models from natural language super- vision. In ICML, 2021. 4

    2021

  25. [33]

    An empirical study of scaling law for ocr

    Miao Rang, Zhenni Bi, Chuanjian Liu, Yunhe Wang, and Kai Han. An empirical study of scaling law for ocr. In CVPR,

  26. [34]

    High-resolution image syn- thesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj¨orn Ommer. High-resolution image syn- thesis with latent diffusion models. In CVPR, 2022. 2

    2022

  27. [35]

    STEFANN: scene text editor using font adap- tive neural network

    Prasun Roy, Saumik Bhattacharya, Subhankar Ghosh, and Umapada Pal. STEFANN: scene text editor using font adap- tive neural network. In CVPR, 2020. 2

    2020

  28. [36]

    Photorealistic text-to-image diffusion models with deep language understanding

    Chitwan Saharia, William Chan, Saurabh Saxena, Lala Li, Jay Whang, Emily L Denton, Kamyar Ghasemipour, Raphael Gontijo Lopes, Burcu Karagol Ayan, Tim Salimans, et al. Photorealistic text-to-image diffusion models with deep language understanding. In NeurIPS, 2022. 2

    2022

  29. [37]

    Resolution-robust large mask inpainting with fourier convolutions

    Roman Suvorov, Elizaveta Logacheva, Anton Mashikhin, Anastasia Remizova, Arsenii Ashukha, Aleksei Silvestrov, Naejin Kong, Harshith Goka, Kiwoong Park, and Victor Lempitsky. Resolution-robust large mask inpainting with fourier convolutions. In WACV, 2022. 3, 8

    2022

  30. [38]

    Llama 2: Open foundation and fine-tuned chat models.arXiv preprint arXiv:2307.09288, 2023

    Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, 15 Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, et al. Llama 2: Open foundation and fine-tuned chat models.arXiv preprint arXiv:2307.09288, 2023. 4

    2023 arXiv

  31. [39]

    Anytext: Multilingual visual text gener- ation and editing

    Yuxiang Tuo, Wangmeng Xiang, Jun-Yan He, Yifeng Geng, and Xuansong Xie. Anytext: Multilingual visual text gener- ation and editing. In ICLR, 2023. 2, 4, 7, 8, 9, 10

    2023

  32. [40]

    Textmaster: Universal control- lable text edit

    Aoqiang Wang, Jian Wang, Zhenyu Yan, Wenxiang Shang, Ran Lin, and Zhao Zhang. Textmaster: Universal control- lable text edit. arXiv preprint arXiv:2410.09879, 2024

    2024

  33. [41]

    High fidelity scene text synthesis

    Yibin Wang, Weizhong Zhang, Jianwei Zheng, and Cheng Jin. High fidelity scene text synthesis. arXiv preprint arXiv:2405.14701, 2024. 2

    2024 arXiv

  34. [42]

    A prompt pattern catalog to enhance prompt engineering with chatgpt

    Jules White, Quchen Fu, Sam Hays, Michael Sandborn, Car- los Olea, Henry Gilbert, Ashraf Elnashar, Jesse Spencer- Smith, and Douglas C Schmidt. A prompt pattern catalog to enhance prompt engineering with chatgpt. arXiv preprint arXiv:2302.11382, 2023. 4

    2023 arXiv

  35. [43]

    Editing text in the wild

    Liang Wu, Chengquan Zhang, Jiaming Liu, Junyu Han, Jing- tuo Liu, Errui Ding, and Xiang Bai. Editing text in the wild. In ACM MM, 2018. 2

    2018

  36. [44]

    Self-correcting llm-controlled diffusion models

    Tsung-Han Wu, Long Lian, Joseph E Gonzalez, Boyi Li, and Trevor Darrell. Self-correcting llm-controlled diffusion models. In CVPR, 2024. 2

    2024

  37. [45]

    Mastering text-to-image dif- fusion: Recaptioning, planning, and generating with multi- modal llms

    Ling Yang, Zhaochen Yu, Chenlin Meng, Minkai Xu, Ste- fano Ermon, and CUI Bin. Mastering text-to-image dif- fusion: Recaptioning, planning, and generating with multi- modal llms. In ICML, 2024. 2

    2024

  38. [46]

    Swaptext: Image based texts transfer in scenes

    Qiangpeng Yang, Jun Huang, and Wei Lin. Swaptext: Image based texts transfer in scenes. InCVPR, pages 14700–14709,

  39. [47]

    Glyphcontrol: Glyph conditional control for visual text generation

    Yukang Yang, Dongnan Gui, Yuhui Yuan, Weicong Liang, Haisong Ding, Han Hu, and Kai Chen. Glyphcontrol: Glyph conditional control for visual text generation. In NeurIPS,

  40. [48]

    DeepSolo: Let transformer decoder with explicit points solo for text spot- ting

    Maoyuan Ye, Jing Zhang, Shanshan Zhao, Juhua Liu, Tongliang Liu, Bo Du, and Dacheng Tao. DeepSolo: Let transformer decoder with explicit points solo for text spot- ting. In CVPR, 2023. 4, 7

    2023

  41. [49]

    Hi-sam: Marrying segment anything model for hierarchical text segmentation

    Maoyuan Ye, Jing Zhang, Juhua Liu, Chenyu Liu, Baocai Yin, Cong Liu, Bo Du, and Dacheng Tao. Hi-sam: Marrying segment anything model for hierarchical text segmentation. arXiv preprint arXiv:2401.17904, 2024. 7

    2024 arXiv

  42. [50]

    TextCtrl: Diffusion-based scene text editing with prior guidance control

    Weichao Zeng, Yan Shu, Zhenhang Li, Dongbao Yang, and Yu Zhou. TextCtrl: Diffusion-based scene text editing with prior guidance control. In NeurIPS, 2024. 2, 10

    2024

  43. [51]

    Brush your text: Synthesize any scene text on im- ages via diffusion model

    Lingjun Zhang, Xinyuan Chen, Yaohui Wang, Yue Lu, and Yu Qiao. Brush your text: Synthesize any scene text on im- ages via diffusion model. In AAAI, 2024. 2

    2024

  44. [52]

    Layout-agnostic scene text image synthesis with diffusion models

    Qilong Zhangli, Jindong Jiang, Di Liu, Licheng Yu, Xi- aoliang Dai, Ankit Ramchandani, Guan Pang, Dimitris N Metaxas, and Praveen Krishnan. Layout-agnostic scene text image synthesis with diffusion models. In CVPR, 2024. 2

    2024

  45. [53]

    Udifftext: A unified frame- work for high-quality text synthesis in arbitrary images via character-aware diffusion models

    Yiming Zhao and Zhouhui Lian. Udifftext: A unified frame- work for high-quality text synthesis in arbitrary images via character-aware diffusion models. In ECCV, 2024. 2, 10

    2024

  46. [54]

    Harmonizing visual text comprehension and gen- eration

    Zhen Zhao, Jingqun Tang, Binghong Wu, Chunhui Lin, Shu Wei, Hao Liu, Xin Tan, Zhizhong Zhang, Can Huang, and Yuan Xie. Harmonizing visual text comprehension and gen- eration. In NeurIPS, 2024. 2

    2024

  47. [55]

    Visual text generation in the wild

    Yuanzhi Zhu, Jiawei Liu, Feiyu Gao, Wenyu Liu, Xinggang Wang, Peng Wang, Fei Huang, Cong Yao, and Zhibo Yang. Visual text generation in the wild. In ECCV, 2024. 2 16 Text-to-Image Text-to-Image Prompt: 'Save the Earth' Prompt: 'London Coins' A159 Lot 419 Birth of Prince Charle...

    2024

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.