- Rating
- 3
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 3
Read review
Paper summary
The paper proposed a VAE-free visual text editing and refinement framework. It generates text patches based on the conditions of the original images and pastes the text patches back into the original image for editing/refinement. To reduce visual artifacts introduced by JiT, the paper adopts uniform timestep sampling during continued training to reduce patch-grid artifacts and modifies the x-prediction sampling update to eliminate residual noise caused by incomplete final denoising. Additionally, the paper introduces the Text Refinement Bench for fine-grained text correction, glyph fidelity, and image preservation evaluation.
Critical strengths
- The paper demonstrates the effectiveness of VAE-free pixel-space generation for localized scene text editing, supported by large-scale training and extensive experiments.
- The paper shows that uniform timestep sampling can mitigate patch-boundary artifacts in pixel-space generation, with improvements demonstrated on both PixText and MiniT2I.
Critical weaknesses
-
PixText shares a similar task setting with existing methods that perform localized text generation and editing, such as UTDesign. The authors should clarify the differences between these approaches and consider including representative methods in the experimental comparison to better demonstrate PixText's advantages.
-
Although the architecture supports variable-sized crops, the paper does not clearly establish how text scale affects editing quality and computational cost. The inference-time resizing policy, supported crop-size range, and treatment of large text regions need clarification. Results broken down by text size would help determine whether the reported advantages hold across scales.
Questions for authors
-
Could the authors include comparisons with existing localized text editing methods, such as UTDesign, or clarify the differences in their task settings and input requirements?
-
Although CCPE supports variable-sized crops, how does PixText perform across different text scales, particularly for large characters or words? A scale-wise evaluation of editing accuracy and computational cost would be helpful.
-
How are the erroneous text regions and target glyphs obtained in real-world applications? Given that OCR may fail to detect subtle glyph errors, does the proposed refinement pipeline require manual intervention?
5 comments
Response to Reviewer Uk3T(part2)
Critical Weakness 2 and Question 2: Accuracy and computational cost across text scales
Reviewer concern. Although CCPE supports variable-sized crops, how does PixText perform across different text scales, particularly for large characters or words? A scale-wise evaluation of editing accuracy and computational cost would be helpful.
Response. We thank the reviewer for this suggestion. We evaluate all seven methods on Text Refinement Bench by target-box short side. OCR is computed per target character, while inference cost is computed per image using the largest target box in that image. Each cell below reports single-character OCR accuracy (%) / amortized inference cost (GPU-s/img). Batch latency is amortized across images under the original evaluation configurations.
For UTDesign, we use the official v1.0 editing weights with 28 steps and CFG 5. Up to three neighboring glyphs from the same line are added as style references: 2,141 targets receive additional references, while 497 fall back to a single reference.
Table 2. OCR accuracy and amortized inference cost by target-box short side
| Short side (px) | AnyText | AnyText2 | FLUX-Text | TextFlux | UTDesign | PixText Teacher | PixText MeanFlow |
|---|---|---|---|---|---|---|---|
| Overall | 25.6 / 4.307 | 48.4 / 8.868 | 52.3 / 16.420 | 28.3 / 14.142 | 83.4 / 5.157 | 94.7 / 4.514 | 94.6 / 0.942 |
| <16 | 0.0 / 4.219 | 1.7 / 8.688 | 1.2 / 16.417 | 0.0 / 14.121 | 46.5 / 5.034 | 75.1 / 4.500 | 76.3 / 0.964 |
| 16–<24 | 0.1 / 4.238 | 12.4 / 8.740 | 13.7 / 16.414 | 5.5 / 14.114 | 76.1 / 5.285 | 95.0 / 4.448 | 94.9 / 0.931 |
| 24–<32 | 8.1 / 4.305 | 44.8 / 8.884 | 54.9 / 16.420 | 20.7 / 14.170 | 86.4 / 6.811 | 97.1 / 4.527 | 96.4 / 0.943 |
| 32–<40 | 30.1 / 4.368 | 73.8 / 8.978 | 81.1 / 16.431 | 37.2 / 14.174 | 90.2 / 7.409 | 97.8 / 4.546 | 98.1 / 0.944 |
| 40–<48 | 54.4 / 4.431 | 87.6 / 9.112 | 83.6 / 16.422 | 46.0 / 14.161 | 92.0 / 4.489 | 97.8 / 4.540 | 97.8 / 0.938 |
| 48–<56 | 71.1 / 4.383 | 90.9 / 9.010 | 90.1 / 16.416 | 59.5 / 14.126 | 93.4 / 2.276 | 98.3 / 4.563 | 99.2 / 0.943 |
| 56–<64 | 85.3 / 4.347 | 93.3 / 8.878 | 93.3 / 16.431 | 64.0 / 14.155 | 97.3 / 3.910 | 93.3 / 4.512 | 92.0 / 0.947 |
| ≥64 | 86.2 / 4.223 | 86.2 / 8.703 | 90.2 / 16.414 | 76.0 / 14.103 | 94.9 / 3.186 | 92.0 / 4.497 | 92.4 / 0.944 |
Two trends emerge. First, PixText’s largest advantage is on small and medium text. Overall, Teacher and MeanFlow exceed UTDesign by 11.23 and 11.15 pp, respectively. Below 16 px, Teacher and MeanFlow achieve 75.1% and 76.3% OCR, compared with 46.5% for UTDesign and 1.7% for the strongest other baseline. At 16–<24 px, they reach 95.0% and 94.9%, compared with 76.1% for UTDesign and 13.7% for the strongest other baseline. PixText also remains ahead in every bin from 24 to 56 px, showing that its main advantage is concentrated where fine strokes are most difficult.
On larger text, PixText remains competitive with UTDesign. At 56–<64 px, UTDesign reaches 97.3%, while Teacher and MeanFlow reach 93.3% and 92.0%. At ≥64 px, UTDesign reaches 94.9%, compared with 92.4% for MeanFlow and 92.0% for Teacher, a difference of only 2.5 pp for MeanFlow. Thus, PixText combines its strongest gains on small and medium text with comparable performance on large characters.
Second, amortized cost remains nearly flat across scales. Teacher stays within 4.177–4.563 GPU-s/img, and MeanFlow stays within 0.931–0.964 GPU-s/img across every bin, including 0.944 GPU-s/img at ≥64 px. These measurements use the largest target per image and the original batch composition, so they describe the evaluated amortized-cost behavior rather than an isolated single-target scaling law.
We additionally test robustness beyond the main training resolution. Using the same PixText checkpoint and inference settings, we upsample the Text Refinement Bench inputs and target regions by 2× and compare performance at the original 1024 scale and at 2048 × 2048.
Table 3. Controlled 2× input-upscaling evaluation
| Input scale | OCR Acc ↑ | CJK-CLIP Pass ↑ |
|---|---|---|
| 1024 main image and targets | 93.86% | 81.54% |
| 2048 main image and targets | 94.88% | 85.56% |
| Improvement | +1.02 pp | +4.02 pp |
Although the main-image area during training is primarily capped at , the model successfully performs localized editing at . Under this controlled setting, OCR increases from 93.86% to 94.88%, and CJK-CLIP Pass increases from 81.54% to 85.56%.
Overall, the scale-wise results support the paper’s central motivation: PixText provides its largest gains on small and medium text, remains comparable on large characters, retains sub-one-GPU-second amortized MeanFlow inference across all tested size bins, and shows promising robustness when evaluated at 2048 resolution.
Response to Reviewer Uk3T(part3)
Question 3: Error detection, target glyph construction, and human involvement
Reviewer concern. How are the erroneous text regions and target glyphs obtained in real-world applications? Since OCR may fail to detect subtle glyph errors, does the proposed refinement pipeline require manual intervention?
Response. We thank the reviewer for raising this practical question. The default PixText pipeline does not require users to manually mark erroneous characters. The target text can be extracted from the user’s generation prompt or supplied by a prompt enhancer. The system automatically proposes repair regions and target text; users may optionally adjust them before generation, but this is not required.
In deployment, an OCR detection head first localizes text regions, and a recognition head produces initial transcriptions. We align these transcriptions with the target text to assign a target string to each region. A transcription mismatch triggers refinement. For character regions whose OCR transcription already matches the target character, CJK-CLIP additionally checks glyph fidelity against the target character. Regions below the character-specific CJK-CLIP threshold are also sent to PixText. This second stage targets cases where OCR reads the character correctly but its strokes are still malformed. Target glyph conditions are then rendered from the aligned target strings, rather than extracted from a reference image.
To evaluate whether the additional verification improves final refinement, we compare two pipelines on Text Refinement Bench: OCR-only error detection and OCR + CJK-CLIP verification. Both use Hungarian matching to assign OCR transcriptions to the target text and share the same PixText repair model. The target text in this experiment is constructed from dataset annotations.
We randomly sample 80 targets with originally incorrect glyphs and 80 with originally correct glyphs. Thirteen reviewers perform an A/B blind evaluation of the two pipelines, with final decisions determined by majority vote.
Table 4. A/B blind evaluation of OCR-only and OCR + CJK-CLIP pipelines
| Original target | Questions | OCR + CJK-CLIP wins | Ties | OCR baseline wins | OCR + CJK-CLIP GSB ↑ | OCR baseline GSB ↑ |
|---|---|---|---|---|---|---|
| Overall | 160 | 40 | 117 | 3 | 1.31 | 0.76 |
| Originally incorrect | 80 | 40 | 39 | 1 | 1.98 | 0.51 |
| Originally correct | 80 | 0 | 78 | 2 | 0.98 | 1.03 |
GSB is computed as , with wins and losses exchanged when scoring the OCR baseline. On originally incorrect glyphs, the CJK-CLIP pipeline wins 40 comparisons, ties 39, and loses only 1, giving a GSB of 1.98 versus 0.51 for OCR alone. On originally correct glyphs, 78 of 80 comparisons are ties and the remaining 2 favor the OCR baseline, indicating near-parity when no repair is needed. Thus, the additional verification improves final refinement on erroneous glyphs without materially degrading correct ones.
Response to Reviewer Uk3T(part1)
Critical Weakness 1 and Question 1: Direct comparison with UTDesign and differences in task settings
Response. We thank the reviewer for suggesting this comparison. UTDesign is a strong and representative localized text-editing method, and its flexible style-reference design makes it an important baseline for our comparison. We have added UTDesign to the related-work discussion and included a formal citation in the revision. We have also evaluated it directly on Text Refinement Bench using the same 1,034 images, 2,638 targets, target boxes, and scoring protocol. For UTDesign, we add up to three neighboring glyphs from the same line as style references.
Table 1. Direct comparison with UTDesign on Text Refinement Bench
Teacher improves over UTDesign by 11.23 pp OCR, 27.03 pp Pass, and 12.66 pp R@1. MeanFlow improves by 11.15 pp, 25.81 pp, and 12.28 pp, respectively. MeanFlow also uses only four sampling steps and achieves 0.942 GPU-s/img, compared with 28 steps and 5.157 GPU-s/img for UTDesign—an approximately 5.5× lower observed generation cost. MeanFlow has 0.954B parameters, while UTDesign’s transformer alone has approximately 1.108B parameters, excluding its visual encoders, VAE, and LaMa background-repair module.
The scale-wise results clarify where the methods differ. PixText leads overall and on small and medium text. On the combined ≥64 px bin, UTDesign reaches 94.9% OCR, compared with 92.4% for MeanFlow and 92.0% for Teacher, indicating that PixText remains comparable on large characters while providing its largest gains on small and medium text.
The two methods share compatible high-level inputs: both can edit using the original image, target regions, and target text. UTDesign’s style references are also extracted from the original image, so it does not require an additional reference image or ground-truth glyph crop. Their task goals, however, differ. UTDesign targets stylized text replacement and generation in graphic design, including layout adjustment. PixText targets localized error correction after image generation: it repairs a small number of incorrect characters or malformed strokes at their original locations, especially small and structurally complex characters, while preserving the existing layout and image content.
Their conditioning and generation mechanisms also differ. PixText directly uses rendered target glyphs as the content condition, eliminating the need for a separate text encoder, and infers font style and appearance from the surrounding image content. Group JiT uses CCPE to align the source image, target glyphs, and repair regions in one coordinate system, then jointly generates RGB patches containing both the corrected characters and their local background before pasting them back. Both training and inference operate in pixel space without a VAE.
UTDesign instead uses DINOv2 to encode the target glyph and CLIP to encode style-reference crops from the original image. It generates glyphs in latent space, decodes them with a transparent VAE into independently editable RGBA text layers, and composites them onto a background after removing the original text. Thus, UTDesign offers reusable and independently editable text layers, whereas PixText directly optimizes localized correction accuracy, fine-stroke preservation, and fast end-to-end repair.