Pith. sign in

REVIEW 4 major objections 5 minor 44 references

WordCraft: Interactive Artistic Typography with Attention Awareness and Noise Blending

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read WordCraft shows that a diffusion model can be made interactive for artistic typography by adding a training-free regional attention mask and a noise-blending rule, enabling precise local edits and iterative refinement from open-ended…

desk verdict A promising interactive typography system whose central attention-mask equation is mathematically wrong as written. read the letter →

arxiv 2507.09573 v1 pith:3N3VD5LV submitted 2025-07-13 cs.CV

Zhe Wang , Jingbo Zhang , Tianyi Wei , Wanchao Su , Can Wang This is my paper
classification cs.CV
keywords artistictypographydiffusionmodelsregionalattentionnoiseblendinginteractiveeditinglargelanguagetext-to-imagegenerationmulti-charactersynthesis
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

WordCraft is an interactive system that generates stylized typography from free-form text prompts and supports fine-grained, iterable edits. The paper's central claim is that two training-free additions to a diffusion model—a regional attention mask and a noise-blending step—give a text-to-image generator precise, independent control over different regions of a letter or word while keeping unedited areas intact. The authors argue this fills a practical gap in prior artistic-typography methods, which either stylize a whole glyph globally or support only a fixed subject/background split and cannot be refined. If the claim holds, designers can localize styles to specific radicals or characters, refine repeatedly, and work across many scripts without retraining the model.

What carries the argument

The central objects are (1) a regional attention mask inserted into the multi-modality attention of the FLUX transformer, whose blocks are set to 1 only for token pairs within the same user-defined region or between that region and its corresponding text or depth tokens, so different regions do not leak information into one another; (2) a noise-blending rule that, at each denoising step, keeps the original predicted noise outside the masked regions and substitutes region-specific newly predicted noise inside, so only the masked areas change; and (3) an LLM-based semantic parser that turns free-form user text into a base prompt and per-region prompts. The input character is parameterized by spline fitting and differentiable rasterization into a clean glyph image, from which a pretrained depth estimator derives the conditioning depth map.

What would settle it

Run an OCR or character-recognition model on the generated images for out-of-domain scripts (Japanese, Korean, Arabic numerals) and compare the recognition accuracy against the input glyphs; if accuracy is near chance or drops sharply relative to in-domain scripts, depth conditioning is not preserving legibility. Alternatively, measure the structural similarity (SSIM) between the unedited regions before and after a regional edit; a substantial drop would contradict the claim that noise blending leaves those regions unchanged.

Watch

Extended reading notes

Core claim

On its own terms, the paper claims that a FLUX diffusion model fine-tuned on a small typography dataset can be made interactive by masking the attention between image tokens and text/depth tokens according to user-drawn regions, and by blending newly predicted noise with the original noise only inside those regions during denoising. Each region of a glyph is then guided by its own text description while the depth map of the plain character keeps the geometry faithful, and repeated local edits are possible because the unchanged regions keep their original noise. A large language model converts abstract or compound user descriptions into a structured base prompt plus per-region prompts. The reported evaluation shows higher CLIP image-text similarity, lower FID, and higher user-study ratings (aesthetic quality, text alignment, legibility) than the VitaGlyph and MetaDesigner baselines, along with qualitative generalization to Japanese, Korean, Arabic numerals, and music symbols.

Load-bearing premise

The load-bearing premise is that depth maps derived from plain, untextured glyph images, together with a LoRA fine-tune on 355 words, are enough to preserve character legibility and structure across regions and across out-of-domain scripts; the paper reports no quantitative measurement of structural preservation.

Editorial extensions

If this is right

  • Users can assign different styles to different radicals, letters, or sub-regions of a word and refine a single region repeatedly without redrawing the rest.
  • The same pipeline works for single- and multi-character inputs and for scripts beyond the training set, such as Japanese, Korean, Arabic numerals, and music symbols.
  • The method removes the need for paired local-editing training data: regional editing and refinement are training-free given a base model fine-tuned for global generation.
  • Compared with prior two-region or single-prompt systems, the regional attention mask reduces semantic confusion between regions, leading to higher text-image alignment (CLIP) and visual realism (FID).
  • The LLM parsing makes abstract, compound, and open-ended descriptions usable as-is, so non-expert users can drive the system without formatting constraints.

Reading between the lines

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

  • If depth conditioning is indeed what preserves legibility, the method's ceiling is set by the depth estimator's ability to read flat, untextured glyphs; scripts with very different stroke topology may need more training data than the 355 words used here.
  • The noise-blending rule implies a trade-off: the more denoising steps already taken before an edit, the stronger the anchor to the original image, so very late edits may not fully adopt the new prompt; this is testable by varying the edit timing.
  • The same regional attention mask could be applied to other diffusion transformer-based generators beyond typography, enabling interactive local editing of logos, diagrams, or mixed-media images.
  • A direct comparison of character recognition accuracy (OCR) on edited outputs would strengthen the legibility claim, which the paper only measures through a subjective user study.
Share X Bluesky LinkedIn Reddit HN

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. WordCraft proposes an interactive artistic typography system built on FLUX.1. The pipeline uses an LLM to decompose open-ended user prompts into structured global/regional prompts, parameterizes characters via FreeType/DiffVG, conditions on depth maps, and fine-tunes FLUX with LoRA on a small dataset. The two core technical contributions are a "regional attention" mask applied inside the transformer's multi-modal attention to separate regions, and a noise-blending strategy that replaces noise only in user-masked regions during iterative denoising. The paper reports CLIP/FID improvements over MetaDesigner and VitaGlyph, a 30-response user study, and extensive qualitative results across languages and scripts.

Significance. If the technical claims are correct, WordCraft would be a meaningful step toward interactive, localized typography generation: it targets real user needs (local edits, iterative refinement, open-ended prompts) that existing single-pass methods such as VitaGlyph and MetaDesigner do not address, and the qualitative figures show a usable interface and plausible multi-region outputs. The idea of region-specific attention in a diffusion transformer is timely, and the noise-blending formulation is simple and well-motivated. However, the paper's central formal definition of the regional attention mask is mathematically incomplete (multiplication by a 0/1 mask is not a valid hard attention mask), and the quantitative evaluation as reported is not sufficient to establish superiority. The contribution is therefore currently promising but not yet convincingly demonstrated.

major comments (4)
  1. [Section 3.2.2, Eq. (3)] The regional attention mask is written as an elementwise multiplication of the logit matrix by a 0/1 mask M. Under softmax, a blocked entry with logit 0 receives probability exp(0)/Σ exp(·), which can be substantial; for example, if the unblocked logit is 2 and the blocked logit is 0, the blocked key receives about 11.9% of the attention mass. This does not implement the "no attention interaction" claimed for the zero blocks in Eq. (8). The correct construction is an additive mask with -inf (or a large negative constant) on blocked entries. As written, the central mechanism of the paper does not provide independent regional control.
  2. [Section 3.2.2, Eq. (4)] The mask blocks for the base-prompt tokens T_b are never defined. The sequence in Eq. (3) is [X; T_b; T_1;...;T_N; D], but the block matrix in Eq. (4) only lists image, regional-text, and depth blocks; the rows/columns for T_b (e.g., M_{T_b2X}, M_{X2T_b}, M_{T_b2T_k}) are absent. Since attention is not stated to be causal in this formulation, the behavior of the base prompt is underspecified and could affect the regional-control claim. This formal gap must be addressed before the method is reproducible.
  3. [Section 4.2, Tables 1-2] The headline quantitative results are not statistically grounded. Table 1 reports CLIP-Score and FID without error bars, confidence intervals, or the number of seeds or samples, and the reference distribution for FID is not described. In addition, CLIP-Score is computed with the same prompts that condition the generation, so the metric is partially circular and the reported +2.23% advantage over VitaGlyph is difficult to interpret. The user study in Table 2 aggregates only 30 responses over 10 questions with no significance tests or variance reporting; the claim that the method "consistently outperforms" the baselines is not supported by the reported evidence.
  4. [Section 4.3 and Section 4.6] The ablations of the regional attention and noise blending are qualitative only, and the claimed generalization to out-of-domain scripts (Japanese, Korean, Arabic numerals, music symbols) is shown only in figures. There is no quantitative measurement of structural preservation, character legibility, or region-independence (e.g., how much the unedited region changes after a local edit). Without such measurements, the contribution of the regional attention mechanism is not isolated from the LoRA fine-tuning and depth conditioning.
minor comments (5)
  1. [Section 3.2.1] The text says "T5 text encoder [21]" but reference [21] is the SDXL paper; the T5 citation is missing.
  2. [Figure 5 caption] The caption says "without rational attention (w/o RA)" but should read "regional attention."
  3. [Section 3.2.2] In the paragraph explaining Eq. (4), the off-diagonal mask blocks are described as "the self-attention mask"; this appears to be a typo for "cross-attention mask."
  4. [Section 4.1] The spellings "OminiControl" and "multi-model attention" should be "Ominicontrol" (or "OmniControl") and "multi-modal attention."
  5. [General terminology] The regional attention is called "training-free," but the overall system includes LoRA fine-tuning on a collected dataset; the paper should clarify the precise sense in which the attention mechanism itself requires no additional training.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivation chain is self-contained, and the self-citations are either inspirational or non-load-bearing.

full rationale

I walked the paper's claimed derivation chain: the LLM-based semantic parsing (Eq. 1), character parameterization, depth-conditioned FLUX generation, regional attention (Eqs. 3-8), and noise blending (Eq. 9). None of these steps reduces to its own inputs by construction. The regional attention mask is defined independently of the evaluation metrics; noise blending is a spatial mixing rule that is definitional but is not presented as a prediction. The CLIP score does measure alignment to the same prompts used for conditioning, which is a mild benchmark dependence shared by all text-to-image methods, but this is not a logical circularity: the generation is still a nontrivial function of the prompts, regions, and depth. The paper cites two prior works by co-authors ([30] Chat2Layout for LLM decomposition and [39] HumanRef for attention-mask inspiration), but neither citation is used as a uniqueness theorem, as a substitute for evidence, or to forbid alternatives; the methods are implemented and evaluated independently. The skeptical concern about Eq. 3 (multiplying logits by a 0/1 mask instead of adding -inf) is a mathematical correctness issue in the attention formulation, not a circularity, because it does not make the output equivalent to the input by definition. No self-citation chain forces the central result, and no fitted parameter is renamed as a prediction. I therefore find no significant circularity and assign score 0.

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

The central claim rests on a fine-tuned diffusion backbone and several unvalidated modeling choices: depth conditioning, attention-mask transfer, and LLM parsing. The paper does not isolate or quantify the contribution of each choice, and the unreported hyperparameters (t_step, guidance scale) make the method under-specified.

free parameters (3)
  • t_step (noise blending timestep)
    Eq. 9 blends noise at a timestep t, but the paper never states which t is used, how the original noise is extracted, or how many denoising steps are run.
  • guidance_scale (classifier-free guidance)
    The text conditioning strength is a key hyperparameter for FLUX generation, yet no value is reported; the behavior of regional attention and noise blending will depend on it.
  • LoRA fine-tuning hyperparameters = 80,000 steps, Prodigy lr=1.0
    The central generation quality depends on a fine-tuned model; the LoRA rank and target modules are not reported, and the fine-tuned weights are not released.
assumptions (3)
  • domain assumption FLUX.1 fine-tuned with LoRA on the authors' 355-word dataset and conditioned on a Depth Anything depth map preserves the structure and legibility of input glyphs.
    Section 3.2 uses the depth map 'to ensure that the fundamental structure of the generated stylized font remains consistent with the input character.' No quantitative structural-preservation metric is reported.
  • domain assumption The attention mask from HumanRef transfers to FLUX's multi-modal attention and exactly isolates regions so that regional prompts do not leak.
    Section 3.2.2 adapts the mask from [39] to FLUX. The paper shows qualitative ablations but no test of whether zeroing cross-region attention causes boundary or coherence artifacts in FLUX.
  • domain assumption GPT-4 reliably converts arbitrary user descriptions and region selections into valid JSON prompts and masks for the diffusion model.
    Section 3.1.1 defines P,c = GPT(Q, I_region). No prompt template, validation, or failure-rate analysis is provided, so the parser's reliability is assumed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of WordCraft: Interactive Artistic Typography with Attention Awareness and Noise Blending." pith.science (2026). https://pith.science/paper/3N3VD5LV

@misc{pith2026250709573,
  author       = {Pith},
  title        = {Pith review of: WordCraft: Interactive Artistic Typography with Attention Awareness and Noise Blending},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3N3VD5LV}},
  note         = {Machine review of arXiv:2507.09573}
}
read the original abstract

Artistic typography aims to stylize input characters with visual effects that are both creative and legible. Traditional approaches rely heavily on manual design, while recent generative models, particularly diffusion-based methods, have enabled automated character stylization. However, existing solutions remain limited in interactivity, lacking support for localized edits, iterative refinement, multi-character composition, and open-ended prompt interpretation. We introduce WordCraft, an interactive artistic typography system that integrates diffusion models to address these limitations. WordCraft features a training-free regional attention mechanism for precise, multi-region generation and a noise blending that supports continuous refinement without compromising visual quality. To support flexible, intent-driven generation, we incorporate a large language model to parse and structure both concrete and abstract user prompts. These components allow our framework to synthesize high-quality, stylized typography across single- and multi-character inputs across multiple languages, supporting diverse user-centered workflows. Our system significantly enhances interactivity in artistic typography synthesis, opening up creative possibilities for artists and designers.

Figures

Figures reproduced from arXiv: 2507.09573 by the authors.

Figure 1
Figure 1. WordCraft introduces an interactive text-driven artistic typography system that provides fine-grained control and high aesthetic quality for both single- and multi-character generation across multiple languages. The first row demonstrates the continuous regional editing process, with modified radicals highlighted in orange rectangles. The third row showcases the results of multi-regional editing, such as the charact… view at source ↗
Figure 2
Figure 2. Framework. Our method supports stylized generation, regional editing, and multi-step user interaction. latent representations and perform diffusion in this space, as demonstrated by models such as Imagen [25], DALLE￾2 [23], and LDM [24]. To enable conditional generation, ControlNet [40] was introduced to guide the denoising pro￾cess using external inputs such as segmentation maps or depth cues, inspiring a series of… view at source ↗
Figure 3
Figure 3. , we first employ GPT-4 as a knowledge engine to deeply understand users’ task-specific intentions and auto￾matically generate semantic prompts adapted to the three editing scenarios described above. This approach lowers the operational barrier for users and minimizes perceptual differences. In essence, users only need to provide a gen￾eral description Q of the desired font stylization and specify the intended editi… view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: Comparison. We compare our method with VitaGlyph and MetaDesigner. Unlike VitaGlyph, which supports only two regions (the subject and the surrounding area), and MetaDesigner, which relies on a single text prompts to describe each region, our method sup￾ports multi-regi…
Figure 5
Figure 5. Figure 5: Ablation Study. We compare our full method with three ablated variants: without the LLM component (w/o LLM), without rational attention (w/o RA), and without noise blending (w/o NB-LB and w/o NB-Inp) [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 7
Figure 7. Figure 7 [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Multi-Regional Generation Results. Our method sup￾ports multi-regional generation by drawing various masks [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Comparison. We compare our method with GPT4o, demonstrating superior semantic consistency in multi-regional generation and better preservation of unedited regions during con￾tinuous editing. cally, it surpasses VitaGlyph by a notable margin in CLIP score (+2.23%), indi…
Figure 10
Figure 10. Figure 10: Multi-regional Generation and Continuous Editing for Multi-character. Both our multi-regional attention and noise blending techniques effectively support multi-character scenarios [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]
Figure 11
Figure 11. Figure 11: User Interface. We design a user interface for our method. 4.3. Ablation Study To assess each technical component in our proposed method, we conduct an ablation study by systematically re￾moving key modules and measuring their impact on overall performance. w/o LLM. I…
Figure 12
Figure 12. Figure 12: Global Generation. Generation on music notations, Arabic numerals, Japanese and Korean characters. 0 1 "base_prompt": "Korean food style with vibrant ingredients", "background_prompt": "plain background", "regional_prompt": "0": "a leaf of spicy kimchi with red chili …
Figure 13
Figure 13. Figure 13 [PITH_FULL_IMAGE:figures/full_fig_p010_13.png]
Figure 14
Figure 14. Figure 14: Results with Diverse Text Prompts. Our method supports diverse text descriptions and for the same text prompt, we can generate various results. for additional data or model retraining. 4.4. Visual Results In [PITH_FULL_IMAGE:figures/full_fig_p011_14.png]
Figure 15
Figure 15. Figure 15: Interactive Support. Our method support long term user continuous editing. More Results on Multi-regional Generation. In [PITH_FULL_IMAGE:figures/full_fig_p012_15.png]
Figure 16
Figure 16. Figure 16: Limitation. Our method may fail when the text de￾scription are very fine-grained and detailed, the results may not be consistent with the semantics. and abstract visual effects. In summary, WordCraft makes significant strides in artistic typography by enhancing user i…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

44 extracted references · 26 canonical work pages

  1. [21]

    Sdxl: Improving latent diffusion mod- els for high-resolution image synthesis.arXiv preprint arXiv:2307.01952, 2023

    Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas M ¨uller, Joe Penna, and Robin Rombach. Sdxl: Improving latent diffusion mod- els for high-resolution image synthesis.arXiv preprint arXiv:2307.01952, 2023. 4

    arXiv 2023

  2. [1]

    Large language models for mathematical reasoning: Progresses and challenges

    Janice Ahn, Rishu Verma, Renze Lou, Di Liu, Rui Zhang, and Wenpeng Yin. Large language models for mathematical reasoning: Progresses and challenges. In18th Conference of the European Chapter of the Association for Computa- tional Linguistics, EACL 2024-Student Research Workshop, SRW 2024, pages 225–237. Association for Computational Linguistics (ACL), 2024. 3

    work page 2024

  3. [2]

    Multi-content gan for few-shot font style transfer

    Samaneh Azadi, Matthew Fisher, Vladimir G Kim, Zhaowen Wang, Eli Shechtman, and Trevor Darrell. Multi-content gan for few-shot font style transfer. InProceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 7564–7573, 2018. 2

    work page 2018

  4. [3]

    Loosec- ontrol: Lifting controlnet for generalized depth conditioning

    Shariq Farooq Bhat, Niloy Mitra, and Peter Wonka. Loosec- ontrol: Lifting controlnet for generalized depth conditioning. InACM SIGGRAPH 2024 Conference Papers, pages 1–11,

    2024

  5. [4]

    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. InForty-first international conference on machine learning,

  6. [5]

    Vitaglyph: Vital- izing artistic typography with flexible dual-branch diffusion models.arXiv preprint arXiv:2410.01738, 2024

    Kailai Feng, Yabo Zhang, Haodong Yu, Zhilong Ji, Jinfeng Bai, Hongzhi Zhang, and Wangmeng Zuo. Vitaglyph: Vital- izing artistic typography with flexible dual-branch diffusion models.arXiv preprint arXiv:2410.01738, 2024. 2, 7

    arXiv 2024

  7. [6]

    Neural trans- formation fields for arbitrary-styled font generation

    Bin Fu, Junjun He, Jianjun Wang, and Yu Qiao. Neural trans- formation fields for arbitrary-styled font generation. InPro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 22438–22447, 2023. 2

    work page 2023

  8. [7]

    Wordart designer: User-driven artistic typography synthesis using large language models

    Jun-Yan He, Zhi-Qi Cheng, Chenyang Li, Jingdong Sun, Wangmeng Xiang, Xianhui Lin, Xiaoyang Kang, Zengke Jin, Yusen Hu, Bin Luo, et al. Wordart designer: User-driven artistic typography synthesis using large language models. InProceedings of the 2023 Conference on Empirical Meth- ods in Natural Language Processing: Industry Track, pages 223–232, 2023. 2, 3

    work page 2023

Show all 44 references
  1. [8]

    Metadesigner: Advancing artistic typography through ai-driven, user-centric, and multilingual wordart synthesis

    Jun-Yan He, Zhi-Qi Cheng, Chenyang Li, Jingdong Sun, Qi He, Wangmeng Xiang, Hanyuan Chen, Jin-Peng Lan, Xian- hui Lin, Bin Luo, et al. Metadesigner: Advancing artistic typography through ai-driven, user-centric, and multilingual wordart synthesis. InThe Thirteenth Internationa...

    2025

  2. [9]

    Gans trained by a two time-scale update rule converge to a local nash equilib- rium.Advances in neural information processing systems, 30, 2017

    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.Advances in neural information processing systems, 30, 2017. 6

    2017

  3. [10]

    Word-as-image for semantic typography.ACM Transactions on Graphics (TOG), 42(4): 1–11, 2023

    Shir Iluz, Yael Vinker, Amir Hertz, Daniel Berio, Daniel Cohen-Or, and Ariel Shamir. Word-as-image for semantic typography.ACM Transactions on Graphics (TOG), 42(4): 1–11, 2023. 2

    2023

  4. [11]

    Image-to-image translation with conditional adver- sarial networks

    Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, and Alexei A Efros. Image-to-image translation with conditional adver- sarial networks. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 1125–1134,

  5. [12]

    Neural style transfer: A review

    Yongcheng Jing, Yezhou Yang, Zunlei Feng, Jingwen Ye, Yizhou Yu, and Mingli Song. Neural style transfer: A review. IEEE transactions on visualization and computer graphics, 26(11):3365–3385, 2019. 2

    2019

  6. [13]

    Flux.https://github.com/ black-forest-labs/flux, 2024

    Black Forest Labs. Flux.https://github.com/ black-forest-labs/flux, 2024. 4, 6

    2024

  7. [14]

    Differentiable vector graphics rasterization for editing and learning.ACM Transactions on Graphics (TOG), 39(6):1–15, 2020

    Tzu-Mao Li, Michal Luk ´aˇc, Micha ¨el Gharbi, and Jonathan Ragan-Kelley. Differentiable vector graphics rasterization for editing and learning.ACM Transactions on Graphics (TOG), 39(6):1–15, 2020. 4

    2020

  8. [15]

    Smartcontrol: Enhancing controlnet for handling rough visual conditions

    Xiaoyu Liu, Yuxiang Wei, Ming Liu, Xianhui Lin, Peiran Ren, Xuansong Xie, and Wangmeng Zuo. Smartcontrol: Enhancing controlnet for handling rough visual conditions. InEuropean Conference on Computer Vision, pages 1–17. Springer, 2024. 3

    2024

  9. [16]

    Intelligent typography: Artistic text style transfer for complex texture and structure.IEEE Transac- tions on Multimedia, 25:6485–6498, 2022

    Wendong Mao, Shuai Yang, Huihong Shi, Jiaying Liu, and Zhongfeng Wang. Intelligent typography: Artistic text style transfer for complex texture and structure.IEEE Transac- tions on Multimedia, 25:6485–6498, 2022. 2

    2022

  10. [17]

    Prodigy: An expeditiously adaptive parameter-free learner

    Konstantin Mishchenko and Aaron Defazio. Prodigy: An expeditiously adaptive parameter-free learner. InForty-first International Conference on Machine Learning, 2024. 6

    2024

  11. [18]

    Fontstudio: shape- adaptive diffusion model for coherent and consistent font ef- fect generation

    Xinzhi Mu, Li Chen, Bohan Chen, Shuyang Gu, Jianmin Bao, Dong Chen, Ji Li, and Yuhui Yuan. Fontstudio: shape- adaptive diffusion model for coherent and consistent font ef- fect generation. InEuropean Conference on Computer Vi- sion, pages 305–322. Springer, 2024. 3

    2024

  12. [19]

    Using an llm to help with code understanding

    Daye Nam, Andrew Macvean, Vincent Hellendoorn, Bogdan Vasilescu, and Brad Myers. Using an llm to help with code understanding. InProceedings of the IEEE/ACM 46th Inter- national Conference on Software Engineering, pages 1–13,

  13. [20]

    Multi- modal attention for speech emotion recognition.arXiv preprint arXiv:2009.04107, 2020

    Zexu Pan, Zhaojie Luo, Jichen Yang, and Haizhou Li. Multi- modal attention for speech emotion recognition.arXiv preprint arXiv:2009.04107, 2020. 4

    2009 arXiv

  14. [22]

    Learning transferable visual models from natural language supervi- sion

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. InInternational conference on machine learning, p...

    2021

  15. [23]

    Hierarchical text-conditional image gener- ation with clip latents.arXiv preprint arXiv:2204.06125, 1 (2):3, 2022

    Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen. Hierarchical text-conditional image gener- ation with clip latents.arXiv preprint arXiv:2204.06125, 1 (2):3, 2022. 3

    2022 arXiv

  16. [24]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image synthesis with latent diffusion models. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695, 2022. 2, 3

    2022

  17. [25]

    Photorealistic text-to-image diffusion models with deep language understanding.Advances in neural information processing systems, 35:36479–36494, 2022

    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.Advances in neural information ...

    2022

  18. [26]

    Ominicontrol: Minimal and uni- versal control for diffusion transformer.arXiv preprint arXiv:2411.15098, 3, 2024

    Zhenxiong Tan, Songhua Liu, Xingyi Yang, Qiaochu Xue, and Xinchao Wang. Ominicontrol: Minimal and uni- versal control for diffusion transformer.arXiv preprint arXiv:2411.15098, 3, 2024. 2, 6

    2024 arXiv

  19. [27]

    Ds-fusion: Artistic typography via discriminated and stylized diffusion

    Maham Tanveer, Yizhi Wang, Ali Mahdavi-Amiri, and Hao Zhang. Ds-fusion: Artistic typography via discriminated and stylized diffusion. InProceedings of the IEEE/CVF Inter- national Conference on Computer Vision, pages 374–384,

  20. [28]

    Freetype, 1996

    David Turner, Robert Wilhelm, and Werner Lemberg. Freetype, 1996. 4

    1996

  21. [29]

    Anything to glyph: artistic font synthe- sis via text-to-image diffusion model

    Changshuo Wang, Lei Wu, Xiaole Liu, Xiang Li, Lei Meng, and Xiangxu Meng. Anything to glyph: artistic font synthe- sis via text-to-image diffusion model. InSIGGRAPH Asia 2023 Conference Papers, pages 1–11, 2023. 2

    2023

  22. [30]

    Chat2layout: Interactive 3d furniture layout with a multimodal llm.arXiv preprint arXiv:2407.21333, 2024

    Can Wang, Hongliang Zhong, Menglei Chai, Mingming He, Dongdong Chen, and Jing Liao. Chat2layout: Interactive 3d furniture layout with a multimodal llm.arXiv preprint arXiv:2407.21333, 2024. 3

    2024 arXiv

  23. [31]

    Typography with decor: Intelligent text style transfer

    Wenjing Wang, Jiaying Liu, Shuai Yang, and Zongming Guo. Typography with decor: Intelligent text style transfer. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5889–5897, 2019. 2

    2019

  24. [32]

    Multi-modality cross attention network for image and sentence matching

    Xi Wei, Tianzhu Zhang, Yan Li, Yongdong Zhang, and Feng Wu. Multi-modality cross attention network for image and sentence matching. InProceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, pages 10941–10950, 2020. 6

    2020

  25. [33]

    Depth anything: Unleashing the power of large-scale unlabeled data

    Lihe Yang, Bingyi Kang, Zilong Huang, Xiaogang Xu, Jiashi Feng, and Hengshuang Zhao. Depth anything: Unleashing the power of large-scale unlabeled data. InCVPR, 2024. 4, 6

    2024

  26. [34]

    Tet-gan: Text effects transfer via stylization and destyl- ization

    Shuai Yang, Jiaying Liu, Wenjing Wang, and Zongming Guo. Tet-gan: Text effects transfer via stylization and destyl- ization. InProceedings of the AAAI Conference on Artificial Intelligence, pages 1238–1245, 2019. 2

    2019

  27. [35]

    Controllable artistic text style transfer via shape-matching gan

    Shuai Yang, Zhangyang Wang, Zhaowen Wang, Ning Xu, Jiaying Liu, and Zongming Guo. Controllable artistic text style transfer via shape-matching gan. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 4442–4451, 2019. 2

    2019

  28. [36]

    Shape- matching gan++: Scale controllable dynamic artistic text style transfer.IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(7):3807–3820, 2021

    Shuai Yang, Zhangyang Wang, and Jiaying Liu. Shape- matching gan++: Scale controllable dynamic artistic text style transfer.IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(7):3807–3820, 2021. 2

    2021

  29. [37]

    Glyphcontrol: glyph conditional control for visual text generation.Advances in Neural Information Processing Systems, 36:44050–44066,

    Yukang Yang, Dongnan Gui, Yuhui Yuan, Weicong Liang, Haisong Ding, Han Hu, and Kai Chen. Glyphcontrol: glyph conditional control for visual text generation.Advances in Neural Information Processing Systems, 36:44050–44066,

  30. [38]

    A survey on large language model (llm) security and privacy: The good, the bad, and the ugly

    Yifan Yao, Jinhao Duan, Kaidi Xu, Yuanfang Cai, Zhibo Sun, and Yue Zhang. A survey on large language model (llm) security and privacy: The good, the bad, and the ugly. High-Confidence Computing, page 100211, 2024. 3

    2024

  31. [39]

    Humanref: Single image to 3d human gen- eration via reference-guided diffusion

    Jingbo Zhang, Xiaoyu Li, Qi Zhang, Yanpei Cao, Ying Shan, and Jing Liao. Humanref: Single image to 3d human gen- eration via reference-guided diffusion. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1844–1854, 2024. 5

    2024

  32. [40]

    Adding conditional control to text-to-image diffusion models

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models. In Proceedings of the IEEE/CVF international conference on computer vision, pages 3836–3847, 2023. 2, 3

    2023

  33. [41]

    Uni-controlnet: All-in-one control to text-to-image diffusion models.Advances in Neural Information Processing Sys- tems, 36:11127–11150, 2023

    Shihao Zhao, Dongdong Chen, Yen-Chun Chen, Jianmin Bao, Shaozhe Hao, Lu Yuan, and Kwan-Yee K Wong. Uni-controlnet: All-in-one control to text-to-image diffusion models.Advances in Neural Information Processing Sys- tems, 36:11127–11150, 2023. 3

    2023

  34. [42]

    Local conditional controlling for text-to-image diffu- sion models

    Yibo Zhao, Liang Peng, Yang Yang, Zekai Luo, Hengjia Li, Yao Chen, Zheng Yang, Xiaofei He, Wei Zhao, Qinglin Lu, et al. Local conditional controlling for text-to-image diffu- sion models. InProceedings of the AAAI Conference on Ar- tificial Intelligence, pages 10492–10500, 2025. 3

    2025

  35. [43]

    Judging llm-as-a-judge with mt-bench and chatbot arena.Advances in Neural Information Processing Systems, 36:46595–46623, 2023

    Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric Xing, et al. Judging llm-as-a-judge with mt-bench and chatbot arena.Advances in Neural Information Processing Systems, 36:46595–46623, 2023. 3

    2023

  36. [44]

    Unpaired image-to-image translation using cycle- consistent adversarial networks

    Jun-Yan Zhu, Taesung Park, Phillip Isola, and Alexei A Efros. Unpaired image-to-image translation using cycle- consistent adversarial networks. InProceedings of the IEEE international conference on computer vision, pages 2223– 2232, 2017. 2

    2017

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.