Pith. sign in

REVIEW 3 major objections 4 minor 56 references

HRDiT: Training-Free High-Resolution Image Generation with Off-the-Shelf Diffusion Transformer Models

T0 review · 3 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read HRDiT lets off-the-shelf diffusion transformers draw 2K, 4K, and 8K images without retraining, by sliding token indices apart and pruning attention heads to their native windows.

desk verdict Strong, honest empirical paper on training-free high-resolution DiT generation, but HAP's pruning component rests on an undefined loss that blocks independent verification. read the letter →

arxiv 2608.07003 v1 pith:W6R5FORI submitted 2026-08-07 cs.CV

Yu Xue , Haoxuan Qu , Zhuoling Li , Hongbin Xu , Jianxiong Yin , Simon See , Hossein Rahmani , Jun Liu This is my paper · ORCID
classification cs.CV
keywords text-to-imagegenerationhigh-resolutionimagesynthesisdiffusiontransformertraining-freeadaptationpositionalembeddingextrapolationattentionpruningFLUXStable3
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

HRDiT argues that off-the-shelf diffusion transformer image generators such as FLUX and Stable Diffusion 3, trained at around 1024×1024, can be pushed to 2K, 4K, and 8K synthesis with no retraining. It attributes the two main blockers to positional embedding collapse and quadratic attention cost. Its response is SPA, which feeds the positional function sliding bundle indices instead of raw token indices, and HAP, which gives each attention head a per-head window chosen under a compute budget. The reported results across 2K, 4K, and 8K show better FID, KID, and CLIP scores and lower latency than every compared training-free baseline, making high-resolution synthesis a plug-in fix rather than a retraining project.

What carries the argument

The load-bearing object is the modified attention contribution $c^{\mathrm{SPA}}_{i,j} = \frac{1}{N}\sum_{n=1}^{N} g(x_i, x_j, f_{\mathrm{pe}}(\phi_{\mathrm{bundle}}^{(N_1=n)}(i), \phi_{\mathrm{bundle}}^{(N_1=n)}(j)))$, in which token indices $i,j$ are mapped to bundle indices by $\phi_{\mathrm{bundle}}$; SPA runs $N$ variants with first-bundle sizes $N_1=1,\ldots,N$ so the bundle boundaries slide, and each token gets a unique tuple of bundle indices. This directly shrinks the number of distinct pairwise positional signals $|S_{\mathrm{pe}}|$ that the positional encoder must separate, which is what Proposition 1 identifies as the bottleneck. The other load-bearing object is the HAP quality estimate $I_q(n_{\mathrm{head}}, n_{\mathrm{scope}})\approx \sum_{(u,v)\in S^{(n_{\mathrm{scope}})}_{\mathrm{omit}}}[\frac{\partial L}{\partial A(u,v)}(-A(u,v)) + \sum_{w\ne v}\frac{\partial L}{\partial A(u,w)}\frac{A(u,v)A(u,w)}{1-A(u,v)}]$, a one-forward-pass Taylor estimate of the quality cost of dropping the attention pairs outside each candidate window; a linear-solver pass over Eq.~8 then assigns each head a window under a total-cost ratio $r_c$. Together these two mechanisms convert resolution extrapolation into a change of input indices and a per-head pruning plan, with no training.

What would settle it

Run the same 4K FLUX experiment twice: once with the authors' scope-selection loss L and once using an explicitly defined held-out perceptual metric, for example FID on a separate prompt set, as the selection loss. If the two selected per-head windows differ markedly, or if using the reported FID as L reproduces the paper's tables exactly, the single-pass quality estimate's role in the result is unsupported.

Watch

Extended reading notes

Core claim

The central claim is that training-free high-resolution generation with off-the-shelf DiT models is not only possible but can beat existing training-free pipelines on quality and speed. Spatial disorder is traced to limited expressiveness of the positional embedding mechanism at large token counts: a pseudo-dimension bound (Proposition 1) shows the number of distinguishable pairwise positional signals cannot keep pace with the growth of the token grid. SPA restores distinguishability by replacing token indices with bundle indices and averaging over N sliding bundle offsets, yielding a unique positional representation for every token. Long generation time is traced to quadratic multi-head attention, and HAP prunes it by estimating, in a single forward pass, the quality degradation caused by omitting each candidate set of attention pairs, then solving an integer program for the per-head window sizes under a cost ratio r_c. Experiments on FLUX and SD3 at 2K, 4K, and 8K report better FID, KID, and CLIP scores and roughly 2–4x lower latency than the compared training-free baselines.

Load-bearing premise

The load-bearing premise is that the paper's unstated loss L, used by Eq. 7 to pick per-head windows, really measures generation-quality loss; if L is unrelated to image quality or is itself one of the reported FID/KID numbers, the pruning-preserves-quality claim is circular.

Editorial extensions

If this is right

  • A released FLUX or Stable Diffusion 3 checkpoint can generate 2K, 4K, and 8K images without any weight update; the only additions are index remapping and a precomputed per-head scope plan.
  • Reported 8K latency drops from 1,708s to 827s for FLUX and from 822s to 454s for SD3, with better FID/KID/CLIP scores than the compared baselines, so resolution scaling becomes a runtime choice rather than a retraining project.
  • Because SPA acts only on token indices before the positional function, the same fix should transfer to any diffusion transformer whose attention consumes token-index positional embeddings.
  • The one-time scope-preparation stage, about 48 minutes for 4K FLUX, is amortized over many images, after which the pruning plan is applied at near-zero runtime cost.

Reading between the lines

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

  • The same bundle/slide trick could be probed on non-generative token-index transformers, such as video diffusion transformers or long-context vision encoders, that also fail at length extrapolation; the paper only demonstrates image generation.
  • HAP's single-pass estimate could be recomputed per prompt or per denoising step to make the speed-quality trade-off adaptive; the paper fixes one plan per resolution from the final denoising step and shows stability only in ablations.
  • If the loss L used in Eq. 7 is left undefined in the released code, the paper's central quality-preservation claim cannot be independently reproduced; defining L as an explicit perceptual metric would turn the method into a self-contained speed-quality optimizer.
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

3 major / 4 minor

Summary. The paper proposes HRDiT, a training-free framework for adapting off-the-shelf Diffusion Transformer (DiT) text-to-image models such as FLUX and Stable Diffusion 3 to high-resolution generation. It identifies two challenges: spatial disorder and long generation time. To address them, HRDiT introduces Spatial Position Alignment (SPA), which replaces token indices with bundle/slide indices and averages over N positional mappings before attention, and Head-adaptive Attention Pruning (HAP), which assigns each attention head a per-head window chosen by optimizing an estimated quality-loss degradation under a computational-cost constraint. Experiments at 2K, 4K, and 8K report that HRDiT outperforms all compared training-free baselines on FID, FIDp, KID, KIDp, and CLIP score while reducing latency by roughly 2x--4x, with ablations showing both components contribute.

Significance. If the claims hold, HRDiT fills an important gap: existing training-free high-resolution methods target U-Net backbones, while the paper tackles the increasingly dominant DiT architecture. The empirical work is extensive, spanning two backbones, three resolutions, five quality metrics, a user study, comparison against a super-resolution baseline, and ablations that separate the contributions of SPA and HAP. The release of code is a further strength. However, the HAP component is optimized against a loss L that is never defined (Eq. 7), which makes the pruning-quality trade-off uncheckable and risks circularity if L coincides with one of the reported metrics. In addition, the theoretical support for SPA establishes per-token uniqueness of bundle-index tuples, not the pairwise distinguishability that Eq. 6 and the h(T)=|S_pe| argument require. These are load-bearing issues, but they are local and fixable, and the empirical foundation is strong enough that the paper merits a major revision rather than rejection.

major comments (3)
  1. [Sec. 3.2, Eq. (7)] The loss L used to define I_q(n_head, n_scope) is never defined in the main paper or the supplementary material. Eq. (7) is a Taylor expansion involving derivatives dL/dA(u,v), but no formula, network, reference image, or metric is given for L. Supplementary B merely states that L is computed from the output of the final denoising step at that resolution. This is load-bearing because the per-head scope assignment in HAP is selected by minimizing the I_q estimate in Eq. (7), and the resulting assignment produces the headline quality/latency numbers in Tables 1 and 3. If L is one of the reported metrics (FID, KID, CLIP, or a feature-space proxy tuned to them), the quality-preservation claim becomes circular; if L is arbitrary, the derivation cannot be checked. The authors should define L explicitly and either prove Eq. (7) or provide the derivation, including how the softmax and the denominator 1-A(u,v) are handled when A(u,v)=1.
  2. [Sec. 3.1, Eq. (6) and Proposition 1] The theoretical justification for SPA is incomplete. Proposition 1 bounds h(T), the number of distinguishable pairwise positional signals under the original function g with f_pe(i,j). The SPA operation in Eq. (6) replaces c_{i,j} with an average over N different bundle-index mappings, so the function whose distinguishability matters is a new averaged function, not the g analyzed in Proposition 1. The proof in Supplementary J shows that each token index i has a unique N-tuple of bundle indices, but it does not show that distinct pairs (i,j) and (i',j') produce averaged outputs that remain separated under the distance dis_g used in Eq. (4). The paper should either provide a formal pairwise-distinguishability statement for the SPA-averaged attention contribution or soften the claim that SPA restores h(T)=|S_pe|.
  3. [Sec. 4.1 and Supplementary Tables 8--11] Several HAP and SPA hyperparameters are selected by sweeping on the same quality metrics that appear in the headline tables. In particular, N_scope (Tab. 8), the bundle size N (Tab. 9), the cost ratio r_c (Tab. 10), and the number of prompts for I_q estimation (Tab. 11) are all chosen by comparing FID, FIDp, KID, KIDp, and CLIP values. The 1,000 evaluation prompts are sampled from LAION-5B, and although the 30 prompts used for I_q estimation are explicitly disjoint from the evaluation prompts, it is not stated whether the hyperparameter sweeps in the supplementary used the same evaluation set. If they did, part of the reported gains reflects tuning to the test set. The authors should clarify the selection protocol and, if needed, re-run the main comparisons with hyperparameters fixed on a held-out validation set.
minor comments (4)
  1. [Sec. 3.1, Eq. (2)] The indexing is inconsistent: the text defines token vectors as {q_t}_{t=0}^{T-1}, but the softmax sum in Eq. (2) is written as running from t=1 to T. Please unify the indexing.
  2. [Sec. 3.2] The text says the precompiled assignment can be applied 'in a cost-free manner,' but the associated per-image latency in Tables 1--3 excludes the one-time preparatory stage reported in Table 7 (about 48 minutes for FLUX at 4K). This should be stated clearly wherever latency numbers are discussed, since 'training-free' and 'cost-free' are distinct claims.
  3. [Sec. 3.1, Fig. 3 and Eq. (5)] The floor notation in Eq. (5) and in the caption of Fig. 3 is rendered inconsistently (e.g., 'T+N-N_1 over N' appears with different bracket styles). Please use a consistent notation such as \lceil\cdot\rceil or \lfloor\cdot\rfloor throughout, and verify the exact intended grouping.
  4. [References] Reference [49] contains a typo: 'Ning, X., , Chen, B.' has an extra comma after 'X.' Please correct.

Circularity Check

0 steps flagged · score 0.0 of 10
Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The framework introduces two fitted components: SPA's bundle size N is tuned per resolution on FID, and HAP's per-head scopes are fitted to an undefined loss L. The theoretical motivation rests on pseudo-dimension bounds and on assumptions about head scopes and quality proxies that are not independently validated.

free parameters (5)
  • Bundle size N = N=3 (2K), N=5 (4K)
    Chosen per resolution based on FID in Supp. Tab. 9; controls the SPA bundle and slide operations.
  • Number of candidate scopes N_scope = 50
    Selected in Supp. Tab. 8 because FID and KID stabilize at N_scope=50.
  • Target attention cost ratio r_c = 0.1
    Selected in Supp. Tab. 10 as the point where further quality gains are trivial while latency still improves.
  • Number of prompts for Iq estimation = 30
    Selected in Supp. Tab. 11 as a compromise between preparation time and quality.
  • Per-head attention scope assignment = One candidate scope per head, found by integer programming per resolution
    The output of HAP's preparatory stage, fitted to the undefined loss L on 30 prompts and reused at inference.
assumptions (4)
  • standard math Pseudo-dimension covering bound (Lemma 1 from [13]) is applicable to the function class G and yields the h(T) upper bound in Eq. 4.
    Used in the proof of Proposition 1 in Sec. 3.1 and Supp. H.
  • domain assumption Attention heads in trained DiT models have distinct, stable attention scopes that can be estimated from a single model pass.
    Borrowed from [5,43,49] and stated in Sec. 3.2 as the basis for HAP.
  • ad hoc to paper The undefined quality loss L computed on the final denoising output is a faithful proxy for generation quality degradation.
    Required by Eq. 7 to estimate Iq; L is never defined or analyzed in the paper.
  • ad hoc to paper Reducing |S_pe| via bundling and averaging via sliding restores h(T)=|S_pe|.
    Central to SPA's motivation in Sec. 3.1, but only per-token uniqueness is proved in Supp. J, not pairwise averaged distinguishability.

how reviews work

0 comments
Cite this review

Pith. "Pith review of HRDiT: Training-Free High-Resolution Image Generation with Off-the-Shelf Diffusion Transformer Models." pith.science (2026). https://pith.science/paper/W6R5FORI

@misc{pith2026260807003,
  author       = {Pith},
  title        = {Pith review of: HRDiT: Training-Free High-Resolution Image Generation with Off-the-Shelf Diffusion Transformer Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W6R5FORI}},
  note         = {Machine review of arXiv:2608.07003}
}
read the original abstract

Training-free text-to-high-resolution image generation has recently attracted growing research attention. However, existing studies on this task primarily focus on adapting off-the-shelf U-Net-based diffusion models to high resolutions, with limited progress on adapting off-the-shelf Diffusion Transformer (DiT) models despite their strong text-to-image generation capabilities at limited resolutions. In this work, we find two key challenges particularly hindering the application of off-the-shelf DiT models for high-resolution image synthesis in a training-free manner, namely, spatial disorder and long generation time. To address these challenges, we propose a novel method tailored to adapt off-the-shelf DiT models for high-resolution image synthesis. Extensive experiments show the efficacy of our method. Our code is available at: https://github.com/zylwithxy/HRDiT.

Figures

Figures reproduced from arXiv: 2608.07003 by the authors.

Figure 1
Figure 1. Results of (1) directly adapting off￾the-shelf text-to-image DiT models to high￾resolution (4096×4096) generation, and (2) applying the DiT-tailored method I-Max [7]. At limited resolution (1024×1024), off-the￾shelf DiT models generate high-quality im￾ages quickly. However, when adapted to higher resolutions through direct adapta￾tion or I-Max, they exhibit spatial disor￾der (see red boxes and zoom-in) and in￾cur lo… view at source ↗
Figure 2
Figure 2. Proportion of time that mainstream off-the￾shelf DiT models (FLUX [22] and Stable Diffusion 3 [8]) spent on multi-head attention computations per image, relative to that on the overall generation process per image, across different image resolutions. These models are adapted to high resolutions in the same direct manner as described in the caption of [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Illustration of the bundle and slide operations, with T = 7 and N = 3. (a) The bundle operation with N1 = 1: us￾ing the mapping function ϕbundle(·) with N1 = 1 (i.e., ϕ (N1=1) bundle (·)), the T = 7 to￾ken indices from 0 to 6 are mapped to  T +N−N1 N  = 3 bundle indices (0 to 2), re￾ducing the input diversity of fpe when its inputs are bundle indices rather than token indices. (b) Additionally applying the slide o… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Qualitative results of the 4K images gener￾ated by our method and by other methods (zoom￾in for better view). More qualitative results are in Supplementary. As shown, across both 2K and 4K generation tasks, our method consistently outper￾forms all baselines on all five…
Figure 5
Figure 5. Figure 5: Qualitative results of the 2K images generated by our method and by other state-of-the-art methods when applied under the same DiT backbone. a commonly used technique: balancing the entropy shift of self-attention [19]. For a fair comparison, all baseline methods in ou…
Figure 6
Figure 6. Figure 6: Qualitative results of the 4K images generated by our method and by other state-of-the-art methods when applied under the same DiT backbone. token index range does not effectively handle spatial disorder. This may be be￾cause downscaling does not reduce |Spe|. Hence, i…
Figure 7
Figure 7. Figure 7: Qualitative results of the 8K images generated by our method and by other state-of-the-art methods when applied under the same DiT backbone. For a fair comparison, we also set rc = 0.1 in this variant. In the third vari￾ant (random window), we randomly assign each head…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 36 canonical work pages

  1. [1]

    arXiv preprint arXiv:2510.20385 (2025)

    Bai, Y., Li, H., Huang, Q.: Positional encoding field. arXiv preprint arXiv:2510.20385 (2025)

    arXiv 2025

  2. [2]

    In: International Conference on Machine Learning

    Bar-Tal, O., Yariv, L., Lipman, Y., Dekel, T.: Multidiffusion: Fusing diffusion paths for controlled image generation. In: International Conference on Machine Learning. pp. 1737–1752. PMLR (2023)

    work page 2023

  3. [3]

    arXiv preprint arXiv:2504.06232 (2025)

    Bu, J., Ling, P., Zhou, Y., Zhang, P., Wu, T., Dong, X., Zang, Y., Cao, Y., Lin, D., Wang, J.: Hiflow: Training-free high-resolution image generation with flow-aligned guidance. arXiv preprint arXiv:2504.06232 (2025)

    arXiv 2025

  4. [4]

    arXiv preprint arXiv:2410.06055 (2024)

    Cao, B., Ye, J., Wei, Y., Shan, H.: Ap-ldm: Attentive and progressive latent dif- fusion model for training-free high-resolution image generation. arXiv preprint arXiv:2410.06055 (2024)

    arXiv 2024

  5. [5]

    In: Forty-second International Conference on Machine Learning (2025),https://openreview.net/forum?id=U74MOXPEJd

    Chen, Y., Zhang, P., Su, R., Ding, H., Stoica, I., Liu, Z., Zhang, H.: Fast video generation with sliding tile attention. In: Forty-second International Conference on Machine Learning (2025),https://openreview.net/forum?id=U74MOXPEJd

    work page 2025

  6. [6]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Du, R., Chang, D., Hospedales, T., Song, Y.Z., Ma, Z.: Demofusion: Democratising high-resolution image generation with no $$$. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 6159–6168 (2024)

    work page 2024

  7. [7]

    arXiv preprint arXiv:2410.07536 (2024)

    Du, R., Liu, D., Zhuo, L., Qi, Q., Li, H., Ma, Z., Gao, P.: I-max: Maximize the resolution potential of pre-trained rectified flow transformers with projected flow. arXiv preprint arXiv:2410.07536 (2024)

    arXiv 2024

  8. [8]

    In: Forty-first international conference on machine learning (2024)

    Esser, P., Kulal, S., Blattmann, A., Entezari, R., Müller, J., Saini, H., Levi, Y., Lorenz, D., Sauer, A., Boesel, F., et al.: Scaling rectified flow transformers for high-resolution image synthesis. In: Forty-first international conference on machine learning (2024)

    2024

Show all 56 references
  1. [9]

    In: European conference on computer vision

    Guo, L., He, Y., Chen, H., Xia, M., Cun, X., Wang, Y., Huang, S., Zhang, Y., Wang, X., Chen, Q., et al.: Make a cheap scaling: A self-cascade diffusion model for higher-resolution adaptation. In: European conference on computer vision. pp. 39–55. Springer (2024)

    2024

  2. [10]

    Gurobi Optimization, L.: Gurobi Optimizer Reference Manual (2025),https:// www.gurobi.com, last accessed 2026/06/29

    2025

  3. [11]

    arXiv preprint arXiv:2308.16137 (2023)

    Han, C., Wang, Q., Peng, H., Xiong, W., Chen, Y., Ji, H., Wang, S.: Lm-infinite: Zero-shot extreme length generalization for large language models. arXiv preprint arXiv:2308.16137 (2023)

    2023 arXiv

  4. [12]

    Hartmann, J., Exner, Y., Domdey, S.: The power of generative marketing: Can generative ai create superhuman visual marketing content? International Journal of Research in Marketing42(1), 13–31 (2025)

    2025

  5. [13]

    Information and computation100(1), 78–150 (1992)

    Haussler, D.: Decision theoretic generalizations of the pac model for neural net and other learning applications. Information and computation100(1), 78–150 (1992)

    1992

  6. [14]

    In: The Twelfth International Conference on Learning Represen- tations (2023)

    He, Y., Yang, S., Chen, H., Cun, X., Xia, M., Zhang, Y., Wang, X., He, R., Chen, Q., Shan, Y.: Scalecrafter: Tuning-free higher-resolution visual generation with diffusion models. In: The Twelfth International Conference on Learning Represen- tations (2023)

    2023

  7. [15]

    In: International Conference on Machine Learning

    Hoogeboom, E., Heek, J., Salimans, T.: simple diffusion: End-to-end diffusion for high resolution images. In: International Conference on Machine Learning. pp. 13213–13232. PMLR (2023)

    2023

  8. [16]

    In: European conference on computer vision

    Huang, L., Fang, R., Zhang, A., Song, G., Liu, S., Liu, Y., Li, H.: Fouriscale: A fre- quency perspective on training-free high-resolution image synthesis. In: European conference on computer vision. pp. 196–212. Springer (2024) HRDiT: Training-Free High-Resolution Image Generation 17

    2024

  9. [17]

    arXiv preprint arXiv:2510.20766 (2025)

    Issachar, N., Yariv, G., Benaim, S., Adi, Y., Lischinski, D., Fattal, R.: Dype: Dy- namic position extrapolation for ultra high resolution diffusion. arXiv preprint arXiv:2510.20766 (2025)

    2025

  10. [18]

    In: Forty-first International Conference on Machine Learning (2024),https://openreview.net/ forum?id=nkOMLBIiI7

    Jin, H., Han, X., Yang, J., Jiang, Z., Liu, Z., Chang, C.Y., Chen, H., Hu, X.: LLM maybe longLM: Selfextend LLM context window without tuning. In: Forty-first International Conference on Machine Learning (2024),https://openreview.net/ forum?id=nkOMLBIiI7

    2024

  11. [19]

    Advances in Neural Information Processing Systems36, 70847–70860 (2023)

    Jin, Z., Shen, X., Li, B., Xue, X.: Training-free diffusion model adaptation for variable-sized text-to-image synthesis. Advances in Neural Information Processing Systems36, 70847–70860 (2023)

    2023

  12. [20]

    In: Proceedings of the AAAI conference on artificial intelligence

    Kim, Y., Hwang, G., Zhang, J., Park, E.: Diffusehigh: Training-free progressive high-resolution image synthesis through structure guidance. In: Proceedings of the AAAI conference on artificial intelligence. vol. 39, pp. 4338–4346 (2025)

    2025

  13. [21]

    arXiv preprint arXiv:2510.25818 (2025)

    Koh, S., Cha, S., Oh, H., Lee, K., Kim, D.J.: Scalediff: Higher-resolution image syn- thesis via efficient and model-agnostic diffusion. arXiv preprint arXiv:2510.25818 (2025)

    2025

  14. [22]

    Labs, B.F.: Flux.https://github.com/black- forest- labs/flux(2024), last accessed 2026/06/29

    2024

  15. [23]

    arXiv preprint arXiv:2602.12769 (2026)

    Lai, H.P., Nguyen, P., Tran, A.: Pixelrush: Ultra-fast, training-free high-resolution image generation via one-step diffusion. arXiv preprint arXiv:2602.12769 (2026)

    2026

  16. [24]

    Advances in Neural Information Processing Systems37, 58610–58636 (2024)

    Lee, Y., Yoon, T., Sung, M.: Groundit: Grounding diffusion transformers via noisy patch transplantation. Advances in Neural Information Processing Systems37, 58610–58636 (2024)

    2024

  17. [25]

    In: The Twelfth International Conference on Learning Representations (2024),https://openreview.net/forum? id=N23A4ybMJr

    Leroy, V., Revaud, J., Lucas, T., Weinzaepfel, P.: Win-win: Training high- resolution vision transformers from two windows. In: The Twelfth International Conference on Learning Representations (2024),https://openreview.net/forum? id=N23A4ybMJr

    2024

  18. [26]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Li, Z., Rahmani, H., Ke, Q., Liu, J.: Longdiff: Training-free long video genera- tion in one go. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 17789–17798 (2025)

    2025

  19. [27]

    In: European Conference on Computer Vision

    Lin, Z., Lin, M., Zhao, M., Ji, R.: Accdiffusion: An accurate method for higher- resolution image generation. In: European Conference on Computer Vision. pp. 38–53. Springer (2024)

    2024

  20. [28]

    arXiv preprint arXiv:2503.18719 (2025)

    Liu, C., Hou, L., Zheng, M., Tao, X., Wan, P., Zhang, D., Gai, K.: Boosting resolu- tion generalization of diffusion transformers with randomized positional encodings. arXiv preprint arXiv:2503.18719 (2025)

    2025

  21. [29]

    arXiv preprint arXiv:2412.16112 (2024)

    Liu, S., Tan, Z., Wang, X.: Clear: Conv-like linearization revs pre-trained diffusion transformers up. arXiv preprint arXiv:2412.16112 (2024)

    2024 arXiv

  22. [30]

    Peebles,W.,Xie,S.:Scalablediffusionmodelswithtransformers.In:Proceedingsof the IEEE/CVF international conference on computer vision. pp. 4195–4205 (2023)

    2023

  23. [31]

    arXiv preprint arXiv:2307.01952 (2023)

    Podell, D., English, Z., Lacey, K., Blattmann, A., Dockhorn, T., Müller, J., Penna, J., Rombach, R.: Sdxl: Improving latent diffusion models for high-resolution image synthesis. arXiv preprint arXiv:2307.01952 (2023)

    2023 arXiv

  24. [32]

    Qin, J.: How does text-to-image ai affect indie game designers and artists? Journal ofInnovationandDevelopment5,107–111(122023).https://doi.org/10.54097/ f7of9f8k

  25. [33]

    arXiv preprint arXiv:2508.15774 (2025)

    Qiu, H., Yu, N., Huang, Z., Debevec, P., Liu, Z.: Cinescale: Free lunch in high- resolution cinematic visual generation. arXiv preprint arXiv:2508.15774 (2025)

    2025 arXiv

  26. [34]

    arXiv preprint arXiv:2412.09626 (2024) 18 Y

    Qiu, H., Zhang, S., Wei, Y., Chu, R., Yuan, H., Wang, X., Zhang, Y., Liu, Z.: Freescale: Unleashing the resolution of diffusion models via tuning-free scale fusion. arXiv preprint arXiv:2412.09626 (2024) 18 Y. Xue et al

    2024 arXiv

  27. [35]

    Advances in Neural Information Processing Systems37, 111131–111171 (2024)

    Ren, J., Li,W., Chen,H., Pei, R.,Shao, B., Guo,Y., Peng, L.,Song, F.,Zhu,L.:Ul- trapixel: Advancing ultra high-resolution image synthesis to new peaks. Advances in Neural Information Processing Systems37, 111131–111171 (2024)

    2024

  28. [36]

    Advances in neural information processing systems35, 25278–25294 (2022)

    Schuhmann, C., Beaumont, R., Vencu, R., Gordon, C., Wightman, R., Cherti, M., Coombes, T., Katta, A., Mullis, C., Wortsman, M., et al.: Laion-5b: An open large- scale dataset for training next generation image-text models. Advances in neural information processing systems35, 2...

    2022

  29. [37]

    arXiv preprint arXiv:2510.27135 (2025)

    Shen, T., Yu, J., Zhou, D., Li, D., Barsoum, E.: E-mmdit: Revisiting multimodal diffusion transformer design for fast image synthesis under limited resources. arXiv preprint arXiv:2510.27135 (2025)

    2025

  30. [38]

    Neurocomputing568, 127063 (2024)

    Su, J., Ahmed, M., Lu, Y., Pan, S., Bo, W., Liu, Y.: Roformer: Enhanced trans- former with rotary position embedding. Neurocomputing568, 127063 (2024)

    2024

  31. [39]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Tang, Y., Han, K., Wang, Y., Xu, C., Guo, J., Xu, C., Tao, D.: Patch slimming for efficient vision transformers. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 12165–12174 (2022)

    2022

  32. [40]

    Advances in neural information pro- cessing systems30(2017)

    Vaswani,A.,Shazeer,N.,Parmar,N.,Uszkoreit,J.,Jones,L.,Gomez,A.N.,Kaiser, Ł., Polosukhin, I.: Attention is all you need. Advances in neural information pro- cessing systems30(2017)

    2017

  33. [41]

    In: Proceedings of the SIG- GRAPH Asia 2025 Conference Papers

    Vontobel,T.,Sadat, S.,Salehi,F., Weber, R.:Hiwave: Training-freehigh-resolution image generation via wavelet-based diffusion sampling. In: Proceedings of the SIG- GRAPH Asia 2025 Conference Papers. pp. 1–11 (2025)

    2025

  34. [42]

    In: 2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)

    Wu, H., Shen, S., Hu, Q., Zhang, X., Zhang, Y., Wang, Y.: Megafusion: Extend diffusion models towards higher-resolution image generation without further tun- ing. In: 2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). pp. 3944–3953. IEEE (2025)

    2025

  35. [43]

    In: The Thirteenth International Confer- ence on Learning Representations (2025),https://openreview.net/forum?id= EytBpUGB1Z

    Wu, W., Wang, Y., Xiao, G., Peng, H., Fu, Y.: Retrieval head mechanisti- cally explains long-context factuality. In: The Thirteenth International Confer- ence on Learning Representations (2025),https://openreview.net/forum?id= EytBpUGB1Z

    2025

  36. [44]

    In: Proceedings of the Computer Vision and Pat- tern Recognition Conference

    Yang, H., Bulat, A., Hadji, I., Pham, H.X., Zhu, X., Tzimiropoulos, G., Martinez, B.: Fam diffusion: Frequency and attention modulation for high-resolution image generation with stable diffusion. In: Proceedings of the Computer Vision and Pat- tern Recognition Conference. pp. ...

    2025

  37. [45]

    arXiv e-prints pp

    Yang, Z., Shen, G., Hou, L., Liu, M., Wang, L., Tao, X., Wan, P., Zhang, D., Chen, Y.C.: Rectifiedhr: Enable efficient high-resolution image generation via energy rec- tification. arXiv e-prints pp. arXiv–2503 (2025)

    2025

  38. [46]

    In: Eighth Conference on Machine Learning and Systems (2025),https://openreview.net/forum?id=RXPofAsL8F

    Ye, Z., Chen, L., Lai, R., Lin, W., Zhang, Y., Wang, S., Chen, T., Kasikci, B., Grover, V., Krishnamurthy, A., Ceze, L.: Flashinfer: Efficient and customizable attention engine for LLM inference serving. In: Eighth Conference on Machine Learning and Systems (2025),https://open...

    2025

  39. [47]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Yu,F.,Gu,J.,Li,Z.,Hu,J.,Kong,X.,Wang,X.,He,J.,Qiao,Y.,Dong,C.:Scaling up to excellence: Practicing model scaling for photo-realistic image restoration in the wild. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 25669–25680 (2024)

    2024

  40. [48]

    Advances in Neural Information Processing Systems37, 1196–1219 (2024)

    Yuan, Z., Zhang, H., Pu, L., Ning, X., Zhang, L., Zhao, T., Yan, S., Dai, G., Wang, Y.: Ditfastattn: Attention compression for diffusion transformer models. Advances in Neural Information Processing Systems37, 1196–1219 (2024)

    2024

  41. [49]

    arXiv preprint arXiv:2406.14909 (2024) HRDiT: Training-Free High-Resolution Image Generation 19

    Zhang, G., Fu, T., Huang, H., Ning, X., , Chen, B., Wu, T., Wang, H., Huang, Z., Li, S., Yan, S., et al.: Moa: Mixture of sparse attention for automatic large language model compression. arXiv preprint arXiv:2406.14909 (2024) HRDiT: Training-Free High-Resolution Image Generation 19

    2024

  42. [50]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Zhang,H.,Su,R.,Yuan,Z.,Chen,P.,Shen,M.,Fan,Y.,Yan,S.,Dai,G.,Wang,Y.: Ditfastattnv2: Head-wise attention compression for multi-modality diffusion trans- formers. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 16399–16409 (2025)

    2025

  43. [51]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Zhang, J., Huang, Q., Liu, J., Guo, X., Huang, D.: Diffusion-4k: Ultra-high- resolution image synthesis with latent diffusion models. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 23464–23473 (2025)

    2025

  44. [52]

    In: European Conference on Computer Vision

    Zhang, S., Chen, Z., Zhao, Z., Chen, Y., Tang, Y., Liang, J.: Hidiffusion: Unlock- ing higher-resolution creativity and efficiency in pretrained diffusion models. In: European Conference on Computer Vision. pp. 145–161. Springer (2024)

    2024

  45. [53]

    arXiv preprint arXiv:2503.04344 (2025)

    Zhang, S., Liang, S., Tan, Y., Chen, Z., Li, L., Wu, G., Chen, Y., Li, S., Zhao, Z., Chen, C., et al.: Ledit: Your length-extrapolatable diffusion transformer without positional encoding. arXiv preprint arXiv:2503.04344 (2025)

    2025

  46. [54]

    In: The Thirteenth International Confer- ence on Learning Representations (2025),https://openreview.net/forum?id= TsBDfe8Ra5

    Zhang, Z., Li, R., Zhang, L.: Frecas: Efficient higher-resolution image generation via frequency-aware cascaded sampling. In: The Thirteenth International Confer- ence on Learning Representations (2025),https://openreview.net/forum?id= TsBDfe8Ra5

    2025

  47. [55]

    arXiv preprint arXiv:2512.04504 (2025)

    Zhao, M., Yan, B., Yang, X., Zhu, H., Zhang, J., Liu, S., Li, C., Zhu, J.: Ultraimage: Rethinking resolution extrapolation in image diffusion transformers. arXiv preprint arXiv:2512.04504 (2025)

    2025

  48. [56]

    Ad- vances in Neural Information Processing Systems35, 9010–9023 (2022) 20 Y

    Zheng, C., Zhang, K., Yang, Z., Tan, W., Xiao, J., Ren, Y., Pu, S., et al.: Savit: Structure-aware vision transformer pruning via collaborative optimization. Ad- vances in Neural Information Processing Systems35, 9010–9023 (2022) 20 Y. Xue et al. HRDiT: Training-Free High-Reso...

    2022

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.