Pith. sign in

REVIEW 5 major objections 5 minor 3 cited by

FreeScale: Unleashing the Resolution of Diffusion Models via Tuning-Free Scale Fusion

T0 review · 5 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Tuning-free scale fusion lets pre-trained diffusion models generate 8K images.

desk verdict A solid engineering paper with a genuinely new frequency-based fusion rule; the 8k resolution claim is not supported by distribution-level evidence. read the letter →

arxiv 2412.09626 v2 pith:SP52SAE7 submitted 2024-12-12 cs.CV

Haonan Qiu , Shiwei Zhang , Yujie Wei , Ruihang Chu , Hangjie Yuan , Xiang Wang , Yingya Zhang , Ziwei Liu This is my paper · ORCID
classification cs.CV
keywords diffusionmodelshigh-resolutionimagegenerationtuning-freeinferencescalefusionfrequencydecompositionself-attentiontext-to-video8K
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

FreeScale is a tuning-free inference method that lets a pre-trained diffusion model produce images and videos at resolutions far above its training resolution. The paper's claim is that repetitive-object artifacts arise because going beyond training resolution increases high-frequency information, and that the fix is to fuse two receptive scales: keep global attention's high-frequency structure and local attention's low-frequency content, exchanging them through a Gaussian-blur split. On a frozen text-to-image model the method yields coherent 8192x8192 images, and on a text-to-video model it yields 640x1024 video, all without retraining. The authors present this as the first tuning-free method to reach 8K-resolution text-to-image generation.

What carries the argument

The central object is the scale-fusion identity in Equation 7, applied to the self-attention output of each transformer block. It combines a global self-attention pass over the full feature map with a local self-attention pass over shifted crops, using Gaussian blur $G$ as the low-pass filter: the final output keeps the high-frequency component of the global pass and the low-frequency component of the local pass. A tailored self-cascade upsampling seeds the high-resolution denoising with a noised, upscaled version of the low-resolution result, and restrained dilated convolution is applied only to down-blocks and mid-blocks. The machinery's job is to balance local and global receptive scales so that neither local repetition nor globally scattered small objects survive.

What would settle it

Generate a test image at 16x resolution with FreeScale and inspect the intermediate signals: if a diagnostic shows repeated objects surviving in the low-frequency band of the local attention output $G(h_{\mathrm{local}})$, or structural collapse in the high-frequency band of the global output $h_{\mathrm{global}} - G(h_{\mathrm{global}})$, the frequency-separability premise fails. A simple variant is to run the fusion with the split inverted (high frequency from local, low frequency from global) and check whether repetitions reappear, which the paper's mechanism predicts they would.

Watch

Extended reading notes

Core claim

The central discovery is a frequency-domain recombination inside the self-attention layer. When a latent diffusion model is asked to denoise at a resolution higher than its training resolution, the self-attention output accumulates dispersed high-frequency signals that read as repeated objects. FreeScale computes self-attention twice, once over the full feature map and once over shifted local crops, then forms its output as $h_{\mathrm{fusion}} = h_{\mathrm{global}} - G(h_{\mathrm{global}}) + G(h_{\mathrm{local}})$, where $G$ is a Gaussian low-pass filter. This keeps the high-frequency part of the global attention output and the low-frequency part of the local attention output. The paper reports that this eliminates both local repetition and globally scattered small objects, allowing a frozen SDXL model to generate 8192x8192 images and VideoCrafter2 to generate 640x1024 videos, with flexible local detail control and flexible aspect ratios as additional capabilities.

Load-bearing premise

The fusion rule assumes that all high-frequency content produced by local attention is harmful repetition while the low-frequency content of global attention is semantically correct; the paper does not analyze whether that frequency split actually separates artifacts from structure.

Editorial extensions

If this is right

  • A frozen text-to-image model can generate 8192x8192 images on a single A800 GPU, with quality metrics better than prior tuning-free baselines and competitive with super-resolution post-processing.
  • A frozen text-to-video model can generate 640x1024 videos with better FVD, dynamic degree, and aesthetic quality than prior tuning-free methods, at lower inference time.
  • Users can control the detail level spatially through a semantic mask and perform local semantic edits, such as changing hair color or facial features, during high-resolution denoising.
  • The method supports flexible aspect ratios that are multiples of 512 and can combine with distilled fast models such as SDXL-Turbo to produce 2048x2048 images in as few as two denoising steps.

Reading between the lines

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

  • Beyond the paper: if the frequency-separability assumption holds across architectures, the same recombination rule could be adapted to DiT-based diffusion models, which the paper notes face blur rather than repetition at high resolution and may need the inverse frequency split.
  • Beyond the paper: because the Gaussian blur kernel is sized in proportion to the resolution multiplier, the optimal blur scale should track the spatial-frequency cutoff where the model's training data ends; varying it could yield a diagnostic for each model's practical resolution ceiling.
  • Beyond the paper: keeping the global high-frequency structure means FreeScale acts as a prior-driven upsampler, which suggests a direct extension to controlled image-to-image and video upscaling beyond the demonstrated FLUX-seeded 8K example.
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

5 major / 5 minor

Summary. FreeScale proposes a tuning-free inference framework that extends pre-trained U-Net diffusion models (SDXL for images, VideoCrafter2 for video) to resolutions far above their training resolution, up to 8192x8192 images and 640x1024 videos. The method combines three components: tailored self-cascade upscaling with a cosine detail-control factor (Eq. 4), restrained dilated convolution applied only to down-blocks and mid-blocks in early timesteps, and scale fusion, which replaces the high-frequency component of local self-attention with the high-frequency component of global self-attention via a Gaussian blur (Eq. 7). Experiments compare against SDXL-DI, ScaleCrafter, DemoFusion, and FouriScale on image metrics (FID/KID/FIDc/KIDc/IS) and video metrics (FVD, VBench), with ablations and user studies. The paper claims to 'unlock 8k-resolution text-to-image generation for the first time.'

Significance. If the central claims hold, FreeScale would be a practically useful, training-free recipe for pushing U-Net diffusion models far beyond their native resolution, with a simple mechanism that also transfers to video and supports flexible aspect ratios and local control. The method is clearly described, the components are individually ablated, and the reported inference times are competitive. However, the headline 8k claim is not supported by quantitative evaluation, and the method's own quantitative comparisons show mixed results on the local-crop metrics that are most relevant to the repetition artifacts the paper targets. The frequency-separable artifact assumption behind scale fusion is untested. As presented, the contribution is a plausible and well-engineered empirical method, but its strongest claims require additional validation before they can be accepted at face value.

major comments (5)
  1. [Section 4.1 and Section C.3] The paper's headline claim is that FreeScale 'unlocks the 8k-resolution text-to-image generation for the first time' (Abstract and Introduction), yet all quantitative evaluations stop at 4096x4096 (Table 1), and the 8k results are confined to qualitative figures (Figures 10 and 12). Moreover, Section A states that for 8k generation alpha is reduced from 2 to 1, so the 8k results correspond to a different hyperparameter regime that has not been quantitatively validated. The authors should either provide distribution-level evidence at 8192x8192 (e.g., a user study on a larger prompt set, or meaningful quantitative metrics) or qualify the 8k claim as a qualitative demonstration.
  2. [Section 4.1, Table 1] On the cropped metrics FIDc and KIDc, which are intended to capture local quality, SDXL-DI (direct inference) outperforms FreeScale at both 2048^2 (FIDc 31.042 vs 36.276; KIDc 0.004 vs 0.006) and 4096^2 (FIDc 42.383 vs 71.369; KIDc 0.009 vs 0.029). The paper explains that duplicated objects in SDXL-DI make its crops closer to the reference distribution, but it does not substantiate this explanation. Since the central motivation is to eliminate repetitions, the method should not be worse on local-crop metrics than the baseline that exhibits repetitions; the authors should analyze this discrepancy (e.g., by comparing crops side-by-side or decomposing the metric contribution) or reconsider the interpretation of these metrics.
  3. [Section 3.4, Eq. (7)] Scale fusion assumes that repeated-object artifacts are carried by the high-frequency component of the local attention output, and that the global attention output provides correct high-frequency structure. This is the core design hypothesis, but the paper provides no evidence that the artifact mechanism is separable in frequency. The ablation in Table 3 removes the entire scale fusion module, which shows a large drop in quality, but it does not isolate the frequency split. A direct test would be to ablate the fusion formula, e.g., swapping the high- and low-frequency contributions, or using a simple weighted sum of global and local outputs, to verify that the frequency-based replacement, rather than merely having two receptive fields, is what drives the improvement.
  4. [Section 3.2 and Section A] The paper describes FreeScale as 'tuning-free,' but it requires setting several hyperparameters: noise timestep K, detail-control exponent alpha, dilation factor d, Gaussian blur kernel size and standard deviation, and local attention window size and stride. The default values are given, but no sensitivity analysis is provided, and alpha is changed for 8k generation (from 2 to 1). For a method that claims to be a turnkey solution, the robustness of these choices should be demonstrated; at minimum, a sweep over alpha and K, or an explicit discussion of how to select them, is needed. Without such analysis, the 'tuning-free' characterization is misleading and the 8k results rely on unvalidated parameter adjustments.
  5. [Section 2 and Section 4.1] The claim of being 'the first' 8k text-to-image generation method is not supported by a comparison with the most recent tuning-free high-resolution generation methods. The related work cites HiDiffusion, ElasticDiffusion, AP-LDM, CutDiffusion, and others, but these are not included in the quantitative or qualitative comparisons. To substantiate the 'first' claim, the authors should compare against these methods at 8192x8192 (or at least at 4096x4096) and demonstrate that they fail or produce inferior results, or they should qualify the novelty claim to reflect the limited comparison set.
minor comments (5)
  1. [Section 4.1, Table 1] All quantitative metrics are reported as single values without standard deviations or confidence intervals. Given that the differences from some baselines are modest (e.g., FIDc at 2048^2), the paper could benefit from multiple runs or at least a bootstrap-based confidence interval to establish that the findings are not noise.
  2. [Section 4.2, Table 2] The video evaluation is performed only on VideoCrafter2; the claim that FreeScale generalizes to video would be strengthened by a second video backbone, even if the results are reported in the supplementary material.
  3. [Section 3.2, Eq. (4) and Figure 4] The spatial variant of the detail-control factor (alpha as a 2D tensor) is described but only demonstrated qualitatively; the paper should clarify how the alpha map is derived from the semantic mask and how it interacts with the rest of the pipeline.
  4. [Throughout] The text contains several typos, including 'higher-solution' where 'higher-resolution' is intended (e.g., Section 4.1 and Section 4.3), and the reference list has formatting artifacts (e.g., '2, 3, 4, 5, 7, 1' at the end of references [14], [20], and [25]). These should be cleaned up.
  5. [Section B.1, Table 4] The comparison with super-resolution post-processing (SDXL+Real-ESRGAN) is useful but somewhat orthogonal to the main contribution; it might be clearer to integrate this into the main paper as a discussion of the difference between upscaling and higher-resolution generation, or to keep it in the supplementary material as is.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: FreeScale's scale-fusion rule is a hand-designed operation evaluated on external benchmarks, and its self-citations are peripheral rather than load-bearing.

full rationale

The core claim is that processing information from different receptive scales and fusing it via Eq. 7, h_fusion = h_global - G(h_global) + G(h_local), yields higher-resolution generation without repetition. This is not circular: Eq. 7 is a designed fusion rule, not a quantity fitted to the evaluation data and then renamed as a prediction. The frequency-decomposition assumption (global high-frequency detail plus local low-frequency semantics) is an empirical design hypothesis, and the paper tests it with FID, KID, IS, FVD, VBench, and user studies against external baselines on LAION and WebVid prompts. No equation in the paper fits parameters to the target result and then reuses that fit as evidence; the hyperparameters K, alpha, and sigma are manually selected and validated, but the central effectiveness claim does not reduce to these choices by construction. The only author-overlapping citations, FreeNoise [41] and FreeU [47], are cited for standard local-attention computation and as an optional compatible enhancement, respectively, and neither supports a uniqueness claim nor forbids alternative designs. The limitation statement that upscaling is bounded by the base model's prior knowledge is an acknowledged boundary, not a self-validating premise. The headline 8k-generation result is supported mainly by qualitative examples rather than distribution-level metrics, but that is an evidence-strength concern, not a circularity. Overall, the derivation chain is self-contained: the method's components are described, combined, and then evaluated against independent baselines, so no load-bearing step reduces to its own input.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on several hand-tuned hyperparameters and on heuristic assumptions about the frequency structure of repetition artifacts. The method is not derived from a theory; it is an empirical recipe whose components are individually motivated by prior work.

free parameters (5)
  • Noise timestep K for cascade = 700
    Set manually; controls how much noise is added to upscaled latent in Eq. 3; chosen by validation across all experiments.
  • Detail control exponent alpha = 2 (default), 1 for 8k, 3/0.5 for masked regions
    Scaling factor in the cosine decay of Eq. 4; hand-tuned per resolution and per region.
  • Dilation factor d = equal to resolution level (2, 4, 16)
    Sets convolution dilation in Eq. 5; assumed proportional to the upscaling factor.
  • Gaussian blur kernel size and standard deviation = kernel = 2*sqrt(H*W)/(1024^2)-1, sigma=1
    Defines the low-pass filter G in Eq. 7; formula chosen heuristically and fixed across experiments.
  • Local attention window size and stride (h, w, dh, dw) = not specified
    The shifted crop sampling S_local in Sec 3.4 is described without concrete window and stride values; a free implementation choice that affects fusion results.
assumptions (4)
  • domain assumption High-frequency information increases when the model generates beyond training resolution, causing repetitive patterns.
    Stated in the abstract and Sec 1 as the key obstacle; not proven, and the paper's method is built on this diagnosis.
  • domain assumption The repetition artifact can be separated by frequency: high-frequency signals from global attention are correct, low-frequency signals from local attention are correct, and their recombination in Eq. 7 removes repetition.
    Central heuristic of Scale Fusion (Sec 3.4, Eq. 7); no theoretical analysis is provided.
  • ad hoc to paper Dilated convolution should be applied only to down-blocks and mid-blocks, and only in early timesteps.
    Empirical design choice reported in Sec 3.3 after observing artifacts with up-block dilation.
  • domain assumption The pre-trained base model contains enough prior knowledge to generate plausible details at the target resolution.
    Acknowledged as a limitation in Sec D of the supplement: beyond the model's prior, no more details can be reasonably added.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FreeScale: Unleashing the Resolution of Diffusion Models via Tuning-Free Scale Fusion." pith.science (2026). https://pith.science/paper/SP52SAE7

@misc{pith2026241209626,
  author       = {Pith},
  title        = {Pith review of: FreeScale: Unleashing the Resolution of Diffusion Models via Tuning-Free Scale Fusion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SP52SAE7}},
  note         = {Machine review of arXiv:2412.09626}
}
read the original abstract

Visual diffusion models achieve remarkable progress, yet they are typically trained at limited resolutions due to the lack of high-resolution data and constrained computation resources, hampering their ability to generate high-fidelity images or videos at higher resolutions. Recent efforts have explored tuning-free strategies to exhibit the untapped potential higher-resolution visual generation of pre-trained models. However, these methods are still prone to producing low-quality visual content with repetitive patterns. The key obstacle lies in the inevitable increase in high-frequency information when the model generates visual content exceeding its training resolution, leading to undesirable repetitive patterns deriving from the accumulated errors. To tackle this challenge, we propose FreeScale, a tuning-free inference paradigm to enable higher-resolution visual generation via scale fusion. Specifically, FreeScale processes information from different receptive scales and then fuses it by extracting desired frequency components. Extensive experiments validate the superiority of our paradigm in extending the capabilities of higher-resolution visual generation for both image and video models. Notably, compared with previous best-performing methods, FreeScale unlocks the 8k-resolution text-to-image generation for the first time.

Figures

Figures reproduced from arXiv: 2412.09626 by the authors.

Figure 1
Figure 1. Gallery of FreeScale. Original SDXL [40] can only generate images with a resolution of up to 10242 without losing quality, while FreeScale successfully extends SDXL to generate 81922 images without any fine-tuning. All generated images are produced using a single A800 GPU. Best viewed ZOOMED-IN. Abstract Visual diffusion models achieve remarkable progress, yet they are typically trained at limited resolutions due to… view at source ↗
Figure 2
Figure 2. Overall framework of FreeScale. (a) Tailored Self-Cascade Upscaling. FreeScale starts with pure Gaussian noise and pro￾gressively denoises it using the training resolution. An image is then generated via the VAE decoder, followed by upscaling to obtain a higher-resolution one. We gradually add noise to the latent of this higher-resolution image and incorporate this forward noise into the denoising process of the hig… view at source ↗
Figure 3
Figure 3. Image qualitative comparisons with other baselines. Our method generates both 20482 and 40962 vivid images with better content coherence and local details. Best viewed ZOOMED-IN. 6 [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Results of flexible control for detail level. A better result will be generated by adding the coefficient weight in the area of Griffons and reducing the coefficient weight in the other regions. Best viewed ZOOMED-IN. 1× Result No Editing Hair Editing Face Editing [PI…
Figure 5
Figure 5. Figure 5: Results of local semantic editing. FreeScale makes the hair purple or edits the face to make this person look more Japanese in the higher-resolution (40962 ). Latent Space Upsampling Ours w/o Scale Fusion Dilated Up-Blocks [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Qualitative image comparisons with ablations. Our full method performs the best. The resolution of results is 40962 for better visualizing the difference between the various strategies. best FIDc and KIDc. The reason is that SDXL-DI tends to generate multiple duplicate…
Figure 7
Figure 7. Figure 7: Video qualitative comparisons with other baselines. While other baselines fail in video generation, FreeScale effectively generates higher-resolution videos with high fidelity. Best viewed ZOOMED-IN [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Image qualitative comparisons with super-resolution. FreeScale is not inferior to SDXL+Real-ESRGAN in visual quality, and adds more details. In addition, SR methods will faithfully follow the low-resolution input while FreeScale can regenerate the original blurred area…
Figure 9
Figure 9. Figure 9: Flexible aspect ratio generation. FreeScale can di￾rectly achieve a flexible aspect ratio (the resolution must be a mul￾tiple of 512) without any adaptation. C. More Results C.1. Flexible Aspect Ratio Generation As shown in [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]
Figure 10
Figure 10. Figure 10: Zoomed in details for the 8k image. FreeScale may regenerate the original blurred areas at low resolution based on the prior knowledge that the model has learned. As shown in the bottom row, two originally chaotic and blurry faces are clearly outlined at 8k resolution…
Figure 11
Figure 11. Figure 11: Fast generation with SDXL-Turbo. FreeScale can help SDXL-Turbo generate results at 20482 resolution with even 2 timesteps. C.2. Fast Generation with SDXL-Turbo FreeScale can easily be compatible with other models with similar structures. SDXL-Turbo [45] is a distilled…
Figure 12
Figure 12. Figure 12: Gallery of generated 8k images. We place the original-resolution result in the lower right corner for reference. FreeScale effectively enhances local details without compromising the visual structure or introducing object repetitions. Best viewed ZOOMED-IN. 4 [PITH_F…
Figure 13
Figure 13. Figure 13: Structure gap. UNet-based LDMs and DiT-based LDMs will face different challenges in the higher-resolution gen￾eration task. UNet-based LDMs face repetition problems while DiT-based LDMs face blur problems. originally chaotic and blurry faces are clearly outlined at 8k…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. CineScale: Free Lunch in High-Resolution Cinematic Visual Generation

    cs.CV 2025-08 conditional novelty 6.0 of 10

    CineScale extends pre-trained diffusion models to 8k image and 4k video generation with mostly tuning-free inference plus a small LoRA adaptation for video.

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

    cs.CV 2026-08 conditional novelty 5.0 of 10

    HRDiT enables off-the-shelf diffusion transformer text-to-image models to produce coherent high-resolution images with far less computation, without any retraining.

  3. UltraImageGen: Efficient Ultra-High-Resolution Image Generation with Hierarchical Local Attention

    cs.CV 2025-10 conditional novelty 5.0 of 10

    A pretrained FLUX diffusion model is adapted with local-window attention plus low-resolution global guidance, allowing 4K text-to-image generation from 1K-only training data at about 2x lower cost.

Reference graph

Works this paper leans on

59 extracted references · 20 canonical work pages · cited by 3 Pith papers

  1. [1]

    Frozen in time: A joint video and image encoder for end-to-end retrieval

    Max Bain, Arsha Nagrani, G ¨ul Varol, and Andrew Zisser- man. Frozen in time: A joint video and image encoder for end-to-end retrieval. In IEEE International Conference on Computer Vision, 2021. 1

    2021

  2. [2]

    Multidiffusion: Fusing diffusion paths for controlled image generation

    Omer Bar-Tal, Lior Yariv, Yaron Lipman, and Tali Dekel. Multidiffusion: Fusing diffusion paths for controlled image generation. arXiv preprint arXiv:2302.08113, 2023. 2, 3, 5

    arXiv 2023

  3. [3]

    Lumiere: A space- time diffusion model for video generation

    Omer Bar-Tal, Hila Chefer, Omer Tov, Charles Her- rmann, Roni Paiss, Shiran Zada, Ariel Ephrat, Junhwa Hur, Yuanzhen Li, Tomer Michaeli, et al. Lumiere: A space- time diffusion model for video generation. arXiv preprint arXiv:2401.12945, 2024. 2

    arXiv 2024

  4. [4]

    Demystifying mmd gans

    Mikołaj Bi ´nkowski, Danica J Sutherland, Michael Arbel, and Arthur Gretton. Demystifying mmd gans. arXiv preprint arXiv:1801.01401, 2018. 5

    arXiv 2018

  5. [5]

    Stable video diffusion: Scaling latent video diffusion models to large datasets

    Andreas Blattmann, Tim Dockhorn, Sumith Kulal, Daniel Mendelevitch, Maciej Kilian, Dominik Lorenz, Yam Levi, Zion English, Vikram V oleti, Adam Letts, et al. Stable video diffusion: Scaling latent video diffusion models to large datasets. arXiv preprint arXiv:2311.15127, 2023. 2

    arXiv 2023

  6. [6]

    Align your latents: High-resolution video synthesis with la- tent diffusion models

    Andreas Blattmann, Robin Rombach, Huan Ling, Tim Dock- horn, Seung Wook Kim, Sanja Fidler, and Karsten Kreis. Align your latents: High-resolution video synthesis with la- tent diffusion models. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 22563–22575, 2023. 2

    2023

  7. [7]

    Ap-ldm: Attentive and progressive latent diffusion model for training-free high-resolution image generation

    Boyuan Cao, Jiaxin Ye, Yujie Wei, and Hongming Shan. Ap-ldm: Attentive and progressive latent diffusion model for training-free high-resolution image generation. arXiv preprint arXiv:2410.06055, 2024. 3

    arXiv 2024

  8. [8]

    Any-resolution training for high- resolution image synthesis

    Lucy Chai, Michael Gharbi, Eli Shechtman, Phillip Isola, and Richard Zhang. Any-resolution training for high- resolution image synthesis. In European Conference on Computer Vision, pages 170–188. Springer, 2022. 5

    work page 2022

Show all 59 references
  1. [9]

    Videocrafter1: Open diffusion models for high-quality video generation

    Haoxin Chen, Menghan Xia, Yingqing He, Yong Zhang, Xiaodong Cun, Shaoshu Yang, Jinbo Xing, Yaofang Liu, Qifeng Chen, Xintao Wang, et al. Videocrafter1: Open diffusion models for high-quality video generation. arXiv preprint arXiv:2310.19512, 2023. 2

    arXiv

  2. [10]

    Videocrafter2: Overcoming data limitations for high-quality video diffusion models, 2024

    Haoxin Chen, Yong Zhang, Xiaodong Cun, Menghan Xia, Xintao Wang, Chao Weng, and Ying Shan. Videocrafter2: Overcoming data limitations for high-quality video diffusion models, 2024. 1, 2, 5, 7

    2024

  3. [11]

    Pixart-α: Fast training of dif- fusion transformer for photorealistic text-to-image synthesis,

    Junsong Chen, Jincheng Yu, Chongjian Ge, Lewei Yao, Enze Xie, Yue Wu, Zhongdao Wang, James Kwok, Ping Luo, Huchuan Lu, and Zhenguo Li. Pixart-α: Fast training of dif- fusion transformer for photorealistic text-to-image synthesis,

  4. [12]

    Resadapter: Domain consistent resolution adapter for diffu- sion models

    Jiaxiang Cheng, Pan Xie, Xin Xia, Jiashi Li, Jie Wu, Yuxi Ren, Huixia Li, Xuefeng Xiao, Min Zheng, and Lean Fu. Resadapter: Domain consistent resolution adapter for diffu- sion models. 2024. 3

    2024

  5. [13]

    Diffusion models beat gans on image synthesis

    Prafulla Dhariwal and Alexander Nichol. Diffusion models beat gans on image synthesis. Advances in neural informa- tion processing systems, 34:8780–8794, 2021. 2

    2021

  6. [14]

    Demofusion: Democratising high- resolution image generation with no $$$

    Ruoyi Du, Dongliang Chang, Timothy Hospedales, Yi-Zhe Song, and Zhanyu Ma. Demofusion: Democratising high- resolution image generation with no $$$. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 6159–6168, 2024. 2, 3, 4, 5, 7, 1

    2024

  7. [15]

    I-max: Maximize the resolu- tion potential of pre-trained rectified flow transformers with projected flow

    Ruoyi Du, Dongyang Liu, Le Zhuo, Qin Qi, Hongsheng Li, Zhanyu Ma, and Peng Gao. I-max: Maximize the resolu- tion potential of pre-trained rectified flow transformers with projected flow. arXiv preprint arXiv:2410.07536, 2024. 5

    2024 arXiv

  8. [16]

    Make a cheap scaling: A self-cascade diffusion model for higher-resolution adapta- tion

    Lanqing Guo, Yingqing He, Haoxin Chen, Menghan Xia, Xiaodong Cun, Yufei Wang, Siyu Huang, Yong Zhang, Xin- tao Wang, Qifeng Chen, et al. Make a cheap scaling: A self-cascade diffusion model for higher-resolution adapta- tion. arXiv preprint arXiv:2402.10491, 2024. 2, 3, 4

    2024 arXiv

  9. [17]

    Animatediff: Animate your personalized text-to-image diffusion models without specific tuning

    Yuwei Guo, Ceyuan Yang, Anyi Rao, Yaohui Wang, Yu Qiao, Dahua Lin, and Bo Dai. Animatediff: Animate your personalized text-to-image diffusion models without specific tuning. arXiv preprint arXiv:2307.04725, 2023. 2

    2023 arXiv

  10. [18]

    Elasticdiffusion: Training-free arbitrary size image genera- tion through global-local content separation

    Moayed Haji-Ali, Guha Balakrishnan, and Vicente Ordonez. Elasticdiffusion: Training-free arbitrary size image genera- tion through global-local content separation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 6603–6612, 2024. 3

    2024

  11. [19]

    Latent video diffusion models for high-fidelity video generation with arbitrary lengths

    Yingqing He, Tianyu Yang, Yong Zhang, Ying Shan, and Qifeng Chen. Latent video diffusion models for high-fidelity video generation with arbitrary lengths. arXiv preprint arXiv:2211.13221, 2022. 2

    2022 arXiv

  12. [20]

    Scalecrafter: Tuning-free higher- resolution visual generation with diffusion models

    Yingqing He, Shaoshu Yang, Haoxin Chen, Xiaodong Cun, Menghan Xia, Yong Zhang, Xintao Wang, Ran He, Qifeng Chen, and Ying Shan. Scalecrafter: Tuning-free higher- resolution visual generation with diffusion models. In The Twelfth International Conference on Learning Representa-...

    2024

  13. [21]

    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. Advances in neural information processing systems , 30, 2017. 5

    2017

  14. [22]

    Denoising diffu- sion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffu- sion probabilistic models. Advances in Neural Information Processing Systems, 33:6840–6851, 2020. 2

    2020

  15. [23]

    Video dif- fusion models

    Jonathan Ho, Tim Salimans, Alexey Gritsenko, William Chan, Mohammad Norouzi, and David J Fleet. Video dif- fusion models. Advances in Neural Information Processing Systems, 35:8633–8646, 2022. 2 9

    2022

  16. [24]

    sim- ple diffusion: End-to-end diffusion for high resolution im- ages

    Emiel Hoogeboom, Jonathan Heek, and Tim Salimans. sim- ple diffusion: End-to-end diffusion for high resolution im- ages. In International Conference on Machine Learning , pages 13213–13232. PMLR, 2023. 3

    2023

  17. [25]

    Fouriscale: A frequency perspective on training-free high-resolution image synthesis

    Linjiang Huang, Rongyao Fang, Aiping Zhang, Guanglu Song, Si Liu, Yu Liu, and Hongsheng Li. Fouriscale: A frequency perspective on training-free high-resolution image synthesis. arXiv preprint arXiv:2403.12963, 2024. 2, 3, 5, 7, 1

    2024 arXiv

  18. [26]

    Vbench: Comprehensive benchmark suite for video generative models

    Ziqi Huang, Yinan He, Jiashuo Yu, Fan Zhang, Chenyang Si, Yuming Jiang, Yuanhan Zhang, Tianxing Wu, Qingyang Jin, Nattapol Chanpaisit, et al. Vbench: Comprehensive benchmark suite for video generative models. arXiv preprint arXiv:2311.17982, 2023. 5

    2023 arXiv

  19. [27]

    Upsample guidance: Scale up diffusion models without training

    Juno Hwang, Yong-Hyun Park, and Junghyo Jo. Upsample guidance: Scale up diffusion models without training. arXiv preprint arXiv:2404.01709, 2024. 3

    2024 arXiv

  20. [28]

    Pyramidal flow matching for effi- cient video generative modeling

    Yang Jin, Zhicheng Sun, Ningyuan Li, Kun Xu, Kun Xu, Hao Jiang, Nan Zhuang, Quzhe Huang, Yang Song, Yadong Mu, and Zhouchen Lin. Pyramidal flow matching for effi- cient video generative modeling. 2024. 2

    2024

  21. [29]

    Training- free diffusion model adaptation for variable-sized text-to- image synthesis

    Zhiyu Jin, Xuli Shen, Bin Li, and Xiangyang Xue. Training- free diffusion model adaptation for variable-sized text-to- image synthesis. Advances in Neural Information Processing Systems, 36:70847–70860, 2023. 3

    2023

  22. [30]

    Beyondscene: Higher-resolution human-centric scene generation with pretrained diffusion

    Gwanghyun Kim, Hayeon Kim, Hoigi Seo, Dong Un Kang, and Se Young Chun. Beyondscene: Higher-resolution human-centric scene generation with pretrained diffusion. In European Conference on Computer Vision, pages 126–142. Springer, 2024. 3

    2024

  23. [31]

    Diffusehigh: Training-free progressive high- resolution image synthesis through structure guidance.arXiv preprint arXiv:2406.18459, 2024

    Younghyun Kim, Geunmin Hwang, Junyu Zhang, and Eun- byung Park. Diffusehigh: Training-free progressive high- resolution image synthesis through structure guidance.arXiv preprint arXiv:2406.18459, 2024. 3

    2024 arXiv

  24. [32]

    Segment any- thing

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C Berg, Wan-Yen Lo, et al. Segment any- thing. In Proceedings of the IEEE/CVF International Con- ference on Computer Vision, pages 4015–4026, 2023. 7

    2023

  25. [33]

    Flux.1 : An advanced state-of-the-art generative deep learning model

    Black Forest Labs. Flux.1 : An advanced state-of-the-art generative deep learning model. Technical report, Black For- est Labs, 2024. 5

    2024

  26. [34]

    Syncdiffusion: Coherent montage via synchronized joint diffusions

    Yuseung Lee, Kunho Kim, Hyunjin Kim, and Minhyuk Sung. Syncdiffusion: Coherent montage via synchronized joint diffusions. Advances in Neural Information Processing Systems, 36:50648–50660, 2023. 3

    2023

  27. [35]

    Cutdiffusion: A simple, fast, cheap, and strong diffusion extrapolation method

    Mingbao Lin, Zhihang Lin, Wengyi Zhan, Liujuan Cao, and Rongrong Ji. Cutdiffusion: A simple, fast, cheap, and strong diffusion extrapolation method. arXiv preprint arXiv:2404.15141, 2024. 3

    2024 arXiv

  28. [36]

    Accdiffusion: An accurate method for higher-resolution im- age generation

    Zhihang Lin, Mingbao Lin, Meng Zhao, and Rongrong Ji. Accdiffusion: An accurate method for higher-resolution im- age generation. arXiv preprint arXiv:2407.10738, 2024. 3

    2024 arXiv

  29. [37]

    Linfusion: 1 gpu, 1 minute, 16k image

    Songhua Liu, Weihao Yu, Zhenxiong Tan, and Xinchao Wang. Linfusion: 1 gpu, 1 minute, 16k image. 2024. 3

    2024

  30. [38]

    Hiprompt: Tuning-free higher-resolution gen- eration with hierarchical mllm prompts

    Xinyu Liu, Yingqing He, Lanqing Guo, Xiang Li, Bu Jin, Peng Li, Yan Li, Chi-Min Chan, Qifeng Chen, Wei Xue, et al. Hiprompt: Tuning-free higher-resolution gen- eration with hierarchical mllm prompts. arXiv preprint arXiv:2409.02919, 2024. 3

    2024 arXiv

  31. [39]

    Scalable diffusion models with transformers

    William Peebles and Saining Xie. Scalable diffusion models with transformers. In Proceedings of the IEEE/CVF inter- national conference on computer vision , pages 4195–4205,

  32. [40]

    Sdxl: Improving latent diffusion mod- els 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 mod- els for high-resolution image synthesis. arXiv preprint arXiv:2307.01952, 2023. 1, 2, 5, 3

    2023 arXiv

  33. [41]

    Freenoise: Tuning-free longer video diffusion via noise rescheduling.arXiv preprint arXiv:2310.15169, 2023

    Haonan Qiu, Menghan Xia, Yong Zhang, Yingqing He, Xin- tao Wang, Ying Shan, and Ziwei Liu. Freenoise: Tuning-free longer video diffusion via noise rescheduling.arXiv preprint arXiv:2310.15169, 2023. 5

    2023 arXiv

  34. [42]

    Ultrapixel: Advancing ultra-high-resolution image synthesis to new peaks

    Jingjing Ren, Wenbo Li, Haoyu Chen, Renjing Pei, Bin Shao, Yong Guo, Long Peng, Fenglong Song, and Lei Zhu. Ultrapixel: Advancing ultra-high-resolution image synthesis to new peaks. arXiv preprint arXiv:2407.02158, 2024. 3

    2024 arXiv

  35. [43]

    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. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10684–10695, 2022. 2

    2022

  36. [44]

    Improved techniques for training gans

    Tim Salimans, Ian Goodfellow, Wojciech Zaremba, Vicki Cheung, Alec Radford, and Xi Chen. Improved techniques for training gans. Advances in neural information processing systems, 29, 2016. 5

    2016

  37. [45]

    Adversarial diffusion distillation

    Axel Sauer, Dominik Lorenz, Andreas Blattmann, and Robin Rombach. Adversarial diffusion distillation. In European Conference on Computer Vision , pages 87–103. Springer,

  38. [46]

    Laion-5b: An open large-scale dataset for training next generation image-text models

    Christoph Schuhmann, Romain Beaumont, Richard Vencu, Cade Gordon, Ross Wightman, Mehdi Cherti, Theo Coombes, Aarush Katta, Clayton Mullis, Mitchell Worts- man, et al. Laion-5b: An open large-scale dataset for training next generation image-text models. Advances in Neural In- f...

    2022

  39. [47]

    Freeu: Free lunch in diffusion u-net

    Chenyang Si, Ziqi Huang, Yuming Jiang, and Ziwei Liu. Freeu: Free lunch in diffusion u-net. In CVPR, 2024. 2, 5, 7

    2024

  40. [48]

    Denoising diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502, 2020. 1

    2010 arXiv

  41. [49]

    Relay diffusion: Unifying diffusion process across resolutions for image syn- thesis

    Jiayan Teng, Wendi Zheng, Ming Ding, Wenyi Hong, Jian- qiao Wangni, Zhuoyi Yang, and Jie Tang. Relay diffusion: Unifying diffusion process across resolutions for image syn- thesis. arXiv preprint arXiv:2309.03350, 2023. 3

    2023 arXiv

  42. [50]

    To- wards accurate generative models of video: A new metric & challenges

    Thomas Unterthiner, Sjoerd van Steenkiste, Karol Kurach, Raphael Marinier, Marcin Michalski, and Sylvain Gelly. To- wards accurate generative models of video: A new metric & challenges. arXiv preprint arXiv:1812.01717, 2018. 5

    2018 arXiv

  43. [51]

    Modelscope text-to-video technical report, 2023

    Jiuniu Wang, Hangjie Yuan, Dayou Chen, Yingya Zhang, Xiang Wang, and Shiwei Zhang. Modelscope text-to-video technical report, 2023. 1, 2

    2023

  44. [52]

    Real-esrgan: Training real-world blind super-resolution with 10 pure synthetic data

    Xintao Wang, Liangbin Xie, Chao Dong, and Ying Shan. Real-esrgan: Training real-world blind super-resolution with 10 pure synthetic data. In Proceedings of the IEEE/CVF inter- national conference on computer vision , pages 1905–1914,

    1905

  45. [53]

    Videocomposer: Compositional video synthesis with motion controllability

    Xiang Wang, Hangjie Yuan, Shiwei Zhang, Dayou Chen, Ji- uniu Wang, Yingya Zhang, Yujun Shen, Deli Zhao, and Jin- gren Zhou. Videocomposer: Compositional video synthesis with motion controllability. NeurIPS, 2023. 2

    2023

  46. [54]

    Dreamvideo: Composing your dream videos with customized subject and motion

    Yujie Wei, Shiwei Zhang, Zhiwu Qing, Hangjie Yuan, Zhi- heng Liu, Yu Liu, Yingya Zhang, Jingren Zhou, and Hong- ming Shan. Dreamvideo: Composing your dream videos with customized subject and motion. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog...

    2024

  47. [55]

    Cogvideox: Text-to-video diffusion models with an expert transformer

    Zhuoyi Yang, Jiayan Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiao- han Zhang, Guanyu Feng, et al. Cogvideox: Text-to-video diffusion models with an expert transformer. arXiv preprint arXiv:2408.06072, 2024. 1, 2, 5

    2024 arXiv

  48. [56]

    Instructvideo: Instructing video dif- fusion models with human feedback

    Hangjie Yuan, Shiwei Zhang, Xiang Wang, Yujie Wei, Tao Feng, Yining Pan, Yingya Zhang, Ziwei Liu, Samuel Al- banie, and Dong Ni. Instructvideo: Instructing video dif- fusion models with human feedback. In CVPR, 2024

    2024

  49. [57]

    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. 1, 2

    2023

  50. [58]

    Hidiffusion: Unlocking higher- resolution creativity and efficiency in pretrained diffusion models

    Shen Zhang, Zhaowei Chen, Zhenyu Zhao, Yuhao Chen, Yao Tang, and Jiajun Liang. Hidiffusion: Unlocking higher- resolution creativity and efficiency in pretrained diffusion models. In European Conference on Computer Vision, pages 145–161. Springer, 2024. 3

    2024

  51. [59]

    FreeScale

    Qingping Zheng, Yuanfan Guo, Jiankang Deng, Jianhua Han, Ying Li, Songcen Xu, and Hang Xu. Any-size- diffusion: Toward efficient text-driven synthesis for any-size hd images. In Proceedings of the AAAI Conference on Arti- ficial Intelligence, pages 7571–7578, 2024. 3 11 FreeSc...

    2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.