Pith. sign in

REVIEW 4 major objections 5 minor 38 references

Accelerating Video Diffusion Models via Distribution Matching

T0 review · 4 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 A two-loss distillation of a video diffusion model produces four-step generation whose FVD and CLIPScore beat the teacher and prior 4-step methods.

desk verdict A legitimate but incremental DMD2-style approach to video distillation; the 2D frame-level SDM trick is clever, but the evaluation is too thin to support the headline claims. read the letter →

arxiv 2412.05899 v1 pith:3QX34H7O submitted 2024-12-08 cs.CV

Yuanzhi Zhu , Hanshu Yan , Huan Yang , Kai Zhang , Junnan Li This is my paper · ORCID
classification cs.CV
keywords text-to-videogenerationdiffusiondistillationdistributionmatchingscoredenoisingGANfew-stepinferenceAnimateDiffframe-level
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

This paper tries to establish that a pre-trained video diffusion model can be distilled into a four-step generator with output quality on par with, or better than, the teacher's multi-step sampling. The recipe is to match two distributions at once: an adversarial video-level GAN loss pulls the generated videos toward real video data, while a novel 2D score distribution matching (SDM) loss, computed on $K$ randomly sampled frames, ties each frame's structure and prompt adherence to a pre-trained 2D image diffusion model. The paper argues the 2D frame loss is indispensable, because a video GAN loss alone produces severe temporal distortions, identity changes, and style shifts. If the claim holds, text-to-video generation would need only four denoising steps and no classifier-free guidance, which would make interactive video generation practical.

What carries the argument

The machine that carries the argument is a two-term distillation objective. The first term, video adversarial distribution matching, uses a denoising GAN discriminator whose encoder is the frozen UNet encoder of the video teacher with a trainable multi-scale prediction head; it operates on noisy latent videos at noise levels $t$ sampled from a range, aligning the student's video distribution with real video data. The second term, 2D score distribution matching, samples $K$ random frames from the student's video, diffuses them, and estimates the gradient of the KL divergence between the student's noisy frame distribution and the 2D teacher's distribution as $w(t)(\epsilon_\phi(\tilde{x}^K_t,t) - \epsilon_\psi(\tilde{x}^K_t,t))\,dG^K_\theta(\epsilon)/d\theta$; this regulates per-frame structure, layout, and prompt following. The generator is trained on $\lambda_{\mathrm{SDM}}L_{\mathrm{SDM}} + \lambda_{\mathrm{ADM}}L_{\mathrm{ADM}}$, with the discriminator and the 2D fake model updated twice as often as the generator, and the backward simulation trick from prior distillation work is adopted to reduce the training-inference mismatch.

What would settle it

Run the same distillation twice on the same teacher and dataset, once with the full objective and once with $\lambda_{\mathrm{SDM}}=0$, then measure FVD and temporal consistency on the WebVid validation set; if the GAN-only model does not show substantially worse temporal coherence, or if a video-level SDM loss at the same compute matches the 2D variant's numbers, the paper's claim that per-frame 2D scoring is necessary for 4-step quality is falsified.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that a combination of video adversarial distribution matching and 2D score distribution matching can distill AnimateDiff into a generator that produces videos with superior frame quality relative to the teacher while requiring only 4 sampling steps at inference. The quantitative evidence is Table 1: with 4 NFEs, the 2D-SDM variant reports FVD 1271.45 and CLIPScore 32.01, beating Motion Consistency Model (1765.30, 28.60), AnimateLCM (1405.79, 28.44), and AnimateDiff-Lightning (1623.98, 29.47). The paper also claims that the 2D SDM loss is necessary, since the diffusion-GAN-only variant displays severe temporal distortions, identity changes, and a style shift, and that the 2D SDM loss can be supplied by any 2D image diffusion model sharing the latent space, allowing style transfer through the choice of 2D teacher. The result is presented as a transfer of distribution matching distillation, previously explored mostly for images, to text-to-video models.

Load-bearing premise

The load-bearing premise is that scoring a handful of randomly chosen single frames with a 2D image model is enough to keep the whole video temporally coherent, because the paper shows the adversarial video loss alone is not sufficient.

Editorial extensions

If this is right

  • Four-step inference from the distilled generator replaces the teacher's 25-step DDIM sampling with no classifier-free guidance at inference, so text-to-video generation becomes cheap enough for interactive use.
  • Because the SDM loss is computed on frames with a 2D teacher sharing the latent space, swapping the 2D teacher changes the output style; the same video backbone can be redirected toward new visual domains without retraining the motion module.
  • The paper's numbers imply that frame-level scoring, not video-level scoring, is enough for few-step video distillation: the 2D SDM variant has slightly better FVD and clearly better CLIPScore than the video SDM variant.
  • Training needs only a real-video dataset for the adversarial term; the score-matching term is data-free with respect to the teacher, so the method can tap the large ecosystem of 2D image diffusion models.
  • The two-timescale update rule and the multi-scale discriminator head carry over directly to other video teacher backbones, not just AnimateDiff.

Reading between the lines

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

  • Editorial inference: if the 2D frame loss is what secures temporal coherence, then ablating the number of sampled frames should show a plateau; at $K=1$ coherence should degrade, which would confirm that temporal coupling is implicit rather than explicitly modeled.
  • Editorial inference: the paper's own limitation notes predict that one-step distillation flickers and that diversity drops; a concrete extension is to add a diversity regularizer or a small temporal discriminator only at low noise levels while keeping the 2D SDM term, testing whether the flicker is caused by the frame-level loss or by the GAN term.
  • Editorial inference: since the 2D SDM variant improves CLIPScore much more than FVD relative to the video SDM variant, the frame loss may be aligning semantics and prompt adherence more than global motion statistics; a separate per-prompt temporal consistency metric would isolate what the 2D loss actually regulates.
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. This paper proposes AVDM2, a distribution-matching distillation framework for accelerating text-to-video diffusion models. The method combines a video adversarial distribution matching (ADM) loss, using a discriminator built on the frozen teacher encoder, with a 2D score distribution matching (SDM) loss applied on K randomly sampled frames. The student is initialized from the teacher and trained with the two losses; the 2D SDM loss is meant to regulate frame quality and prompt adherence, while the ADM loss aligns the video distribution with real data. Experiments on AnimateDiff as teacher report FVD and CLIPScore on 100 WebVid videos, showing that the 4-step student outperforms the baselines Motion Consistency Model, AnimateLCM, and AnimateDiff-Lightning on both metrics. The conclusion claims superior frame quality relative to the teacher with only 4 inference steps.

Significance. If the quantitative results are reliable, the paper offers a practical method to distill text-to-video diffusion models to 4 steps while improving per-frame quality and CLIPScore, with flexibility in choosing the 2D base model. The combination of a video GAN loss with a 2D score distribution loss is a sensible extension of image distillation methods to video, and the authors openly discuss limitations such as one-step generation difficulty and diversity loss. The paper also provides a complete algorithmic description (Algorithm 1) and detailed training details that would facilitate reproduction. However, the evidence as presented is not yet sufficient to substantiate the stronger claims of surpassing the teacher and of temporal consistency, as detailed in the major comments.

major comments (4)
  1. [Table 1 and Conclusion (Section 7)] Table 1 does not include a teacher baseline. The conclusion (Section 7) claims "superior frame quality compared to the teacher models," but the quantitative table compares only 4-step methods; no FVD/CLIPScore for AnimateDiff with 25 steps is reported. The qualitative comparison (Figure 2) shows the teacher 25-step output, but no metric is given. Please add a teacher row to Table 1 and explicitly state whether the improvement over the teacher is statistically significant.
  2. [Section 5.2 and Eq. (8)] The temporal-consistency claim is not tested. Section 5.2 states that the diffusion-GAN-alone variant exhibits "severe distortions, identity changes" and that the 2D SDM loss mitigates these issues. However, the SDM loss in Eq. (8) is a sum of per-frame score-matching terms computed on K randomly sampled frames, with no coupling between frames. The paper does not report a temporal consistency metric (e.g., frame-to-frame distance, VBench temporal score) on either the full model or the ablation. The text itself admits that "existing metrics fail to capture flickering" (Section 5.2). Without a temporal metric, the central claim that the 4-step model produces high-quality videos (rather than only high-quality frames) is not established.
  3. [Section 5.1 and Table 1] The quantitative evaluation is based on a small validation set with no error bars. Only 100 videos from WebVid10M are used for FVD and CLIPScore (Section 5.1). FVD computed on 100 samples has high variance, and the paper does not report confidence intervals or multiple seeds. Therefore, the differences reported in Table 1 (e.g., 1271.45 vs. 1405.79) may not be statistically meaningful. Please provide confidence intervals, multiple seeds, or statistical tests.
  4. [Figure 4 and Section 5.2] The ablation "Importance of SDM" is qualitative only. Figure 4 shows a few frames from the diffusion-GAN-alone variant and the full method, but no FVD/CLIPScore or temporal metric is given for the ablation. Since the GAN-alone variant is admitted to be insufficient, a quantitative comparison would strengthen the claim that the SDM term is necessary.
minor comments (5)
  1. [Section 4.2 and Algorithm 1] The notation for the SDM teacher and fake model is inconsistent: Eq. (8) uses ϵϕ and ϵψ, but the text later refers to "the SDM fake model ϵϕ" (Section 4.2) and Algorithm 1 line 14 updates "ϕ2D". Please use distinct, consistent names (e.g., ϵφ2D and ϵψ2D).
  2. [Abstract] The phrase "potentially improving—generation quality" is vague; please specify which quality is improved (e.g., frame quality, motion consistency) and avoid hedged claims in the abstract.
  3. [Table 1 and Section 5.2] The rows "Ours (with video SDM)" and "Ours (with 2D SDM)" are not clearly distinguished. Please add a sentence explaining what "video SDM" means (e.g., applying Eq. (8) to the whole video) and why the 2D version is preferred.
  4. [Eq. (7)] The word "fasion" is a typo and should be "fashion."
  5. [References] There are duplicate entries for Liu (2022) and a few other reference formatting issues; please clean up the bibliography.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: AVDM2's 4-step video distillation is trained against real data and frozen teachers, with no fitted constant relabeled as a prediction.

full rationale

The paper's derivation chain is self-contained with respect to circularity patterns. The generator is trained by Eq. (9), a weighted sum of L_ADM (Eq. 7), a denoising-GAN adversarial loss against real videos, and L_SDM (Eq. 8), the standard score-distribution-matching gradient computed on K randomly sampled frames using a frozen 2D teacher and a separately trained 2D fake model. No parameter is fitted to the reported FVD/CLIPScore values and then renamed as a prediction; Table 1 is an external comparison on a held-out WebVid validation set. The two self-citations (Yan et al. 2024; Zhu et al. 2025) appear only in a survey paragraph on rectified flows and are not load-bearing. The fact that the 2D SDM term has no explicit temporal coupling is a testable assumption about what regulates temporal consistency, not a case where Eq. X equals Eq. Y by construction or where a fitted input is reported as a prediction. No circular step can be quoted and exhibited.

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

The central claim rests on standard DMD-style machinery: a pre-trained teacher, an online fake score model, and a GAN discriminator. The main unstated choices are the loss weights, the subframe count K, and the compatibility of 2D and video latent spaces. No new physical or conceptual entities are introduced.

free parameters (3)
  • lambda_SDM and lambda_ADM loss weights
    Equation (9) combines the two losses; the paper calls them importance weights but never reports their values, so the balance is tuned by hand.
  • K, number of randomly sampled subframes
    Section 4.2 uses K random subframes for the frame SDM loss, but no value or sensitivity analysis is given; K controls how much of the video the 2D loss sees.
  • Time intervals for ADM and SDM = t ~ U[0, 0.5T] for ADM; t' ~ U[0.2T, 0.98T] for SDM
    Set by hand in Section 5.1; these intervals determine which noise levels the discriminator and 2D teacher score, and the paper notes tmax must be chosen carefully.
assumptions (3)
  • domain assumption The 2D image diffusion teacher and the video teacher share a compatible latent space, so video frame latents can be scored by the 2D model.
    Section 4.2 and Figure 1 feed video frame latents to a 2D teacher; this requires the AnimateDiff video latents to be scoreable by a Stable Diffusion-style 2D latent model, asserted without verification.
  • standard math The score distribution matching gradient (Equation 8) is a valid unbiased gradient for the KL divergence between noisy frame distributions.
    Borrowed from VSD and DMD (Equation 4); the paper does not rederive it for the video case, relying on prior derivations from Luo et al. and Yin et al.
  • domain assumption Backward simulation produces training-time sample paths that match inference-time generation paths.
    Adopted from Kohler et al. and Yin et al. in Section 4; if the training simulator and the few-step generator diverge, the distillation objective is mismatched.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Accelerating Video Diffusion Models via Distribution Matching." pith.science (2026). https://pith.science/paper/3QX34H7O

@misc{pith2026241205899,
  author       = {Pith},
  title        = {Pith review of: Accelerating Video Diffusion Models via Distribution Matching},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3QX34H7O}},
  note         = {Machine review of arXiv:2412.05899}
}
read the original abstract

Generative models, particularly diffusion models, have made significant success in data synthesis across various modalities, including images, videos, and 3D assets. However, current diffusion models are computationally intensive, often requiring numerous sampling steps that limit their practical application, especially in video generation. This work introduces a novel framework for diffusion distillation and distribution matching that dramatically reduces the number of inference steps while maintaining-and potentially improving-generation quality. Our approach focuses on distilling pre-trained diffusion models into a more efficient few-step generator, specifically targeting video generation. By leveraging a combination of video GAN loss and a novel 2D score distribution matching loss, we demonstrate the potential to generate high-quality video frames with substantially fewer sampling steps. To be specific, the proposed method incorporates a denoising GAN discriminator to distil from the real data and a pre-trained image diffusion model to enhance the frame quality and the prompt-following capabilities. Experimental results using AnimateDiff as the teacher model showcase the method's effectiveness, achieving superior performance in just four sampling steps compared to existing techniques.

Figures

Figures reproduced from arXiv: 2412.05899 by the authors.

Figure 1
Figure 1. Illustration of proposed distribution matching loss: The generator produces a video from random noise input. This generated video undergoes forward diffusion to create a noisy video, which is then input to the discriminator for GAN loss computation. Simultaneously, K random frames from the same video are diffused with noise and fed into the 2D teacher and fake model to construct the SDM loss. The discriminator is tr… view at source ↗
Figure 2
Figure 2. Comparison between our method and teacher AnimateDiff model with different sampling [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Qualitative comparison on base model AnimateDiff. From top to bottom the text prompts are: 1) a dog with big expressive eyes running in a city park; 2) a majestic horse with a long flowing tail running at a tranquil beach; 3) a red car, moving on the road, mountain, green grass and trees; 4) Origami dancers in white paper, 3D render, ultra-detailed, on white background, studio shot, dancing modern dance. 5 RESULTS A… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Comparison between our method and diffusion GAN alone training on 4 step generation. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Visual results of our method with different 2D SDM models. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 1 canonical work pages

  1. [2]

    Blattmann, T

    A. Blattmann, T. Dockhorn, S. Kulal, D. Mendelevitch, M. Kilian, D. Lorenz, Y . Levi, Z. English, V . V oleti, A. Letts, et al. Stable video diffusion: Scaling latent video diffusion models to large datasets. arXiv preprint arXiv:2311.15127, 2023a. A. Blattmann, R. Rombach, H. Ling, T. Dockhorn, S. W. Kim, S. Fidler, and K. Kreis. Align your latents: High...

    arXiv

  2. [6]

    Hessel, A

    J. Hessel, A. Holtzman, M. Forbes, R. L. Bras, and Y . Choi. Clipscore: A reference-free evaluation metric for image captioning. arXiv preprint arXiv:2104.08718,

    arXiv

  3. [7]

    J. Ho, W. Chan, C. Saharia, J. Whang, R. Gao, A. Gritsenko, D. P. Kingma, B. Poole, M. Norouzi, D. J. Fleet, et al. Imagen video: High definition video generation with diffusion models. arXiv preprint arXiv:2210.02303, 2022a. J. Ho, T. Salimans, A. Gritsenko, W. Chan, M. Norouzi, and D. J. Fleet. Video diffusion models. Advances in Neural Information Proc...

    arXiv

  4. [9]

    D. P. Kingma and M. Welling. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114,

    arXiv

  5. [11]

    Z. Kong, W. Ping, J. Huang, K. Zhao, and B. Catanzaro. Diffwave: A versatile diffusion model for audio synthesis. arXiv preprint arXiv:2009.09761,

    arXiv 2009

  6. [12]

    J. Li, W. Feng, T.-J. Fu, X. Wang, S. Basu, W. Chen, and W. Y . Wang. T2v-turbo: Breaking the quality bottleneck of video consistency model with mixed reward feedback. arXiv preprint arXiv:2405.18750,

    arXiv

  7. [13]

    Lin and X

    S. Lin and X. Yang. Animatediff-lightning: Cross-model diffusion distillation. arXiv preprint arXiv:2403.12706,

    arXiv

  8. [14]

    S. Lin, A. Wang, and X. Yang. Sdxl-lightning: Progressive adversarial diffusion distillation. arXiv preprint arXiv:2402.13929,

    arXiv

Show all 38 references
  1. [15]

    Q. Liu. Rectified flow: A marginal preserving approach to optimal transport. arXiv preprint arXiv:2209.14577,

    arXiv

  2. [17]

    Luhman and T

    E. Luhman and T. Luhman. Knowledge distillation in iterative generative models for improved sampling speed. arXiv preprint arXiv:2101.02388,

    arXiv

  3. [18]

    S. Luo, Y . Tan, L. Huang, J. Li, and H. Zhao. Latent consistency models: Synthesizing high-resolution images with few-step inference. arXiv preprint arXiv:2310.04378, 2023a. W. Luo, T. Hu, S. Zhang, J. Sun, Z. Li, and Z. Zhang. Diff-instruct: A universal approach for transfer...

    arXiv

  4. [19]

    X. Mao, Z. Jiang, F.-Y . Wang, W. Zhu, J. Zhang, H. Chen, M. Chi, and Y . Wang. Osv: One step is enough for high-quality image to video generation. arXiv preprint arXiv:2409.11367,

    arXiv

  5. [20]

    K. Nan, R. Xie, P. Zhou, T. Fan, Z. Yang, Z. Chen, X. Li, J. Yang, and Y . Tai. Openvid-1m: A large-scale high-quality dataset for text-to-video generation. arXiv preprint arXiv:2407.02371,

    arXiv

  6. [21]

    Open-sora-plan

    PKU-Yuan Lab and Tuzhan AI. Open-sora-plan. https://doi.org/10.5281/zenodo. 10948109, Apr

    doi

  7. [22]

    Podell, Z

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

    arXiv

  8. [23]

    Poole, A

    B. Poole, A. Jain, J. T. Barron, and B. Mildenhall. Dreamfusion: Text-to-3d using 2d diffusion. arXiv preprint arXiv:2209.14988,

    arXiv

  9. [24]

    Ramesh, P

    A. Ramesh, P. Dhariwal, A. Nichol, C. Chu, and M. Chen. Hierarchical text-conditional image generation with clip latents. arXiv preprint arXiv:2204.06125,

    arXiv

  10. [26]

    Singer, A

    U. Singer, A. Polyak, T. Hayes, X. Yin, J. An, S. Zhang, Q. Hu, H. Yang, O. Ashual, O. Gafni, et al. Make-a-video: Text-to-video generation without text-video data. arXiv preprint arXiv:2209.14792,

    arXiv

  11. [28]

    Unterthiner, S

    T. Unterthiner, S. Van Steenkiste, K. Kurach, R. Marinier, M. Michalski, and S. Gelly. Towards accurate generative models of video: A new metric & challenges.arXiv preprint arXiv:1812.01717,

    arXiv

  12. [30]

    13 Technical Report Z. Wang, J. Lorraine, Y . Wang, H. Su, J. Zhu, S. Fidler, and X. Zeng. Llama-mesh: Unifying 3d mesh generation with language models. arXiv preprint arXiv:2411.09595, 2024b. Z. Wang, C. Lu, Y . Wang, F. Bao, C. Li, H. Su, and J. Zhu. Prolificdreamer: High-fi...

    arXiv

  13. [31]

    Z. Yang, J. Teng, W. Zheng, M. Ding, S. Huang, J. Xu, Y . Yang, W. Hong, X. Zhang, G. Feng, et al. Cogvideox: Text-to-video diffusion models with an expert transformer. arXiv preprint arXiv:2408.06072,

    arXiv

  14. [32]

    J. Yim, B. L. Trippe, V . De Bortoli, E. Mathieu, A. Doucet, R. Barzilay, and T. Jaakkola. Se (3) diffusion model with application to protein backbone generation. arXiv preprint arXiv:2302.02277,

    arXiv

  15. [33]

    T. Yin, M. Gharbi, T. Park, R. Zhang, E. Shechtman, F. Durand, and W. T. Freeman. Improved distribution matching distillation for fast image synthesis. arXiv preprint arXiv:2405.14867, 2024a. T. Yin, M. Gharbi, R. Zhang, E. Shechtman, F. Durand, W. T. Freeman, and T. Park. One...

    arXiv

  16. [34]

    Y . Zhai, K. Lin, Z. Yang, L. Li, J. Wang, C.-C. Lin, D. Doermann, J. Yuan, and L. Wang. Motion consistency model: Accelerating video diffusion with disentangled motion-appearance distillation. arXiv preprint arXiv:2406.06890,

    arXiv

  17. [35]

    D. J. Zhang, J. Z. Wu, J.-W. Liu, R. Zhao, L. Ran, Y . Gu, D. Gao, and M. Z. Shou. Show-1: Marrying pixel and latent diffusion models for text-to-video generation. International Journal of Computer Vision, pages 1–15, 2024a. Q. Zhang and Y . Chen. Fast sampling of diffusion mo...

    arXiv

  18. [36]

    Zhang, J

    S. Zhang, J. Wang, Y . Zhang, K. Zhao, H. Yuan, Z. Qin, X. Wang, D. Zhao, and J. Zhou. I2vgen-xl: High-quality image-to-video synthesis via cascaded diffusion models. arXiv preprint arXiv:2311.04145,

    arXiv

  19. [37]

    Zhang, Y

    Z. Zhang, Y . Li, Y . Wu, Y . Xu, A. Kag, I. Skorokhodov, W. Menapace, A. Siarohin, J. Cao, D. Metaxas, et al. Sf-v: Single forward video generation model. arXiv preprint arXiv:2406.04324, 2024b. D. Zhou, W. Wang, H. Yan, W. Lv, Y . Zhu, and J. Feng. Magicvideo: Efficient vide...

    arXiv

  20. [38]

    M. Zhou, Z. Wang, H. Zheng, and H. Huang. Long and short guidance in score identity distillation for one-step text-to-image generation. arXiv preprint arXiv:2406.01561, 2024a. M. Zhou, H. Zheng, Y . Gu, Z. Wang, and H. Huang. Adversarial score identity distillation: Rapidly su...

    arXiv

  21. [1992]

    Salimans and J

    T. Salimans and J. Ho. Progressive distillation for fast sampling of diffusion models. arXiv preprint arXiv:2202.00512,

    arXiv

  22. [2013]

    Kohler, A

    J. Kohler, A. Pumarola, E. Schönfeld, A. Sanakoyeu, R. Sumbaly, P. Vajda, and A. Thabet. Imagine flash: Accelerating emu diffusion models with backward distillation. arXiv preprint arXiv:2405.05224,

    arXiv

  23. [2015]

    J. Song, C. Meng, and S. Ermon. Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502, 2020a. Y . Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, and B. Poole. Score-based generative modeling through stochastic differential equations. arXiv preprint a...

    2010 arXiv

  24. [2018]

    F.-Y . Wang, Z. Huang, X. Shi, W. Bian, G. Song, Y . Liu, and H. Li. Animatelcm: Accelerating the animation of personalized diffusion models and adapters with decoupled consistency learning. arXiv preprint arXiv:2402.00769, 2024a. J. Wang, H. Yuan, D. Chen, Y . Zhang, X. Wang,...

    arXiv

  25. [2019]

    Kim, C.-H

    11 Technical Report D. Kim, C.-H. Lai, W.-H. Liao, N. Murata, Y . Takida, T. Uesaka, Y . He, Y . Mitsufuji, and S. Ermon. Consistency trajectory models: Learning probability flow ode trajectory of diffusion.arXiv preprint arXiv:2310.02279,

    arXiv

  26. [2020]

    J. Gu, S. Zhai, Y . Zhang, L. Liu, and J. M. Susskind. Boot: Data-free distillation of denoising diffusion models with bootstrapping. In ICML 2023 Workshop on Structured Probabilistic Inference{\&} Generative Modeling,

    2023

  27. [2021]

    L. Dinh, J. Sohl-Dickstein, and S. Bengio. Density estimation using real nvp. arXiv preprint arXiv:1605.08803,

    arXiv

  28. [2022]

    X. Liu, X. Zhang, J. Ma, J. Peng, and Q. Liu. Instaflow: One step is enough for high-quality diffusion-based text-to-image generation. arXiv preprint arXiv:2309.06380,

    arXiv

  29. [2023]

    Y . Guo, C. Yang, A. Rao, Z. Liang, Y . Wang, Y . Qiao, M. Agrawala, D. Lin, and B. Dai. Animatediff: Animate your personalized text-to-image diffusion models without specific tuning. arXiv preprint arXiv:2307.04725,

    arXiv

  30. [2024]

    Arjovsky and L

    M. Arjovsky and L. Bottou. Towards principled methods for training generative adversarial networks. arXiv preprint arXiv:1701.04862,

    arXiv

Pith tools

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