Pith. sign in

REVIEW 3 major objections 5 minor 48 references

FreePCA: Integrating Consistency Information across Long-short Frames in Training-free Long Video Generation via Principal Component Analysis

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read FreePCA claims that PCA decouples global and local temporal-attention features into consistent appearance and motion intensity, and that progressively fusing them yields long videos that are both consistent and high quality without any…

desk verdict A useful training-free long video generation method whose engineering is solid but whose central 'precise decoupling' claim outruns the qualitative evidence; still deserves peer review. read the letter →

arxiv 2505.01172 v1 pith:D6ZH3EK4 submitted 2025-05-02 cs.CV

Jiangtong Tan , Hu Yu , Jie Huang , Jie Xiao , Feng Zhao This is my paper · ORCID
classification cs.CV
keywords training-freelongvideogenerationprincipalcomponentanalysistemporalattentiondiffusionmodelsconsistencyprogressivefusionnoisereschedulingtext-to-video
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

Long video generation from models trained on short clips faces a trade-off: running the model on the full long sequence keeps appearance consistent but degrades quality and motion, while stitching sliding-window short segments preserves quality but breaks consistency. The paper claims that applying principal component analysis to temporal attention features separates these two entangled signals: components with high cosine similarity between the global and local features carry consistent appearance, and the remaining components carry motion intensity. FreePCA uses that split to transplant the global consistent-appearance components into the local motion features, progressively increasing the proportion as the window slides, and reuses the mean statistics of the initial noise for extra stability. The claimed upshot is a training-free, model-agnostic method that achieves both high consistency and high quality on 64-frame continuations of 16-frame video diffusion models, with better VBench consistency and motion scores than direct sampling, FreeNoise, and FreeLong.

What carries the argument

The machinery is a PCA projection along the frame dimension of temporal-attention features. For each sliding window, the global feature computed on the full long sequence and sliced to the window, and the local feature computed on the short window, are projected with the transformation matrix P obtained from the global feature. Cosine similarity per principal component ranks the f components; the top k global components become consistency features, the remaining local components become motion features, and concatenation followed by the transpose P^T maps the fused feature back to the original space. Progressive fusion controls k = min(i, Kmax) with Kmax=3, and Reuse Mean Statistics replaces the temporal mean of later noise windows with the first window's mean. The temporal transformer in the pre-trained U-Net is the only place where the fusion happens; the first 25 of 50 DDIM steps use complete FreePCA and the last 25 use the local method.

What would settle it

Construct synthetic long videos or feature sequences with known appearance and motion labels, run FreePCA's decomposition, and check whether the top cosine-similarity components track the true appearance subspace while the discarded components track optical-flow magnitude; a quantitative mismatch when appearance is fixed or when only motion changes would refute the decoupling claim.

Watch

Extended reading notes

Core claim

The central claim is that global and local information in long-video generation can be precisely decoupled by PCA: after projecting the temporal-attention features from long-frame and short-frame branches into a shared principal component space, the top-k components ranked by cosine similarity between the two branches are consistent appearance features, while the remaining local components are motion intensity features. On this basis the paper proposes to keep the local short-frame features as the quality backbone, replace the selected local appearance components with the smoother global appearance components, and progressively raise k up to Kmax=3 as the sliding window advances, then average overlapping windows. It also replaces the mean of later noise frames with the mean of the first f frames, using a shuffle step, to improve appearance consistency. The paper reports that this training-free procedure improves subject, background, and overall consistency while also improving motion smoothness, dynamic degree, and imaging quality on VideoCrafter2 and LaVie for 64-frame generation.

Load-bearing premise

The load-bearing premise is that cosine similarity in PCA space cleanly separates appearance from motion, so the top-k global components are genuinely the appearance-consistency information rather than a heuristic mix.

Editorial extensions

If this is right

  • If the decoupling claim is right, training-free long video generation no longer has to choose between global consistency and local quality, and the same PCA fusion recipe should transfer to any diffusion model with temporal attention.
  • FreePCA upgrades 16-frame models to 64-frame generation with gains on both consistency and motion or quality metrics, so longer clips can be produced without retraining or fine-tuning.
  • Because the method keeps the local branch as the quality backbone and only swaps appearance components, it preserves the original model's motion richness rather than freezing the scene.
  • The technique extends to multi-prompt video generation and to continuing a given video via DDIM inversion, so consistency control is not limited to text-to-long-video generation.
  • The reuse of initial-noise mean statistics adds appearance consistency while avoiding the input restrictions of pure noise rescheduling, supporting richer scenes.

Reading between the lines

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

  • Not stated in the paper, but if the PCA decoupling is as general as claimed, the same component-selection rule should apply to DiT-based backbones and to much longer horizons, both of which are testable without retraining.
  • Not stated in the paper, but the top-k shared components can be read as a low-dimensional appearance manifold, which could be reused for identity-preserving editing or interpolation between generated videos.
  • Not stated in the paper, but the choice of Kmax=3 suggests that only a few principal components carry appearance consistency, implying a low-rank structure in video diffusion features that could simplify other consistency mechanisms.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes FreePCA, a training-free method for long video generation from short-video diffusion models. The method runs both a global branch (the full long frame sequence) and a local branch (sliding windows of the original short frame count) through temporal attention, projects the resulting features into a PCA space defined by the global feature, and then uses cosine similarity between global and local principal components to divide the components into "consistent appearance" features (top-k global components) and "motion intensity" features (remaining local components). These are fused progressively as the sliding window advances, and additional consistency is obtained by reusing the mean statistics of the initial noise. Experiments on VideoCrafter2 and LaVie report improvements over Direct Sampling, FreeNoise, and FreeLong on VBench metrics, together with ablations on Kmax, PCA removal, random selection, fixed k, and mean-statistics reuse.

Significance. If the central decoupling claim is correct, FreePCA would be a simple and appealing training-free recipe for combining global appearance consistency with local motion quality in long-video generation, with plausible extensions to multi-prompt and continuing generation. The paper has real strengths: the code is released, the method is evaluated with independent VBench metrics on two base models, and the ablations in Table 2 show that random component selection, fixed k, and removal of the proposed components all degrade performance, which is nontrivial supporting evidence for the design. The main weakness is that the paper's headline claim -- that PCA "precisely decouples" appearance and motion -- rests on qualitative visualizations and an ad hoc PSNR threshold rather than any quantitative validation against ground-truth motion or appearance, and the empirical results are reported from a single run with no variance estimates. The contribution is therefore plausible but not yet fully established.

major comments (3)
  1. [Sec. 3, Eqs. (5)-(8)] The central claim that PCA decouples global and local information into "consistent appearance" and "motion intensity" is not quantitatively validated. The evidence in Sec. 3 consists of edge-map visualizations (Figs. 2-3) and a PSNR>35 dB threshold for counting "consistency components"; neither measures whether the components selected by cosine similarity actually correspond to appearance stability versus motion. Because the "consistent appearance" construct is defined by the same cosine-similarity rule used in the method, the argument has a circular element. I ask for a component-level validation: for example, compute optical flow or temporal variance for each PCA component and correlate those quantities with the cosine-similarity ranking, or use synthetic videos with known appearance/motion separation. Without such evidence, the labels assigned in Eqs. (7)-(8) are asserted, and the observed gains could stem from the progressive fusion schedule or mean-statistics reuse rather than from the PCA selection.
  2. [Tables 1 and 2] All quantitative results appear to come from a single run with no error bars or significance tests. The key margins (e.g., Subject Consistency 95.54 vs. 93.77, Dynamic Degree 59.72 vs. 52.77 in Table 1) may be within run-to-run variance, so the "substantial improvements" claim is not yet statistically grounded. In addition, Kmax=3 and the 25-step FreePCA/25-step local schedule are selected on the same 326 VBench prompts used for the headline comparison, which risks selection bias. Please report results over multiple seeds or random prompt subsets, perform a significance test, and either use a held-out prompt set for hyperparameter selection or show a sensitivity curve for Kmax and the step schedule.
  3. [Sec. 5.2, Table 1] The experimental comparison omits FIFO-Diffusion [26], which is described in the related work as another training-free long-video generation method. Given the paper's claim of state-of-the-art results, the comparison against only Direct Sampling, FreeNoise, and FreeLong is insufficient. Please add this baseline to the comparison or explicitly justify its exclusion (for example, if it targets a different setting or has different requirements).
minor comments (5)
  1. [Sec. 4.1] The scaling factor is written as "λ = plogfF" in the text; this is not a standard formula and appears to be a typo. Please write the expression properly (likely λ = sqrt(log(f/F))) and define all symbols clearly.
  2. [Fig. 2(d)] The percentages in Fig. 2(d) (e.g., 37%, 63%, 64%, 36%) are not explained in the caption or body text. Please specify what the two groups are and how the percentages are computed over the 100 prompts.
  3. [Eqs. (13)-(14)] The index ranges in Eqs. (13)-(14) are ambiguous: the first shuffled segment is written as ϵ′_{f+1:2f+1}, which could denote f+1 frames depending on the slice convention. Please clarify whether Python-style slicing or mathematical inclusive indexing is used, and adjust the ranges accordingly.
  4. [Sec. 4 and Table 2] The ablation configurations (1)-(6) in Table 2 are not mapped to concrete method variants in the text. Please add a sentence or a table column that describes each configuration (e.g., "no PCA", "random selection", "fixed k=3", "direct reuse without shuffling") so the reader can interpret the ablation without consulting the supplementary material.
  5. [Sec. 1, Sec. 4] The term "train-free" appears in a few places (e.g., Sec. 4) while the abstract and title use "training-free"; please use one consistent term throughout.

Circularity Check

1 steps flagged · score 3.0 of 10

Mild self-definitional circularity: 'consistent appearance' is defined by the cosine-similarity selection itself, though the final VBench evaluation is independent.

  1. self definitional [Section 4.1, Eq. (7); cf. Section 3, Fig. 3 and Fig. 2d]
    "To extract consistency features, we compare the cosine similarity of each component of global and local features in the principal component space and select the topk most similar components from the zi global, treating their corresponding features as consistency features ... z_i_con = z_i_global[n(1), n(2), ..., n(k)]."

    The paper's central claim is that PCA 'precisely decouples' global/local features into 'consistent appearance and motion intensity information.' Operationally, Eq. (7) defines the consistency feature as the top-k components of the global feature by cosine similarity against the local feature, and Eq. (8) defines the motion feature as the remaining local components. Thus the semantic labels 'consistent appearance' and 'motion intensity' are assigned by the same selection rule that constitutes the method, rather than by any independent ground-truth measure (e.g., optical flow, background masks, temporal variance). The support offered (Fig. 3 edge/frame-difference visualizations, PSNR>35 dB counting in Fig. 2d) is qualitative and heuristic.

full rationale

The paper does not rely on any load-bearing self-citation: the PCA background-subtraction motivation [1,10,11], FreeNoise [31], FreeLong [28], and the mean-statistics observation [43] are all external prior work. The final evaluation uses VBench, an external benchmark, and ablations (random component selection, fixed k, no PCA) show that the cosine-similarity selection and progressive fusion contribute to the reported scores, so the method is not a pure tautology. The only circular element is the central 'decoupling' claim: the paper defines consistency features as the top-k highest-cosine-similarity global components (Eq. 7) and motion features as the remaining local components (Eq. 8), then 'reveals' that this procedure precisely separates consistent appearance from motion intensity. That semantic labeling is validated only by qualitative visualizations and a PSNR>35 dB heuristic, not by independent ground-truth measures of appearance or motion; the claim is therefore partly a restatement of the construction. The choice Kmax=3 is selected from the same VBench benchmark (Table 2), which is a hyperparameter-tuning concern rather than a circular derivation.

Assumptions & free parameters 3 free parameters · 5 assumptions · 2 invented entities

The central result depends on several method-internal constructs and tuned hyperparameters. The main free parameters, Kmax and the step schedule, are selected on the evaluation benchmark, and the decoupling premise is asserted from visualizations rather than derived. No new physical entities are introduced; the consistency feature and motion intensity feature are algorithmic roles.

free parameters (3)
  • Kmax = 3
    Upper bound on consistency components selected per window (Eq. 9). Chosen via ablation on the VBench evaluation set (Table 2); no held-out validation.
  • FreePCA active steps = 25 of 50 DDIM steps
    FreePCA is used for the first 25 denoising steps and the local method for the last 25 (Sec. 4.2). This split is justified by a citation to [6] but not ablated.
  • Scaling factor lambda = sqrt(log(F)/log(f))
    Applied to amplify the global query in temporal attention (Sec. 4.1). Formula based on attention entropy [25]; details deferred to the supplementary.
assumptions (5)
  • domain assumption PCA can linearly separate video features into consistent appearance and motion intensity components in the temporal dimension.
    Underlies the entire method; supported only by qualitative visualizations and heuristic PSNR/Canny analysis in Sec. 3, not by a formal model.
  • domain assumption The principal component basis computed from global features is appropriate for projecting local features.
    Eq. (3-4) compute P from x_global and apply it to x_local; assumes both lie in the same subspace.
  • domain assumption Temporal attention output features carry appearance and motion information in a form where cosine similarity between global and local components measures consistency.
    Used in Eq. (5-8) to select consistency features.
  • domain assumption Diffusion models generate global layout early and fine details late [6].
    Motivates using FreePCA for only the first 25 of 50 steps (Sec. 4.2).
  • domain assumption VBench automatic metrics are valid proxies for video consistency and quality.
    Used in Sec. 5.1; no human evaluation is reported.
invented entities (2)
  • Consistent appearance feature (top-k high-cosine-similarity PCA components)
    purpose: Carries stable appearance information from global features into local features during fusion.
    The construct is defined by the method's own cosine-similarity rule (Eq. 7). The paper offers an ablation with random selection, but no external benchmark or ground-truth labels confirm that these components correspond to appearance consistency.
  • Motion intensity feature (remaining local PCA components)
    purpose: Preserves motion information from the local feature.
    Similarly defined by the residual components after removing high-similarity ones (Eq. 8); no external criterion independent of the method validates this semantic label.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FreePCA: Integrating Consistency Information across Long-short Frames in Training-free Long Video Generation via Principal Component Analysis." pith.science (2026). https://pith.science/paper/D6ZH3EK4

@misc{pith2026250501172,
  author       = {Pith},
  title        = {Pith review of: FreePCA: Integrating Consistency Information across Long-short Frames in Training-free Long Video Generation via Principal Component Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/D6ZH3EK4}},
  note         = {Machine review of arXiv:2505.01172}
}
read the original abstract

Long video generation involves generating extended videos using models trained on short videos, suffering from distribution shifts due to varying frame counts. It necessitates the use of local information from the original short frames to enhance visual and motion quality, and global information from the entire long frames to ensure appearance consistency. Existing training-free methods struggle to effectively integrate the benefits of both, as appearance and motion in videos are closely coupled, leading to motion inconsistency and visual quality. In this paper, we reveal that global and local information can be precisely decoupled into consistent appearance and motion intensity information by applying Principal Component Analysis (PCA), allowing for refined complementary integration of global consistency and local quality. With this insight, we propose FreePCA, a training-free long video generation paradigm based on PCA that simultaneously achieves high consistency and quality. Concretely, we decouple consistent appearance and motion intensity features by measuring cosine similarity in the principal component space. Critically, we progressively integrate these features to preserve original quality and ensure smooth transitions, while further enhancing consistency by reusing the mean statistics of the initial noise. Experiments demonstrate that FreePCA can be applied to various video diffusion models without requiring training, leading to substantial improvements. Code is available at https://github.com/JosephTiTan/FreePCA.

Figures

Figures reproduced from arXiv: 2505.01172 by the authors.

Figure 1
Figure 1. Illustration of different training-free methods for generating long videos. (a) Global aligned method, which inputs the entire [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. (a-c) Illustration of consistency information extraction [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Visualization of consistency features extracted in the [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Overview of our method. For noise initialization, we extend short initial noise into long initial noise using a reuse mean statistics [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Illustration of Progressive Fusion. As the window slides, [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Qualitative comparison using VideoCrafter2 as base model. Direct sampling leads to a loss of detail and semantics, while [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Qualitative comparison using LaVie as base model. Direct sampling results in content blurriness and slow motion, while [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Result of multi-prompt video generation. [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Result of continuing generation based on a given video. [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 18 canonical work pages

  1. [26]

    Fifo-diffusion: Generating infinite videos from text without training

    Jihwan Kim, Junoh Kang, Jinyoung Choi, and Bohyung Han. Fifo-diffusion: Generating infinite videos from text without training. arXiv preprint arXiv:2405.11473, 2024. 3

    arXiv 2024

  2. [1]

    Learning Spatial-Temporal Regularized Tensor Sparse RPCA for Background Subtraction

    Basit Alawode and Sajid Javed. Learning spatial-temporal regularized tensor sparse rpca for background subtraction. arXiv preprint arXiv:2309.15576, 2023. 2, 3

    work page Pith review arXiv 2023

  3. [2]

    Canny edge detec- tion enhancement by scale multiplication

    Paul Bao, Lei Zhang, and Xiaolin Wu. Canny edge detec- tion enhancement by scale multiplication. IEEE Transac- tions on Pattern Analysis and Machine Intelligence , 27(9): 1485–1490, 2005. 3

    work page 2005

  4. [3]

    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. 3

    arXiv 2023

  5. [4]

    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. 3

    2023

  6. [5]

    Generating long videos of dynamic scenes

    Tim Brooks, Janne Hellsten, Miika Aittala, Ting-Chun Wang, Timo Aila, Jaakko Lehtinen, Ming-Yu Liu, Alexei Efros, and Tero Karras. Generating long videos of dynamic scenes. Advances in Neural Information Processing Systems, 35:31769–31781, 2022. 3

    2022

  7. [6]

    Masactrl: Tuning-free mu- tual self-attention control for consistent image synthesis and editing

    Mingdeng Cao, Xintao Wang, Zhongang Qi, Ying Shan, Xi- aohu Qie, and Yinqiang Zheng. Masactrl: Tuning-free mu- tual self-attention control for consistent image synthesis and editing. In Proceedings of the IEEE/CVF International Con- ference on Computer Vision, pages 22560–22570, 2023. 6

    work page 2023

  8. [7]

    Videocrafter2: Overcoming data limitations for high-quality video diffu- sion models

    Haoxin Chen, Yong Zhang, Xiaodong Cun, Menghan Xia, Xintao Wang, Chao Weng, and Ying Shan. Videocrafter2: Overcoming data limitations for high-quality video diffu- sion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 7310– 7320, 2024. 1, 3, 6

    work page 2024

Show all 48 references
  1. [8]

    Slicedit: Zero- shot video editing with text-to-image diffusion models us- ing spatio-temporal slices

    Nathaniel Cohen, Vladimir Kulikov, Matan Kleiner, Inbar Huberman-Spiegelglas, and Tomer Michaeli. Slicedit: Zero- shot video editing with text-to-image diffusion models us- ing spatio-temporal slices. arXiv preprint arXiv:2405.12211,

    arXiv

  2. [9]

    Irc- gan: Introspective recurrent convolutional gan for text-to- video generation

    Kangle Deng, Tianyi Fei, Xin Huang, and Yuxin Peng. Irc- gan: Introspective recurrent convolutional gan for text-to- video generation. In International Joint Conference on Arti- ficial Intelligence, pages 2216–2222, 2019. 3

    2019

  3. [10]

    Block- sparse rpca for salient motion detection

    Zhi Gao, Loong-Fah Cheong, and Yu-Xiang Wang. Block- sparse rpca for salient motion detection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 36(10):1975– 1987, 2014. 2, 3

    1975

  4. [11]

    Robust low-rank tensor recovery: Models and algorithms

    Donald Goldfarb and Zhiwei Qin. Robust low-rank tensor recovery: Models and algorithms. SIAM Journal on Matrix Analysis and Applications, 35(1):225–253, 2014. 2, 3

    2014

  5. [12]

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

    Yuwei Guo, Ceyuan Yang, Anyi Rao, Zhengyang Liang, Yaohui Wang, Yu Qiao, Maneesh Agrawala, Dahua Lin, and Bo Dai. Animatediff: Animate your personalized text- to-image diffusion models without specific tuning. arXiv preprint arXiv:2307.04725, 2023. 1, 3

    2023 arXiv

  6. [13]

    Sparsectrl: Adding sparse controls to text-to-video diffusion models

    Yuwei Guo, Ceyuan Yang, Anyi Rao, Maneesh Agrawala, Dahua Lin, and Bo Dai. Sparsectrl: Adding sparse controls to text-to-video diffusion models. In European Conference on Computer Vision, pages 330–348. Springer, 2025. 1

    2025

  7. [14]

    Flexible diffusion modeling of long videos

    William Harvey, Saeid Naderiparizi, Vaden Masrani, Chris- tian Weilbach, and Frank Wood. Flexible diffusion modeling of long videos. Advances in Neural Information Processing Systems, 35:27953–27965, 2022. 3

    2022

  8. [15]

    Cameractrl: Enabling camera control for text-to-video generation

    Hao He, Yinghao Xu, Yuwei Guo, Gordon Wetzstein, Bo Dai, Hongsheng Li, and Ceyuan Yang. Cameractrl: Enabling camera control for text-to-video generation. arXiv preprint arXiv:2404.02101, 2024. 1

    2024 arXiv

  9. [16]

    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, 2(3):4, 2022. 3

    2022 arXiv

  10. [17]

    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. 1, 3

    2020

  11. [18]

    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. 1, 3

    2022

  12. [19]

    Image quality metrics: Psnr vs

    Alain Hore and Djemel Ziou. Image quality metrics: Psnr vs. ssim. In 2010 20th International Conference on Pattern Recognition, pages 2366–2369. IEEE, 2010. 3

    2010

  13. [20]

    Vbench: Comprehensive bench- mark 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 bench- mark suite for video generative models. In Proceedings of the IEEE/CVF Conference on Computer Vision and P...

    2024

  14. [21]

    Structural and statistical texture knowledge distillation for semantic segmentation

    Deyi Ji, Haoran Wang, Mingyuan Tao, Jianqiang Huang, Xian-Sheng Hua, and Hongtao Lu. Structural and statistical texture knowledge distillation for semantic segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 16876–16885, 2022. 3

    2022

  15. [22]

    Ultra-high resolution segmentation with ultra-rich con- text: A novel benchmark

    Deyi Ji, Feng Zhao, Hongtao Lu, Mingyuan Tao, and Jieping Ye. Ultra-high resolution segmentation with ultra-rich con- text: A novel benchmark. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 23621–23630, 2023. 3

    2023

  16. [23]

    Ppt- former: Pseudo multi-perspective transformer for uav seg- mentation

    Deyi Ji, Wenwei Jin, Hongtao Lu, and Feng Zhao. Ppt- former: Pseudo multi-perspective transformer for uav seg- mentation. International Joint Conference on Artificial In- telligence, pages 893–901, 2024. 3

    2024

  17. [24]

    Discrete latent perspective learning for seg- mentation and detection

    Deyi Ji, Feng Zhao, Lanyun Zhu, Wenwei Jin, Hongtao Lu, and Jieping Ye. Discrete latent perspective learning for seg- mentation and detection. InInternational Conference on Ma- chine Learning, pages 21719–21730, 2024. 3

    2024

  18. [25]

    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. 5

    2023

  19. [27]

    Video-p2p: Video editing with cross-attention control

    Shaoteng Liu, Yuechen Zhang, Wenbo Li, Zhe Lin, and Jiaya Jia. Video-p2p: Video editing with cross-attention control. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8599–8608, 2024. 1

    2024

  20. [28]

    Freelong: Training-free long video generation with spectralblend tem- poral attention

    Yu Lu, Yuanzhi Liang, Linchao Zhu, and Yi Yang. Freelong: Training-free long video generation with spectralblend tem- poral attention. arXiv preprint arXiv:2407.19918, 2024. 1, 3, 4, 6, 7

    2024 arXiv

  21. [29]

    Sync-draw: Automatic video generation using deep recurrent attentive architectures

    Gaurav Mittal, Tanya Marwah, and Vineeth N Balasubrama- nian. Sync-draw: Automatic video generation using deep recurrent attentive architectures. In Proceedings of the 25th ACM international conference on Multimedia , pages 1096– 1104, 2017. 3

    2017

  22. [30]

    To create what you tell: Generating videos from cap- tions

    Yingwei Pan, Zhaofan Qiu, Ting Yao, Houqiang Li, and Tao Mei. To create what you tell: Generating videos from cap- tions. In Proceedings of the 25th ACM international confer- ence on Multimedia, pages 1789–1798, 2017. 3

    2017

  23. [31]

    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. 1, 3, 4, 6, 7

    2023 arXiv

  24. [32]

    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. 1

    2022

  25. [33]

    Make-a-video: Text-to-video generation without text-video data

    Uriel Singer, Adam Polyak, Thomas Hayes, Xi Yin, Jie An, Songyang Zhang, Qiyuan Hu, Harry Yang, Oron Ashual, Oran Gafni, et al. Make-a-video: Text-to-video generation without text-video data. arXiv preprint arXiv:2209.14792 ,

    arXiv

  26. [34]

    Stylegan-v: A continuous video generator with the price, image quality and perks of stylegan2

    Ivan Skorokhodov, Sergey Tulyakov, and Mohamed Elho- seiny. Stylegan-v: A continuous video generator with the price, image quality and perks of stylegan2. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 3626–3636, 2022. 3

    2022

  27. [35]

    Consistent view synthe- sis with pose-guided diffusion models

    Hung-Yu Tseng, Qinbo Li, Changil Kim, Suhib Alsisan, Jia- Bin Huang, and Johannes Kopf. Consistent view synthe- sis with pose-guided diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 16773–16783, 2023. 3

    2023

  28. [36]

    Neural discrete representation learning

    Aaron Van Den Oord, Oriol Vinyals, et al. Neural discrete representation learning. arXiv preprint arXiv:1711.00937 ,

    arXiv

  29. [37]

    Mcvd-masked conditional video diffusion for prediction, generation, and interpolation

    Vikram V oleti, Alexia Jolicoeur-Martineau, and Chris Pal. Mcvd-masked conditional video diffusion for prediction, generation, and interpolation. Advances in Neural Informa- tion Processing Systems, 35:23371–23385, 2022. 3

    2022

  30. [38]

    Gen-l-video: Multi-text to long video generation via temporal co-denoising

    Fu-Yun Wang, Wenshuo Chen, Guanglu Song, Han-Jia Ye, Yu Liu, and Hongsheng Li. Gen-l-video: Multi-text to long video generation via temporal co-denoising. arXiv preprint arXiv:2305.18264, 2023. 1, 3

    2023 arXiv

  31. [39]

    Videocomposer: Compositional video synthesis with motion controllability

    Xiang Wang, Hangjie Yuan, Shiwei Zhang, Dayou Chen, Jiuniu Wang, Yingya Zhang, Yujun Shen, Deli Zhao, and Jingren Zhou. Videocomposer: Compositional video synthesis with motion controllability. arXiv preprint arXiv:2306.02018, 2024. 1

    2024 arXiv

  32. [40]

    Lavie: High-quality video gener- ation with cascaded latent diffusion models

    Yaohui Wang, Xinyuan Chen, Xin Ma, Shangchen Zhou, Ziqi Huang, Yi Wang, Ceyuan Yang, Yinan He, Jiashuo Yu, Peiqing Yang, et al. Lavie: High-quality video gener- ation with cascaded latent diffusion models. arXiv preprint arXiv:2309.15103, 2023. 1, 6

    2023 arXiv

  33. [41]

    Godiva: Gen- erating open-domain videos from natural descriptions

    Chenfei Wu, Lun Huang, Qianxi Zhang, Binyang Li, Lei Ji, Fan Yang, Guillermo Sapiro, and Nan Duan. Godiva: Gen- erating open-domain videos from natural descriptions. arXiv preprint arXiv:2104.14806, 2021. 3

    2021 arXiv

  34. [42]

    Tune-a-video: One-shot tuning of image diffusion models for text-to-video generation

    Jay Zhangjie Wu, Yixiao Ge, Xintao Wang, Stan Weixian Lei, Yuchao Gu, Yufei Shi, Wynne Hsu, Ying Shan, Xiaohu Qie, and Mike Zheng Shou. Tune-a-video: One-shot tuning of image diffusion models for text-to-video generation. In Proceedings of the IEEE/CVF International Conference...

    2023

  35. [43]

    Video diffusion models are training-free motion interpreter and controller

    Zeqi Xiao, Yifan Zhou, Shuai Yang, and Xingang Pan. Video diffusion models are training-free motion interpreter and controller. arXiv preprint arXiv:2405.14864, 2024. 1, 6

    2024 arXiv

  36. [44]

    Nuwa-xl: Diffusion over diffusion for extremely long video generation.arXiv preprint arXiv:2303.12346, 2023

    Shengming Yin, Chenfei Wu, Huan Yang, Jianfeng Wang, Xiaodong Wang, Minheng Ni, Zhengyuan Yang, Linjie Li, Shuguang Liu, Fan Yang, et al. Nuwa-xl: Diffusion over diffusion for extremely long video generation.arXiv preprint arXiv:2303.12346, 2023. 1, 3

    2023 arXiv

  37. [45]

    Make pixels dance: High- dynamic video generation

    Yan Zeng, Guoqiang Wei, Jiani Zheng, Jiaxin Zou, Yang Wei, Yuchen Zhang, and Hang Li. Make pixels dance: High- dynamic video generation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 8850–8860, 2024. 3

    2024

  38. [46]

    Show-1: Marrying pixel and latent dif- fusion models for text-to-video generation

    David Junhao Zhang, Jay Zhangjie Wu, Jia-Wei Liu, Rui Zhao, Lingmin Ran, Yuchao Gu, Difei Gao, and Mike Zheng Shou. Show-1: Marrying pixel and latent dif- fusion models for text-to-video generation. arXiv preprint arXiv:2309.15818, 2023. 3

    2023 arXiv

  39. [47]

    Magicvideo: Efficient video generation with latent diffusion models

    Daquan Zhou, Weimin Wang, Hanshu Yan, Weiwei Lv, Yizhe Zhu, and Jiashi Feng. Magicvideo: Efficient video generation with latent diffusion models. arXiv preprint arXiv:2211.11018, 2022. 1, 3

    2022 arXiv

  40. [48]

    Storydiffusion: Consistent self- attention for long-range image and video generation

    Yupeng Zhou, Daquan Zhou, Ming-Ming Cheng, Jiashi Feng, and Qibin Hou. Storydiffusion: Consistent self- attention for long-range image and video generation. arXiv preprint arXiv:2405.01434, 2024. 3

    2024 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.