Eulerian Motion Guidance: Robust Image Animation via Bidirectional Geometric Consistency

Chunyan Miao; Cong-Duy Nguyen; Khoi M. Le; Luu Anh Tuan; See-kiong Ng; Thong Nguyen

arxiv: 2605.06280 · v4 · pith:SC4LSGB3new · submitted 2026-05-07 · 💻 cs.CV

Eulerian Motion Guidance: Robust Image Animation via Bidirectional Geometric Consistency

Thong Nguyen , Khoi M. Le , Cong-Duy Nguyen , Luu Anh Tuan , See-Kiong Ng , Chunyan Miao This is my paper

Pith reviewed 2026-06-30 23:23 UTC · model grok-4.3

classification 💻 cs.CV

keywords image animationEulerian motion fieldsbidirectional geometric consistencyoptical flowdiffusion modelsocclusion maskingtemporal coherencewarping objectives

0 comments

The pith

Adjacent-frame Eulerian motion fields plus bidirectional cycle checks let diffusion models train in parallel while masking occlusions to cut animation artifacts.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tries to show that replacing global Lagrangian optical flow with short-hop Eulerian motion fields between adjacent frames gives bounded-error supervision and allows parallel training. It further claims that a forward-backward geometric consistency check can mathematically locate and mask occluded regions so the model does not learn wrong warping targets. If both pieces work, the result is faster training, better frame-to-frame coherence, and fewer dynamic errors than methods that rely on a single reference frame. A sympathetic reader would care because current image-animation pipelines often suffer from drift and slow sequential training when motion signals accumulate over long sequences.

Core claim

The paper claims that Eulerian motion fields computed between adjacent frames supply local supervision signals that support parallelized training and keep error bounds tight throughout generation, while the Bidirectional Geometric Consistency mechanism uses a forward-backward cycle check to identify and mask occluded regions, thereby preventing the model from optimizing incorrect warping objectives.

What carries the argument

Bidirectional Geometric Consistency mechanism, which performs a forward-backward cycle check on adjacent-frame Eulerian motion fields to mathematically detect and mask occluded regions.

If this is right

Training becomes parallelizable across frames instead of sequential.
Supervision error stays bounded at each short temporal hop.
Occluded regions are explicitly masked so incorrect warping objectives are not learned.
Output animations exhibit preserved temporal coherence and fewer dynamic artifacts than reference-based baselines.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same local supervision pattern could be chained across longer video lengths without the error accumulation typical of single-reference Lagrangian methods.
The cycle-check masking step might transfer to other motion-conditioned generative tasks that currently suffer from occlusion-induced drift.

Load-bearing premise

The forward-backward cycle check reliably identifies occluded regions without introducing new errors or masking valid motion information.

What would settle it

Generate animations on test videos containing known large occlusions; if the cycle-masked regions do not align with actual occluded areas or if visible warping artifacts still appear inside the unmasked motion regions, the central claim is falsified.

Figures

Figures reproduced from arXiv: 2605.06280 by Chunyan Miao, Cong-Duy Nguyen, Khoi M. Le, Luu Anh Tuan, See-kiong Ng, Thong Nguyen.

**Figure 1.** Figure 1: Long-horizon qualitative comparison. We show the reference image and generated frames at view at source ↗

**Figure 2.** Figure 2: Eulerian Motion Guidance with Bidirectional Geometric Consistency. Given a reference image and sparse motion view at source ↗

**Figure 3.** Figure 3: Occlusion masking from bidirectional cycle energy. view at source ↗

**Figure 4.** Figure 4: Qualitative comparison with ImageConductor [ view at source ↗

**Figure 5.** Figure 5: Qualitative comparison on keypoint-based animation. We compare our Eulerian Motion Guidance against the state-of view at source ↗

**Figure 6.** Figure 6: Qualitative ablation of geometric consistency. We compare training with (a) no consistency enforcement, (b) forward view at source ↗

**Figure 7.** Figure 7: Training Efficiency Analysis. comparison of per view at source ↗

**Figure 8.** Figure 8: Robustness to Large Displacement. We compare view at source ↗

**Figure 9.** Figure 9: Extended Qualitative Evaluation on Landmark view at source ↗

**Figure 9.** Figure 9: Extended Qualitative Evaluation on Landmark [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗

read the original abstract

Recent advancements in image animation have utilized diffusion models to breathe life into static images. However, existing controllable frameworks typically rely on Lagrangian motion guidance, where optical flow is estimated relative to the initial frame. This paper revisits the same optical-flow primitive through a more local supervision design: we use adjacent-frame Eulerian motion fields to guide generation, where the motion signal always describes a short temporal hop. This shift enables parallelized training and provides bounded-error supervision throughout the generation process. To mitigate the drift artifacts common in adjacent frame generation, we introduce a Bidirectional Geometric Consistency mechanism, which computes a forward-backward cycle check to mathematically identify and mask occluded regions, preventing the model from learning incorrect warping objectives. Extensive experiments demonstrate that our approach accelerates training, preserves temporal coherence, and reduces dynamic artifacts compared to reference-based baselines.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper swaps Lagrangian optical flow for adjacent-frame Eulerian fields plus a bidirectional cycle mask, which looks like a practical tweak for parallel training in diffusion animation but rests on an untested assumption about what the cycle check actually catches.

read the letter

The main move here is replacing the usual reference-frame Lagrangian guidance with short-hop Eulerian motion fields between adjacent frames. That change is meant to let training run in parallel and keep supervision error bounded instead of accumulating. They add a forward-backward cycle check to mask regions that fail consistency, on the idea that this catches occlusions and stops the model from learning bad warps. If the cycle check works as described, it could cut down on drift and artifacts while speeding things up.

What stands out is the explicit contrast with prior Lagrangian setups and the claim that the Eulerian design gives bounded-error signals throughout generation. The bidirectional mask is presented as a mathematical fix rather than another learned component, which keeps the method relatively lightweight.

The soft spot is the occlusion identification step. The stress-test note is right to flag that cycle inconsistency can come from motion estimation errors (aperture problems, lighting shifts) as easily as from true occlusions. If the mask removes valid motion signals, the bounded-error guarantee weakens and the training benefit shrinks. The abstract says extensive experiments back the claims, but without numbers, ablations, or details on how they validated the mask, it is hard to judge whether the mechanism delivers what it promises.

This is a targeted technique paper inside controllable diffusion animation. A reader already working on motion-guided video generation would get the most out of it; someone outside that niche would not. The idea is coherent on its own terms and cites the relevant line of work, so it clears the bar for a serious referee even if the experiments need tightening.

Referee Report

2 major / 0 minor

Summary. The manuscript proposes Eulerian Motion Guidance for image animation with diffusion models. It replaces Lagrangian motion guidance (optical flow from the initial frame) with adjacent-frame Eulerian motion fields for local, short-hop supervision. This design is claimed to enable parallelized training and bounded-error supervision. A Bidirectional Geometric Consistency mechanism is introduced that performs a forward-backward cycle check to mathematically identify and mask occluded regions, thereby preventing the model from learning incorrect warping objectives. The paper asserts that the approach accelerates training, preserves temporal coherence, and reduces dynamic artifacts relative to reference-based baselines.

Significance. If validated, the shift to adjacent-frame Eulerian fields could meaningfully improve training efficiency via parallelization while supplying more stable supervision signals than long-range Lagrangian guidance. The Bidirectional Geometric Consistency idea addresses a known source of drift in adjacent-frame generation and, if the cycle check reliably isolates occlusions, would constitute a useful technical contribution to motion-guided animation. The emphasis on bounded-error supervision is a clear conceptual strength.

major comments (2)

[Bidirectional Geometric Consistency] Bidirectional Geometric Consistency section: The central claim that the forward-backward cycle check 'mathematically identifies and masks occluded regions' to prevent incorrect warping objectives rests on the assumption that any cycle inconsistency is caused exclusively by occlusion. The manuscript provides no analysis or experiments showing that estimation errors in the Eulerian motion fields (e.g., aperture problems, lighting variation) do not produce false-positive masks that remove valid adjacent-frame motion signals, which would undermine the bounded-error supervision guarantee.
[Experiments] Experimental evaluation: The abstract states that 'extensive experiments demonstrate' accelerated training, preserved coherence, and reduced artifacts, yet the provided manuscript text supplies no quantitative metrics, ablation tables, or implementation details (e.g., datasets, baselines, or error measures). This absence makes it impossible to verify whether the proposed mechanisms actually deliver the claimed benefits.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful and constructive comments. We address each major point below and outline the revisions we will make.

read point-by-point responses

Referee: [Bidirectional Geometric Consistency] Bidirectional Geometric Consistency section: The central claim that the forward-backward cycle check 'mathematically identifies and masks occluded regions' to prevent incorrect warping objectives rests on the assumption that any cycle inconsistency is caused exclusively by occlusion. The manuscript provides no analysis or experiments showing that estimation errors in the Eulerian motion fields (e.g., aperture problems, lighting variation) do not produce false-positive masks that remove valid adjacent-frame motion signals, which would undermine the bounded-error supervision guarantee.

Authors: We appreciate the referee drawing attention to this subtlety. The forward-backward cycle is derived from the geometric identity that, for visible regions, the composition of adjacent-frame Eulerian fields should recover the original coordinates up to discretization error. The mask is therefore a conservative filter that discards any location exhibiting inconsistency, regardless of cause. While we agree that motion-estimation artifacts (aperture problems, illumination changes) can trigger false positives, such locations are precisely those where the supervision signal is unreliable; discarding them still supports the bounded-error claim for the retained regions. Nevertheless, we acknowledge that the manuscript lacks explicit robustness analysis and will add a dedicated paragraph plus a small diagnostic experiment quantifying the fraction of masks attributable to estimation error versus true occlusion. revision: yes
Referee: [Experiments] Experimental evaluation: The abstract states that 'extensive experiments demonstrate' accelerated training, preserved coherence, and reduced artifacts, yet the provided manuscript text supplies no quantitative metrics, ablation tables, or implementation details (e.g., datasets, baselines, or error measures). This absence makes it impossible to verify whether the proposed mechanisms actually deliver the claimed benefits.

Authors: We apologize for the incomplete presentation in the version the referee received. The complete manuscript contains quantitative results (FID, FVD, temporal coherence scores), ablation tables isolating the Eulerian supervision and bidirectional mask, implementation details (datasets, training schedule, baselines), and error measures. To eliminate any ambiguity, we will reorganize the experimental section so that all metrics, tables, and reproducibility information appear in the main body with clear references from the abstract and introduction. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper describes a methodological shift to adjacent-frame Eulerian motion fields for parallel training and bounded-error supervision, plus a Bidirectional Geometric Consistency mechanism based on forward-backward cycle checks for occlusion masking. No equations, fitted parameters, or self-citations are presented that reduce any claimed result (e.g., accelerated training or reduced artifacts) to quantities defined by the method's own inputs or prior author work. The central claims rest on the design choices and empirical validation rather than self-referential definitions or renamings, making the derivation self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no free parameters, axioms, or invented entities can be extracted.

pith-pipeline@v0.9.1-grok · 5684 in / 1048 out tokens · 27353 ms · 2026-06-30T23:23:26.512570+00:00 · methodology

Review history (4 revisions) →

discussion (0)

Reference graph

Works this paper leans on

42 extracted references · 14 canonical work pages · 6 internal anchors

[1]

Omer Bar-Tal, Hila Chefer, Omer Tov, Charles Herrmann, Roni Paiss, Shiran Zada, Ariel Ephrat, Junhwa Hur, Guanghui Liu, Amit Raj, et al. 2024. Lumiere: A space-time diffusion model for video generation. InSIGGRAPH Asia 2024 Conference Papers. 1–11

2024
[2]

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

work page internal anchor Pith review Pith/arXiv arXiv 2023
[3]

Ryan Burgert, Yuancheng Xu, Wenqi Xian, Oliver Pilarski, Pascal Clausen, Ming- ming He, Li Ma, Yitong Deng, Lingxiao Li, Mohsen Mousavi, et al. 2025. Go-with- the-flow: Motion-controllable video diffusion models using real-time warped noise. InProceedings of the Computer Vision and Pattern Recognition Conference. 13–23

2025
[4]

Zhiyuan Chen, Jiajiong Cao, Zhiquan Chen, Yuming Li, and Chenguang Ma. 2025. Echomimic: Lifelike audio-driven portrait animations through editable landmark conditions. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 2403–2410

2025
[5]

Jiahao Cui, Hui Li, Yun Zhan, Hanlin Shang, Kaihui Cheng, Yuqi Ma, Shan Mu, Hang Zhou, Jingdong Wang, and Siyu Zhu. 2025. Hallo3: Highly dynamic and realistic portrait image animation with video diffusion transformer. InProceedings of the Computer Vision and Pattern Recognition Conference. 21086–21095

2025
[6]

Jiankang Deng, Jia Guo, Niannan Xue, and Stefanos Zafeiriou. 2019. Arcface: Additive angular margin loss for deep face recognition. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 4690–4699

2019
[7]

Wanquan Feng, Tianhao Qi, Jiawei Liu, Mingzhen Sun, Pengqi Tu, Tianxiang Ma, Fei Dai, Songtao Zhao, Siyu Zhou, and Qian He. 2025. I2vcontrol: Disentan- gled and unified video motion synthesis control. InProceedings of the IEEE/CVF International Conference on Computer Vision. 14051–14060

2025
[8]

Craig G Fraser. 2005. Leonhard Euler, book on the calculus of variations (1744). InLandmark Writings in Western Mathematics 1640-1940. Elsevier, 168–180

2005
[9]

Junyao Gao, Yanan Sun, Fei Shen, Xin Jiang, Zhening Xing, Kai Chen, and Cairong Zhao. 2025. Faceshot: Bring any character into life.arXiv preprint arXiv:2503.00740(2025)

work page arXiv 2025
[10]

Sicheng Gao, Yutang Feng, Linlin Yang, Xuhui Liu, Zichen Zhu, David S Do- ermann, and Baochang Zhang. 2022. MagFormer: Hybrid Video Motion Mag- nification Transformer from Eulerian and Lagrangian Perspectives.. InBMVC. 444

2022
[11]

Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. 2017. Gans trained by a two time-scale update rule converge to a local nash equilibrium.Advances in neural information processing systems30 (2017)

2017
[12]

Li Hu. 2024. Animate anyone: Consistent and controllable image-to-video syn- thesis for character animation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 8153–8163

2024
[13]

Longbin Ji, Lei Zhong, Pengfei Wei, and Changjian Li. 2025. PoseTraj: Pose- Aware Trajectory Control in Video Diffusion. InProceedings of the Computer Vision and Pattern Recognition Conference. 22776–22785

2025
[14]

Hoiyeong Jin, Hyojin Jang, Jeongho Kim, Junha Hyung, Kinam Kim, Dongjin Kim, Huijin Choi, Hyeonji Kim, and Jaegul Choo. 2025. InsertAnywhere: Bridging 4D Scene Geometry and Diffusion Models for Realistic Video Object Insertion. arXiv preprint arXiv:2512.17504(2025)

work page internal anchor Pith review Pith/arXiv arXiv 2025
[15]

Wei-Sheng Lai, Jia-Bin Huang, Oliver Wang, Eli Shechtman, Ersin Yumer, and Ming-Hsuan Yang. 2018. Learning blind video temporal consistency. InProceed- ings of the European conference on computer vision (ECCV). 170–185

2018
[16]

Yaowei Li, Xintao Wang, Zhaoyang Zhang, Zhouxia Wang, Ziyang Yuan, Liang- bin Xie, Ying Shan, and Yuexian Zou. 2025. Image conductor: Precision control for interactive video synthesis. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 5031–5038

2025
[17]

Jingyun Liang, Yuchen Fan, Kai Zhang, Radu Timofte, Luc Van Gool, and Rakesh Ranjan. 2024. Movideo: Motion-aware video generation with diffusion model. In European Conference on Computer Vision. Springer, 56–74

2024
[18]

Ilya Loshchilov and Frank Hutter. 2017. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101(2017)

work page internal anchor Pith review Pith/arXiv arXiv 2017
[19]

Xin Ma, Yaohui Wang, Gengyun Jia, Xinyuan Chen, Yuan-Fang Li, Cunjian Chen, and Yu Qiao. 2024. Cinemo: Consistent and controllable image animation with motion diffusion models.arXiv preprint arXiv:2407.15642(2024)

work page arXiv 2024
[20]

Koichi Namekata, Sherwin Bahmani, Ziyi Wu, Yash Kant, Igor Gilitschenski, and David B Lindell. 2024. Sg-i2v: Self-guided trajectory control in image-to-video generation.arXiv preprint arXiv:2411.04989(2024)

work page arXiv 2024
[21]

Niranjan D Narvekar and Lina J Karam. 2011. A no-reference image blur metric based on the cumulative probability of blur detection (CPBD).IEEE Transactions on Image Processing20, 9 (2011), 2678–2683

2011
[22]

Muyao Niu, Xiaodong Cun, Xintao Wang, Yong Zhang, Ying Shan, and Yinqiang Zheng. 2024. Mofa-video: Controllable image animation via generative motion field adaptions in frozen image-to-video diffusion model. InEuropean Conference on Computer Vision. Springer, 111–128

2024
[23]

Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sand- hini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al
[24]

In International Conference on Machine Learning (ICML)

Learning transferable visual models from natural language supervision. In International Conference on Machine Learning (ICML). PMLR
[25]

Jiapeng Tang, Kai Li, Chengxiang Yin, Liuhao Ge, Fei Jiang, Jiu Xu, Matthias Nießner, Christian Häne, Timur Bagautdinov, Egor Zakharov, et al . 2025. Fac- torPortrait: Controllable Portrait Animation via Disentangled Expression, Pose, and Viewpoint.arXiv preprint arXiv:2512.11645(2025)

work page arXiv 2025
[26]

Zachary Teed and Jia Deng. 2020. Raft: Recurrent all-pairs field transforms for optical flow. InEuropean conference on computer vision. Springer, 402–419

2020
[27]

Carl Vondrick, Hamed Pirsiavash, and Antonio Torralba. 2016. Generating videos with scene dynamics.Advances in neural information processing systems 29 (2016)

2016
[28]

Team Wan, Ang Wang, Baole Ai, Bin Wen, Chaojie Mao, Chen-Wei Xie, Di Chen, Feiwu Yu, Haiming Zhao, Jianxiao Yang, et al. 2025. Wan: Open and advanced large-scale video generative models.arXiv preprint arXiv:2503.20314(2025)

work page internal anchor Pith review Pith/arXiv arXiv 2025
[29]

Angtian Wang, Haibin Huang, Jacob Zhiyuan Fang, Yiding Yang, and Chongyang Ma. 2025. ATI: Any Trajectory Instruction for Controllable Video Generation. arXiv preprint arXiv:2505.22944(2025)

work page arXiv 2025
[30]

Zhouxia Wang, Ziyang Yuan, Xintao Wang, Yaowei Li, Tianshui Chen, Menghan Xia, Ping Luo, and Ying Shan. 2024. Motionctrl: A unified and flexible motion controller for video generation. InACM SIGGRAPH 2024 Conference Papers. 1–11

2024
[31]

Thaddäus Wiedemer, Yuxuan Li, Paul Vicol, Shixiang Shane Gu, Nick Matarese, Kevin Swersky, Been Kim, Priyank Jaini, and Robert Geirhos. 2025. Video models are zero-shot learners and reasoners.arXiv preprint arXiv:2509.20328(2025)

work page internal anchor Pith review Pith/arXiv arXiv 2025
[32]

Mingwang Xu, Hui Li, Qingkun Su, Hanlin Shang, Liwei Zhang, Ce Liu, Jingdong Wang, Yao Yao, and Siyu Zhu. 2024. Hallo: Hierarchical audio-driven visual synthesis for portrait image animation.arXiv preprint arXiv:2406.08801(2024)

work page arXiv 2024
[33]

Zhongcong Xu, Jianfeng Zhang, Jun Hao Liew, Hanshu Yan, Jia-Wei Liu, Chenxu Zhang, Jiashi Feng, and Mike Zheng Shou. 2024. Magicanimate: Temporally consistent human image animation using diffusion model. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 1481–1490

2024
[34]

Fei Yin, Vikram Voleti, Nikita Drobyshev, Maksim Lapin, Aaryaman Vasishta, Varun Jampani, et al . 2025. Stable Video-Driven Portraits.arXiv preprint arXiv:2509.17476(2025)

work page arXiv 2025
[35]

Fei Yin, Yong Zhang, Xiaodong Cun, Mingdeng Cao, Yanbo Fan, Xuan Wang, Qingyan Bai, Baoyuan Wu, Jue Wang, and Yujiu Yang. 2022. Styleheat: One- shot high-resolution editable talking face generation via pre-trained stylegan. In European conference on computer vision. Springer, 85–101

2022
[36]

Shengming Yin, Chenfei Wu, Jian Liang, Jie Shi, Houqiang Li, Gong Ming, and Nan Duan. 2023. Dragnuwa: Fine-grained control in video generation by inte- grating text, image, and trajectory.arXiv preprint arXiv:2308.08089(2023)

work page internal anchor Pith review Pith/arXiv arXiv 2023
[37]

Sihyun Yu, Jihoon Tack, Sangwoo Mo, Hyunsu Kim, Junho Kim, Jung-Woo Ha, and Jinwoo Shin. 2022. Generating videos with dynamics-aware implicit generative adversarial networks.arXiv preprint arXiv:2202.10571(2022)

work page arXiv 2022
[38]

Zhongrui Yu, Martina Megaro-Boldini, Robert W Sumner, and Abdelaziz Djelouah. 2025. Unboxed: Geometrically and Temporally Consistent Video Outpainting. InProceedings of the Computer Vision and Pattern Recognition Con- ference. 7309–7319

2025
[39]

Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang
[40]

The Unreasonable Effectiveness of Deep Features as a Perceptual Metric. InCVPR
[41]

Wenxuan Zhang, Xiaodong Cun, Xuan Wang, Yong Zhang, Xi Shen, Yu Guo, Ying Shan, and Fei Wang. 2023. Sadtalker: Learning realistic 3d motion coefficients for stylized audio-driven single image talking face animation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 8652–8661

2023
[42]

vanishing subject

Guangcong Zheng, Xianpan Zhou, Xuewei Li, Zhongang Qi, Ying Shan, and Xi Li. 2023. Layoutdiffusion: Controllable diffusion model for layout-to-image generation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 22490–22499. Trovato et al. A Proof of Theorem 1 Theorem 1. (Expected Error Accumulation in Lagrangian Motion F...

2023

[1] [1]

Omer Bar-Tal, Hila Chefer, Omer Tov, Charles Herrmann, Roni Paiss, Shiran Zada, Ariel Ephrat, Junhwa Hur, Guanghui Liu, Amit Raj, et al. 2024. Lumiere: A space-time diffusion model for video generation. InSIGGRAPH Asia 2024 Conference Papers. 1–11

2024

[2] [2]

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

work page internal anchor Pith review Pith/arXiv arXiv 2023

[3] [3]

Ryan Burgert, Yuancheng Xu, Wenqi Xian, Oliver Pilarski, Pascal Clausen, Ming- ming He, Li Ma, Yitong Deng, Lingxiao Li, Mohsen Mousavi, et al. 2025. Go-with- the-flow: Motion-controllable video diffusion models using real-time warped noise. InProceedings of the Computer Vision and Pattern Recognition Conference. 13–23

2025

[4] [4]

Zhiyuan Chen, Jiajiong Cao, Zhiquan Chen, Yuming Li, and Chenguang Ma. 2025. Echomimic: Lifelike audio-driven portrait animations through editable landmark conditions. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 2403–2410

2025

[5] [5]

Jiahao Cui, Hui Li, Yun Zhan, Hanlin Shang, Kaihui Cheng, Yuqi Ma, Shan Mu, Hang Zhou, Jingdong Wang, and Siyu Zhu. 2025. Hallo3: Highly dynamic and realistic portrait image animation with video diffusion transformer. InProceedings of the Computer Vision and Pattern Recognition Conference. 21086–21095

2025

[6] [6]

Jiankang Deng, Jia Guo, Niannan Xue, and Stefanos Zafeiriou. 2019. Arcface: Additive angular margin loss for deep face recognition. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 4690–4699

2019

[7] [7]

Wanquan Feng, Tianhao Qi, Jiawei Liu, Mingzhen Sun, Pengqi Tu, Tianxiang Ma, Fei Dai, Songtao Zhao, Siyu Zhou, and Qian He. 2025. I2vcontrol: Disentan- gled and unified video motion synthesis control. InProceedings of the IEEE/CVF International Conference on Computer Vision. 14051–14060

2025

[8] [8]

Craig G Fraser. 2005. Leonhard Euler, book on the calculus of variations (1744). InLandmark Writings in Western Mathematics 1640-1940. Elsevier, 168–180

2005

[9] [9]

Junyao Gao, Yanan Sun, Fei Shen, Xin Jiang, Zhening Xing, Kai Chen, and Cairong Zhao. 2025. Faceshot: Bring any character into life.arXiv preprint arXiv:2503.00740(2025)

work page arXiv 2025

[10] [10]

Sicheng Gao, Yutang Feng, Linlin Yang, Xuhui Liu, Zichen Zhu, David S Do- ermann, and Baochang Zhang. 2022. MagFormer: Hybrid Video Motion Mag- nification Transformer from Eulerian and Lagrangian Perspectives.. InBMVC. 444

2022

[11] [11]

Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. 2017. Gans trained by a two time-scale update rule converge to a local nash equilibrium.Advances in neural information processing systems30 (2017)

2017

[12] [12]

Li Hu. 2024. Animate anyone: Consistent and controllable image-to-video syn- thesis for character animation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 8153–8163

2024

[13] [13]

Longbin Ji, Lei Zhong, Pengfei Wei, and Changjian Li. 2025. PoseTraj: Pose- Aware Trajectory Control in Video Diffusion. InProceedings of the Computer Vision and Pattern Recognition Conference. 22776–22785

2025

[14] [14]

Hoiyeong Jin, Hyojin Jang, Jeongho Kim, Junha Hyung, Kinam Kim, Dongjin Kim, Huijin Choi, Hyeonji Kim, and Jaegul Choo. 2025. InsertAnywhere: Bridging 4D Scene Geometry and Diffusion Models for Realistic Video Object Insertion. arXiv preprint arXiv:2512.17504(2025)

work page internal anchor Pith review Pith/arXiv arXiv 2025

[15] [15]

Wei-Sheng Lai, Jia-Bin Huang, Oliver Wang, Eli Shechtman, Ersin Yumer, and Ming-Hsuan Yang. 2018. Learning blind video temporal consistency. InProceed- ings of the European conference on computer vision (ECCV). 170–185

2018

[16] [16]

Yaowei Li, Xintao Wang, Zhaoyang Zhang, Zhouxia Wang, Ziyang Yuan, Liang- bin Xie, Ying Shan, and Yuexian Zou. 2025. Image conductor: Precision control for interactive video synthesis. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 5031–5038

2025

[17] [17]

Jingyun Liang, Yuchen Fan, Kai Zhang, Radu Timofte, Luc Van Gool, and Rakesh Ranjan. 2024. Movideo: Motion-aware video generation with diffusion model. In European Conference on Computer Vision. Springer, 56–74

2024

[18] [18]

Ilya Loshchilov and Frank Hutter. 2017. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101(2017)

work page internal anchor Pith review Pith/arXiv arXiv 2017

[19] [19]

Xin Ma, Yaohui Wang, Gengyun Jia, Xinyuan Chen, Yuan-Fang Li, Cunjian Chen, and Yu Qiao. 2024. Cinemo: Consistent and controllable image animation with motion diffusion models.arXiv preprint arXiv:2407.15642(2024)

work page arXiv 2024

[20] [20]

Koichi Namekata, Sherwin Bahmani, Ziyi Wu, Yash Kant, Igor Gilitschenski, and David B Lindell. 2024. Sg-i2v: Self-guided trajectory control in image-to-video generation.arXiv preprint arXiv:2411.04989(2024)

work page arXiv 2024

[21] [21]

Niranjan D Narvekar and Lina J Karam. 2011. A no-reference image blur metric based on the cumulative probability of blur detection (CPBD).IEEE Transactions on Image Processing20, 9 (2011), 2678–2683

2011

[22] [22]

Muyao Niu, Xiaodong Cun, Xintao Wang, Yong Zhang, Ying Shan, and Yinqiang Zheng. 2024. Mofa-video: Controllable image animation via generative motion field adaptions in frozen image-to-video diffusion model. InEuropean Conference on Computer Vision. Springer, 111–128

2024

[23] [23]

Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sand- hini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al

[24] [24]

In International Conference on Machine Learning (ICML)

Learning transferable visual models from natural language supervision. In International Conference on Machine Learning (ICML). PMLR

[25] [25]

Jiapeng Tang, Kai Li, Chengxiang Yin, Liuhao Ge, Fei Jiang, Jiu Xu, Matthias Nießner, Christian Häne, Timur Bagautdinov, Egor Zakharov, et al . 2025. Fac- torPortrait: Controllable Portrait Animation via Disentangled Expression, Pose, and Viewpoint.arXiv preprint arXiv:2512.11645(2025)

work page arXiv 2025

[26] [26]

Zachary Teed and Jia Deng. 2020. Raft: Recurrent all-pairs field transforms for optical flow. InEuropean conference on computer vision. Springer, 402–419

2020

[27] [27]

Carl Vondrick, Hamed Pirsiavash, and Antonio Torralba. 2016. Generating videos with scene dynamics.Advances in neural information processing systems 29 (2016)

2016

[28] [28]

Team Wan, Ang Wang, Baole Ai, Bin Wen, Chaojie Mao, Chen-Wei Xie, Di Chen, Feiwu Yu, Haiming Zhao, Jianxiao Yang, et al. 2025. Wan: Open and advanced large-scale video generative models.arXiv preprint arXiv:2503.20314(2025)

work page internal anchor Pith review Pith/arXiv arXiv 2025

[29] [29]

Angtian Wang, Haibin Huang, Jacob Zhiyuan Fang, Yiding Yang, and Chongyang Ma. 2025. ATI: Any Trajectory Instruction for Controllable Video Generation. arXiv preprint arXiv:2505.22944(2025)

work page arXiv 2025

[30] [30]

Zhouxia Wang, Ziyang Yuan, Xintao Wang, Yaowei Li, Tianshui Chen, Menghan Xia, Ping Luo, and Ying Shan. 2024. Motionctrl: A unified and flexible motion controller for video generation. InACM SIGGRAPH 2024 Conference Papers. 1–11

2024

[31] [31]

Thaddäus Wiedemer, Yuxuan Li, Paul Vicol, Shixiang Shane Gu, Nick Matarese, Kevin Swersky, Been Kim, Priyank Jaini, and Robert Geirhos. 2025. Video models are zero-shot learners and reasoners.arXiv preprint arXiv:2509.20328(2025)

work page internal anchor Pith review Pith/arXiv arXiv 2025

[32] [32]

Mingwang Xu, Hui Li, Qingkun Su, Hanlin Shang, Liwei Zhang, Ce Liu, Jingdong Wang, Yao Yao, and Siyu Zhu. 2024. Hallo: Hierarchical audio-driven visual synthesis for portrait image animation.arXiv preprint arXiv:2406.08801(2024)

work page arXiv 2024

[33] [33]

Zhongcong Xu, Jianfeng Zhang, Jun Hao Liew, Hanshu Yan, Jia-Wei Liu, Chenxu Zhang, Jiashi Feng, and Mike Zheng Shou. 2024. Magicanimate: Temporally consistent human image animation using diffusion model. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 1481–1490

2024

[34] [34]

Fei Yin, Vikram Voleti, Nikita Drobyshev, Maksim Lapin, Aaryaman Vasishta, Varun Jampani, et al . 2025. Stable Video-Driven Portraits.arXiv preprint arXiv:2509.17476(2025)

work page arXiv 2025

[35] [35]

Fei Yin, Yong Zhang, Xiaodong Cun, Mingdeng Cao, Yanbo Fan, Xuan Wang, Qingyan Bai, Baoyuan Wu, Jue Wang, and Yujiu Yang. 2022. Styleheat: One- shot high-resolution editable talking face generation via pre-trained stylegan. In European conference on computer vision. Springer, 85–101

2022

[36] [36]

Shengming Yin, Chenfei Wu, Jian Liang, Jie Shi, Houqiang Li, Gong Ming, and Nan Duan. 2023. Dragnuwa: Fine-grained control in video generation by inte- grating text, image, and trajectory.arXiv preprint arXiv:2308.08089(2023)

work page internal anchor Pith review Pith/arXiv arXiv 2023

[37] [37]

Sihyun Yu, Jihoon Tack, Sangwoo Mo, Hyunsu Kim, Junho Kim, Jung-Woo Ha, and Jinwoo Shin. 2022. Generating videos with dynamics-aware implicit generative adversarial networks.arXiv preprint arXiv:2202.10571(2022)

work page arXiv 2022

[38] [38]

Zhongrui Yu, Martina Megaro-Boldini, Robert W Sumner, and Abdelaziz Djelouah. 2025. Unboxed: Geometrically and Temporally Consistent Video Outpainting. InProceedings of the Computer Vision and Pattern Recognition Con- ference. 7309–7319

2025

[39] [39]

Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang

[40] [40]

The Unreasonable Effectiveness of Deep Features as a Perceptual Metric. InCVPR

[41] [41]

Wenxuan Zhang, Xiaodong Cun, Xuan Wang, Yong Zhang, Xi Shen, Yu Guo, Ying Shan, and Fei Wang. 2023. Sadtalker: Learning realistic 3d motion coefficients for stylized audio-driven single image talking face animation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 8652–8661

2023

[42] [42]

vanishing subject

Guangcong Zheng, Xianpan Zhou, Xuewei Li, Zhongang Qi, Ying Shan, and Xi Li. 2023. Layoutdiffusion: Controllable diffusion model for layout-to-image generation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 22490–22499. Trovato et al. A Proof of Theorem 1 Theorem 1. (Expected Error Accumulation in Lagrangian Motion F...

2023