Can Pose Transfer Models Generate Realistic Human Motion?
Not on the map. Why?
Abstract (arXiv)
Recent pose-transfer methods aim to generate temporally consistent and fully controllable videos of human action where the motion from a reference video is reenacted by a new identity. We evaluate three state-of-the-art pose-transfer methods -- AnimateAnyone, MagicAnimate, and ExAvatar -- by generating videos with actions and identities outside the training distribution and conducting a participant study about the quality of these videos. In a controlled environment of 20 distinct human actions, we find that participants, presented with the pose-transferred videos, correctly identify the desired action only 42.92% of the time. Moreover, the participants find the actions in the generated videos consistent with the reference (source) videos only 36.46% of the time. These results vary by method: participants find the splatting-based ExAvatar more consistent and photorealistic than the diffusion-based AnimateAnyone and MagicAnimate.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 30 of its 52 references that are in the map. First posted on arXiv 26 Jan 2025.
Builds on
30 of 52 references.
References · 30
- 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
- 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesiscited by 4,356
- 2023Animate Anyone: Consistent and Controllable Image-to-Video Synthesis for Character Animationcited by 688
- 2018DensePose: Dense Human Pose Estimation In The Wildcited by 575
- 2023MagicAnimate: Temporally Consistent Human Image Animation using Diffusion Modelcited by 330
- 2021Motion Representations for Articulated Animationcited by 255
- 2017Pose Guided Person Image Generationcited by 251
- 2020Deep Learning-Based Human Pose Estimation: A Surveycited by 207
- 2023DreamPose: Fashion Image-to-Video Synthesis via Stable Diffusioncited by 167
- 2017Deformable GANs for Pose-based Human Image Generationcited by 166
- 2019Liquid Warping GAN: A Unified Framework for Human Motion Imitation, Appearance Transfer and Novel View Synthesiscited by 161
- 2019Progressive Pose Attention Transfer for Person Image Generationcited by 138
18 more
- 2019DwNet: Dense warp-based network for pose-guided human video generationcited by 115
- 2023Human Motion Generation: A Surveycited by 96
- 2024Expressive Whole-Body 3D Gaussian Avatarcited by 89
- 2023X-Avatar: Expressive Human Avatarscited by 66
- 2020High-Fidelity Neural Human Motion Transfer from Monocular Videocited by 31
- 2023Bidirectionally Deformable Motion Modulation For Video-based Human Pose Transfercited by 25
- 2023Image-Based Virtual Try-On: A Surveycited by 19
- 2020Human Motion Transfer from Poses in the Wildcited by 17
- 2023Edit Temporal-Consistent Videos with Image Diffusion Modelcited by 12
- 2023Cinematic Behavior Transfer via NeRF-based Differentiable Filmingcited by 12
- 2021Transfer Learning for Pose Estimation of Illustrated Characterscited by 10
- 2022Multi-scale Attention Guided Pose Transfercited by 10
- 2023PoseExaminer: Automated Testing of Out-of-Distribution Robustness in Human Pose and Shape Estimationcited by 10
- 2022VL4Pose: Active Learning Through Out-Of-Distribution Detection For Pose Estimationcited by 9
- 2022Collecting The Puzzle Pieces: Disentangled Self-Driven Human Pose Transfer by Permuting Texturescited by 6
- 2023Open-World Pose Transfer via Sequential Test-Time Adaptioncited by 4
- 2024Human Pose Descriptions and Subject-Focused Attention for Improved Zero-Shot Transfer in Human-Centric Classification Taskscited by 3
- 2021Human Pose Transfer with Augmented Disentangled Feature Consistencycited by 1
Built on by
Citing papers per year: 2025: 1.
1 paper.