M3DDM+: An improved video outpainting by a modified masking strategy
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917206268313600 |
|---|---|
| author | Murakawa, Takuya Fukuzawa, Takumi Ding, Ning Tamaki, Toru |
| author_facet | Murakawa, Takuya Fukuzawa, Takumi Ding, Ning Tamaki, Toru |
| contents | M3DDM provides a computationally efficient framework for video outpainting via latent diffusion modeling. However, it exhibits significant quality degradation -- manifested as spatial blur and temporal inconsistency -- under challenging scenarios characterized by limited camera motion or large outpainting regions, where inter-frame information is limited. We identify the cause as a training-inference mismatch in the masking strategy: M3DDM's training applies random mask directions and widths across frames, whereas inference requires consistent directional outpainting throughout the video. To address this, we propose M3DDM+, which applies uniform mask direction and width across all frames during training, followed by fine-tuning of the pretrained M3DDM model. Experiments demonstrate that M3DDM+ substantially improves visual fidelity and temporal coherence in information-limited scenarios while maintaining computational efficiency. The code is available at https://github.com/tamaki-lab/M3DDM-Plus. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_11048 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | M3DDM+: An improved video outpainting by a modified masking strategy Murakawa, Takuya Fukuzawa, Takumi Ding, Ning Tamaki, Toru Computer Vision and Pattern Recognition M3DDM provides a computationally efficient framework for video outpainting via latent diffusion modeling. However, it exhibits significant quality degradation -- manifested as spatial blur and temporal inconsistency -- under challenging scenarios characterized by limited camera motion or large outpainting regions, where inter-frame information is limited. We identify the cause as a training-inference mismatch in the masking strategy: M3DDM's training applies random mask directions and widths across frames, whereas inference requires consistent directional outpainting throughout the video. To address this, we propose M3DDM+, which applies uniform mask direction and width across all frames during training, followed by fine-tuning of the pretrained M3DDM model. Experiments demonstrate that M3DDM+ substantially improves visual fidelity and temporal coherence in information-limited scenarios while maintaining computational efficiency. The code is available at https://github.com/tamaki-lab/M3DDM-Plus. |
| title | M3DDM+: An improved video outpainting by a modified masking strategy |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2601.11048 |