#192 · Primary category: Video & Animation
ReCamMaster
[ICCV'25 Best Paper Finalist] ReCamMaster: Camera-Controlled Generative Rendering from A Single Video
Project last updated:11/28/25
GitHub Stars
1.9K
Forks
98
Contributors
5
License
MIT
Why we included this project
ReCamMaster tackles a deceptively common post-production problem: you have footage shot from one angle and you want the same scene from another, without a reshoot. Feed it a single clip plus a target camera trajectory and it synthesizes a plausible re-rendered version of the scene, which makes it interesting for filmmakers, virtual-production crews, and anyone fixing framing in the edit. The repo includes training and inference code and a checkpoint, and it comes out of a peer-reviewed ICCV paper, so it works as both a practical reference and a study base for camera-controlled video synthesis. One honest caveat from the authors: the open-source build may not match the paper's best model, so treat it as a research-grade baseline. You will want a capable GPU and some comfort running generative video models.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
yt-dlp
A feature-rich command-line audio/video downloader
MoneyPrinterTurbo
Generate HD short videos from a topic or keyword with an automated AI workflow.
Deep-Live-Cam
real time face swap and one-click video deepfake with only a single image
manim
Animation engine for explanatory math videos
anime
JavaScript animation engine