Views
No views yet
1CameraNoise-I2V/
2 1024x576/
3 cameranoise_i2v_wan2.1_1024x576_lora.safetensors
4 i2v_demo_results/
5 demo1
6 demo2
7 demo3
8 ...
9 demo101git clone https://github.com/gulucaptain/CameraNoise
2cd CameraNoise
3pip install -r requirements.txt1VGGT checkpoint
2QwenVL checkpoint
3Wan2.1-I2V-14B-720P checkpoint
4CameraNoise-I2V LoRA checkpointoutputs/. Put the reference image and reference video in the inputs/ folder:1outputs/demo1/
2 inputs/
3 example.jpg # reference image
4 example.mp4 # reference video for camera motion1python cameranoise_i2v.py \
2 --demo-dir outputs/demo1 \
3 --vggt-ckpt /path/to/VGGT-1B \
4 --cameranoise-config cameranoise_warping/configs/default.yaml \
5 --qwenvl-model-path /path/to/Qwen2-VL-7B-Instruct \
6 --model-root /path/to/Wan2.1-I2V-14B-720P \
7 --lora-path /path/to/CameraNoise-I2V/1024x576/cameranoise_lora.safetensors \
8 --height 576 \
9 --width 1024 \
10 --frames 49 \
11 --cfg 3.5 \
12 --device cuda \
13 --output-type singlecameranoise_downscale_size = [height // 8, width // 8]1576 x 1024 -> [72, 128]
2768 x 768 -> [96, 96]--cameranoise-downscale-size 72,1281outputs/demo1/
2 conditions/
3 caption.txt
4 noises/
5 *_noises.npy
6 *_visualization.mp4
7 samples/
8 demo1.mp4
9 demo1_compare.mp4
10 manifest.json1@inproceedings{zhao2026cameranoise,
2 title = {CameraNoise: Enabling Faithful Camera Control in Video Diffusion through Geometry-Flow-Guided Noise Warping},
3 author = {Zhao, Haoyu and Gu, Jiaxi and Chen, Haoran and Zheng, Qingping and Jin, Yeying and Yang, Hongyi and Cheng, Junqi and Zhang, Yuang and Lu, Zenghui and Yu, Huan and Jiang, Jie and Shu, Peng and Wu, Zuxuan and Jiang, Yu-Gang},
4 booktitle = {Proceedings of the Forty-third International Conference on Machine Learning},
5 year = {2026}
6}
## Disclaimer
This model is released for research purposes. Please refer to the GitHub repository for the complete codebase, detailed installation instructions, and inference scripts.