FreeNoise: Tuning-Free Longer Video Diffusion via Noise Rescheduling
Authors: Haonan Qiu, Menghan Xia, Yong Zhang, Yingqing He, Xintao Wang, Ying Shan, Ziwei Liu
ICLR 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments validate the superiority of our paradigm in extending the generative capabilities of video diffusion models. |
| Researcher Affiliation | Collaboration | 1Nanyang Technological University 2Tencent AI Lab 3Hong Kong University of Science and Technology |
| Pseudocode | No | The paper describes the methods using text, equations, and diagrams (e.g., Figure 3), but it does not contain a dedicated 'Pseudocode' or 'Algorithm' section or clearly formatted algorithm blocks. |
| Open Source Code | Yes | Video Crafter (Chen et al., 2023): https://github.com/AILab-CVC/Free Noise. Animate Diff (Guo et al., 2023): https://github.com/arthur-qiu/Free Noise-Animate Diff. La Vie (Wang et al., 2023d): https://github.com/arthur-qiu/Free Noise-La Vie. |
| Open Datasets | Yes | utilizing 512 prompts (from a standard evaluation paper Eval Crafter (Liu et al., 2023)) |
| Dataset Splits | No | The paper references model training and inference lengths ('trained on 16 frames', 'sample 64 frames'), and discusses evaluation metrics, but it does not specify explicit dataset splits (e.g., percentages or counts) for training, validation, or testing. |
| Hardware Specification | Yes | In addition, we also compare the operation time of those methods on NVIDIA A100. |
| Software Dependencies | No | The paper mentions implementing the method on existing models (Video Crafter, Animate Diff, La Vie) but does not provide specific version numbers for any software dependencies like programming languages or libraries (e.g., Python, PyTorch, TensorFlow, CUDA). |
| Experiment Setup | Yes | During sampling, we perform DDIM sampling (Song et al., 2020) with 50 denoising steps, setting DDIM eta to 0. The inference resolution is fixed at 256 256 pixels. The scale of the classifier-free guidance is set to 15. |