TeaCache can speedup ConsisID 2.1x without much visual quality degradation, in a training-free manner. The following video shows the results generated by TeaCache-ConsisID with various rel_l1_thresh values: 0 (original), 0.1 (1.6x speedup), 0.15 (2.1x speedup), and 0.2 (2.7x speedup).
consisid_teacache_example_reencoded.mp4
| ConsisID | TeaCache (0.1) | TeaCache (0.15) | TeaCache (0.20) |
|---|---|---|---|
| ~110 s | ~70 s | ~53 s | ~41 s |
Follow ConsisID to clone the repo and finish the installation, then you can modify the rel_l1_thresh to obtain your desired trade-off between latency and visul quality, and change the ckpts_path, prompt, image to customize your identity-preserving video.
For single-gpu inference, you can use the following command:
cd TeaCache4ConsisID
python3 teacache_sample_video.py \
--rel_l1_thresh 0.1 \
--ckpts_path BestWishYsh/ConsisID-preview \
--image "https://github.com/PKU-YuanGroup/ConsisID/blob/main/asserts/example_images/2.png?raw=true" \
--prompt "The video captures a boy walking along a city street, filmed in black and white on a classic 35mm camera. His expression is thoughtful, his brow slightly furrowed as if he's lost in contemplation. The film grain adds a textured, timeless quality to the image, evoking a sense of nostalgia. Around him, the cityscape is filled with vintage buildings, cobblestone sidewalks, and softly blurred figures passing by, their outlines faint and indistinct. Streetlights cast a gentle glow, while shadows play across the boy\'s path, adding depth to the scene. The lighting highlights the boy\'s subtle smile, hinting at a fleeting moment of curiosity. The overall cinematic atmosphere, complete with classic film still aesthetics and dramatic contrasts, gives the scene an evocative and introspective feel." \
--seed 42 \
--num_infer_steps 50 \
--output_path ./teacache_resultsTo generate a video with 8 GPUs, you can following here.
Learn more about ConsisID with the following resources.
- A video demonstrating ConsisID's main features.
- The research paper, Identity-Preserving Text-to-Video Generation by Frequency Decomposition for more details.
If you find TeaCache is useful in your research or applications, please consider giving us a star 🌟 and citing it by the following BibTeX entry.
@article{liu2024timestep,
title={Timestep Embedding Tells: It's Time to Cache for Video Diffusion Model},
author={Liu, Feng and Zhang, Shiwei and Wang, Xiaofeng and Wei, Yujie and Qiu, Haonan and Zhao, Yuzhong and Zhang, Yingya and Ye, Qixiang and Wan, Fang},
journal={arXiv preprint arXiv:2411.19108},
year={2024}
}
We would like to thank the contributors to the ConsisID.