Instructions to use circlestone-labs/Anima with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusion Single File
How to use circlestone-labs/Anima with Diffusion Single File:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Anima's INT8 Acceleration Plugin and Advanced Timestep Parameters
The INT8 acceleration plugin can significantly increase generation speed on GPUs with strong INT8 computing power, such as the RTX 30 series, while keeping the quality almost unchanged. When generating with an RTX 3090 GPU, the generation time is reduced to 0.5-0.6x compared to when it is disabled.
Disabled
Enabled
Original plugin: https://github.com/SparknightLLC/ComfyUI-INT8-Toolkit
forge-neo: https://github.com/kongbai-0421/anima-int8-toolkit
The advanced timestep plugin allows for the configuration of advanced timestep sampling parameters.
https://github.com/kongbai-0421/anima-timestep-advanced
Looking forward to better FP8 support though. I get nowhere near 2x speed when using FP8 quants :(
INT8 only improved performance by 10% (1.66s/it -> 1.49s/it) on my B580, but reduced VRAM usage by 4G.
The overall image quality hasn’t dropped significantly, but there’s a marked difference in the finer details.

