Instructions to use Kijai/LTX2.3_comfy with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusion Single File
How to use Kijai/LTX2.3_comfy 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
Will there be a v2 version of mxfp8 format
#38
by lanranjun - opened
This morning I opened Comfyui, but no other programs were opened except for a Google Chrome browser. The workflow of MXFP8 in 8 steps is as follows:
I2V, 10 second duration:
- 1280 * 720: 138s
- 1536 * 832: 175s
The previous two days of testing was probably due to opening other programs or not cleaning up the memory and video memory: - 1280 * 720: 180s
- 1536 * 832: 250s
The quality of video image generation is quite good, but the real scene is a bit cold, and compared to it, there is no dev-fp8-transformer-only color that looks good. Pixar 3D anime has similar effects.
my pc:
5060 ti 16G VRAM
64G RAM
When Comfyui is running, the computer's memory usage is about 50GB, and the video memory usage is about 14-15GB