LTX Video 13B In ComfyUI - Still The Most Fastest AI VIdeo & Getting Better!
Added 2025-05-10 18:04:34 +0000 UTC
Video : https://youtu.be/2asTwiDn0y0
In this video, we explore the latest update of the LTX Video AI model, version 0.9.7 with 13 billion parameters. Discover how to run this powerful model locally using FP8 or GGUF-quantized versions, optimized for consumer-grade GPUs like the Nvidia 4090 and 5090. Learn about the new custom nodes, Spatiotemporal Skip Guidance (STG), and the improved prompt enhancer powered by Llama 3.2.3B. We also cover workflow optimizations, upscaling techniques, and performance comparisons to help you generate high-quality videos efficiently.
Who is this for?
AI enthusiasts and researchers
Video creators exploring generative AI tools
Developers working with ComfyUI and Hugging Face models
Anyone interested in running large AI models locally
Why it matters:
The LTX Video 13B model offers advanced text-to-video and image-to-video capabilities, with significant improvements in coherence and detail. Understanding how to leverage its features can enhance creative workflows and reduce reliance on cloud-based solutions.
Resources:
ltxv-13b-0.9.7-dev : https://huggingface.co/Lightricks/LTX-Video/blob/main/ltxv-13b-0.9.7-dev.safetensors
FP8 : https://huggingface.co/Lightricks/LTX-Video/blob/main/ltxv-13b-0.9.7-dev-fp8.safetensors
Requires installing for using FP8: https://github.com/Lightricks/LTX-Video-Q8-Kernels
ComfyUI-LTXVideo https://github.com/Lightricks/ComfyUI-LTXVideo
LTXVideo Upscalers:
ltxv-spatial-upscaler-0.9.7
https://huggingface.co/Lightricks/LTX-Video/blob/main/ltxv-spatial-upscaler-0.9.7.safetensors
ltxv-temporal-upscaler-0.9.7
https://huggingface.co/Lightricks/LTX-Video/blob/main/ltxv-temporal-upscaler-0.9.7.safetensors
More Eaier way is using GGUF
GGUF: https://huggingface.co/wsbagnsv1/ltxv-13b-0.9.7-dev-GGUF
Workflow:
https://huggingface.co/wsbagnsv1/ltxv-13b-0.9.7-dev-GGUF/blob/main/exampleworkflow.json
