Instructions to use Lightricks/LTX-2.3-nvfp4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use Lightricks/LTX-2.3-nvfp4 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline from diffusers.utils import load_image, export_to_video # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("Lightricks/LTX-2.3-nvfp4", dtype=torch.bfloat16, device_map="cuda") pipe.to("cuda") prompt = "A man with short gray hair plays a red electric guitar." image = load_image( "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/guitar-man.png" ) output = pipe(image=image, prompt=prompt).frames[0] export_to_video(output, "output.mp4") - Notebooks
- Google Colab
- Kaggle
distill-nvfp4
We are really waiting for distill nvfp4. Pls! Dev is not needed — low quality on dev is nonsense. But the speed in distill for a simple video makes sense!
We are really waiting for distill nvfp4. Pls! Dev is not needed — low quality on dev is nonsense. But the speed in distill for a simple video makes sense!
I converted the transformer only of the distilled model in nvfp4 if you want (note: you'll need to load the video vae, audio vae and projector in the dual clip loader with Gemma): https://huggingface.co/Winnougan/LTX-2.3-INT8/blob/main/ltx-2.3-22b-distilled_transformer_only_NVFP4.safetensors
There are different formats for LTX-2.3 as well - https://huggingface.co/Hippotes/LTX-2.3-various-formats
lol i have one aswell.
https://huggingface.co/MrReclusive/LTX-2.3-FP4/tree/main/FP4ME
requires vae's and projection as well.
Is there a workflow "image-in-video" process for NVFP4 with a looped video? For the Comfyui desktop version...