This is not my model. Don't ask me questions. I don't know the answers.
Go to the official LTX-2.3 HuggingFace repo for answers.
I've only uploaded the model here as a way to tag the videos I generate with it. It will most likely be taken down once the official owners or CivitAI upload their own version, so be warned.
The following is the original README as of time of publishing.
LTX-2.3 is a DiT-based audio-video foundation model designed to generate synchronized video and audio within a single model. It brings together the core building blocks of modern video generation, with open weights and a focus on practical, local execution.
Model Checkpoints
NameNotesltx-2.3-22b-devThe full model, flexible and trainable in bf16ltx-2.3-22b-distilledThe distilled version of the full model, 8 steps, CFG=1ltx-2.3-22b-distilled-lora-384A LoRA version of the distilled model applicable to the full modelltx-2.3-spatial-upscaler-x2-1.0An x2 spatial upscaler for the ltx-2.3 latents, used in multi stage (multiscale) pipelines for higher resolutionltx-2.3-spatial-upscaler-x1.5-1.0An x1.5 spatial upscaler for the ltx-2.3 latents, used in multi stage (multiscale) pipelines for higher resolutionltx-2.3-temporal-upscaler-x2-1.0An x2 temporal upscaler for the ltx-2.3 latents, used in multi stage (multiscale) pipelines for higher FPS
Model Details
Developed by: Lightricks
Model type: Diffusion-based audio-video foundation model
Language(s): English
Online demo
LTX-2.3 is accessible right away via the API Playground.
Run locally
Direct use license
You can use the models - full, distilled, upscalers and any derivatives of the models - for purposes under the license.
Description
LTX2.3 FP8 Distilled from https://huggingface.co/Lightricks/LTX-2.3-fp8/blob/main/ltx-2.3-22b-distilled-fp8.safetensors
FAQ
Comments (14)
I love these "interview on the street" shorts!
Any good workflow to reommend for easy lora insertion? I disllike the "modern" once where they put everthing onto a separate subgraph.
Maybe my VideoFlow workflow fits your needs: https://civitai.com/models/1815300?modelVersionId=2775952
Loras are easy to add. Subgraphs are used, but there is no need to open them.
@ai839 wow. trying it out now, seems really good with the distilled model. Lots of room for tinkering :) Thank you mate.
@snobbias124 Have fun ♥️🙂! I linked the 1.0 version but meanwhile version 2.0 is available and much better.
where is the workflow for this ??
If you are using ComfyUI you find workflows on the "Browse Templates" page. You can also try an optimized workflow like my VideoFlow: https://civitai.com/models/1815300/videoflow-ltx-23-wan-2221-i2v-image-to-video-img2vid-workflow
there are some from official LTX github
https://github.com/Lightricks/ComfyUI-LTXVideo/tree/master/example_workflows
or after downloading the comfyui nodes
/ComfyUI/custom_nodes/ComfyUI-LTXVideo/example_workflows
FULL impossible to download, this buggy website keeps stopping the download at the end... ((
tip: Get a download manager.
Same here, honestly, I also use performance while downloading here, quite spuckie
Good tutorial for LTX video is here: https://www.youtube.com/watch?v=G2Ec3h5CfA8
The physics-based impact of the punch and kick in version 2.3 is really good; looking forward to the next version.