Instructions to use RuneXX/LTX-2.3-Workflows with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- LTX.io
How to use RuneXX/LTX-2.3-Workflows with LTX.io:
# Install the LTX-2 pipelines git clone https://github.com/Lightricks/LTX-2.git cd LTX-2 uv sync --frozen
# Download the weights from this repo, plus the Gemma text encoder hf download RuneXX/LTX-2.3-Workflows --local-dir models/LTX-2.3-Workflows hf download google/gemma-3-12b-it-qat-q4_0-unquantized --local-dir models/gemma-3-12b
# Fast pipeline (distilled model, no distilled LoRA needed) uv run python -m ltx_pipelines.distilled \ --distilled-checkpoint-path models/LTX-2.3-Workflows/<distilled-checkpoint>.safetensors \ --spatial-upsampler-path models/LTX-2.3-Workflows/<spatial-upsampler>.safetensors \ --gemma-root models/gemma-3-12b \ --prompt "A beautiful sunset over the ocean" \ --output-path output.mp4 # For image-to-video, add: --image path/to/image.jpg 0 0.8# HQ pipeline (two-stage, higher quality) uv run python -m ltx_pipelines.ti2vid_two_stages_hq \ --checkpoint-path models/LTX-2.3-Workflows/<checkpoint>.safetensors \ --distilled-lora models/LTX-2.3-Workflows/<distilled-lora>.safetensors 0.8 \ --spatial-upsampler-path models/LTX-2.3-Workflows/<spatial-upsampler>.safetensors \ --gemma-root models/gemma-3-12b \ --prompt "A beautiful sunset over the ocean" \ --output-path output.mp4 # For image-to-video, add: --image path/to/image.jpg 0 0.8 - Notebooks
- Google Colab
- Kaggle
LTX-2.3-All In One.json - low res output?
HI, I have just downloaded your new ALL IN ONE workflow - great idea! However compared to the other workflows that create high resolution output in my ComfyUI setup, this new workflow seems to create low resolution output videos. When I choose a resolution of 720 (width) x 1280 (height) it uses 720 as the height and it outputs a video with a resolution of 384 (width) x 704 (height). So the resolution is wrong, but also I note a lot of distortions in the video. Is the CFG factor too high? With the distilled model I think it needs to be 1.0
My other question is - how can I change the number of steps? I notice that 10 steps give a bit better quality than 8, so I normally use 10 steps
have just downloaded your new ALL IN ONE workflow
Not sure what workflow you mean. ALL IN ONE workflow? At least i dont have any by that name ;)
But maybe you mean a different one or you got that from someone else perhaps? Think one of the youtubers have this concept "all in one"
My other question is - how can I change the number of steps? I notice that 10 steps give a bit better quality than 8, so I normally use 10 steps
The earlier workflows its exactly how LTX themselves do it, with "hardcoded" sigma values to 8 steps.
But I also like being more flexible, so lately I started to do basic scheduler instead, and works great..
It might already be there in the workflow as a collapsed node, for those that want the option. Just connect the BasicScheduler to the sampler.
If not, you can just add ** BasicScheduler** node, and connect that to the sigma input at the sampler.
(I'll update my older workflows to have this option as well)
Or if you want it a bit more "advanced", something like this:
Ah, then it must someone else's workflow, they posted a link to your repo however. But thanks for getting back on this.
