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
Bernini Workflow
Bernini's facial consistency is excellent; could you please take a look at her?
been using it a bit, its a super model. But Wan, not LTX
Maybe I'll make a few Wan workflows too.
Wan has a huge user base (LTX too by now, but not in the start), so there are so many workflows around (And my reason for making the LTX wf's was the lack of workflows around)
The Bernini model I bet will even have a default workflow inside comfy too real soon (support was just added couple of days ago)
But since the model is so good, and interesting, i might make some variations ;-) its a very good model for many things
Hey Rune, do you know if this model works with SVI 2 Pro? If not, how does it fare when using the last image of the video for the starting image of a new video in order to extend it? Does the color become all saturated and bright like Wan? SVI is the only thing that I am aware of for wan where you can make long videos without the image getting all messed up. Appreciate any info you may have on this. I just don't want to waste my time downloading this model if its no good when it comes to making long videos. Thanks.
Not tried with SVI, but maybe.. will try.
But it does work with Context Window, so you can create a bit longer videos than the usual Wan. But eventually it does become saturated and all . .. its Wan.
There is a new Bernini coming (the FULL model), and that looks like its even better...
(its released already, but not yet comfy ready)
Not tried with SVI, but maybe.. will try.
But it does work with Context Window, so you can create a bit longer videos than the usual Wan. But eventually it does become saturated and all . .. its Wan.
There is a new Bernini coming (the FULL model), and that looks like its even better...
(its released already, but not yet comfy ready)
https://huggingface.co/Rudra-ai/Bernini-ComfyUI-weights/tree/main
Thanks for the info. I was always told it had to do with the VAE Decoder but LTX doesn't have this problem and it uses the same decoder as WAN. Not sure how SVI is able to get around it.
