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
Workflow: T2V R2V Character Sheet with Best Face ID Lora
T2V R2V Character Sheet with Best Face ID Lora
Create consistent character videos from a character sheet and use T2V prompts to set the scene, action and motion with the new loras from @Alissonerdx
Strongly suggested to read the guide on Best Face ID lora repro for tips and tricks, and requirements to image inputs
The lora is a bit new, so might be some changes to workflow. But feel free to experiment with the lora strength, and use with or without the extra face id only lora for 2nd pass.
Get the lora here: https://huggingface.co/Alissonerdx/LTX-Best-Face-ID (face id character sheet + face id lora)
Get the BFS node here: https://github.com/alisson-anjos/ComfyUI-BFSNodes
Helper Workflow to create character sheet: https://huggingface.co/RuneXX/LTX-2.3-Workflows/tree/main/Multi-ref-character-sheet/Helper-wf
(or you can use chatGPT, Nano Banana etc with the prompt instructions here: https://huggingface.co/Alissonerdx/LTX-Best-Face-ID/blob/main/character_sheet_prompt.md )
Try out workflow here:
https://huggingface.co/RuneXX/LTX-2.3-Workflows/tree/main/Multi-ref-character-sheet
(a face id only workflow coming asap.. you can also try out workflows here https://huggingface.co/Alissonerdx/LTX-Best-Face-ID/tree/main/workflows )
How do I combine with background and another character?
How do I combine with background and another character?
out of the box this is a single character lora.
But was intending to try see if i could "hack" that, since the lora is more capable that it might first seem (you can even use regular i2v with the lora etc). Will try some experiments
For now, if you want to combine background and another character, perhaps look into LTX-Ingredients lora instead or the MSR-2 (Multi-Subject Reference lora).
Might even be possible to combine but i havent tried that
What about the sampler? Try use the euler_ancestral_cfg in first pass, and the manual sigmas.
no that seems fine. 2nd pass works good with euler_cfg_pp
In the FIRST pass, check if its euler_ancestral_cfg_pp there. And manual sigma.
(i'll download the wf too. Sometimes i leave it at Basic Scheduler, so its ready for more steps, but since you are on distilled model, manual sigma and euler_ancestral_cfg_pp is more proven to usually be good)
and all seems fine from the screens. Will run some test here also. LTX is a bit soft (And euler too), but perhaps not as much as your output
the first pass in default no changes anything
yes thats default LTX. So thats all ok. ;-)
will try the wf just to see what it could be
Yeeaaay it's time to move to Minimax H3 and delete everything about LTX2.3 :)

