Instructions to use Lightricks/LTX-2.3-22b-IC-LoRA-Pixel-Spatial-Upscaler with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- LTX-2
How to use Lightricks/LTX-2.3-22b-IC-LoRA-Pixel-Spatial-Upscaler with LTX-2:
# 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 Lightricks/LTX-2.3-22b-IC-LoRA-Pixel-Spatial-Upscaler --local-dir models/LTX-2.3-22b-IC-LoRA-Pixel-Spatial-Upscaler hf download google/gemma-3-12b-it-qat-q4_0-unquantized --local-dir models/gemma-3-12b
# Video-to-video with the IC-LoRA (runs on the distilled base model) uv run python -m ltx_pipelines.ic_lora \ --distilled-checkpoint-path path/to/distilled_checkpoint.safetensors \ --spatial-upsampler-path path/to/spatial_upsampler.safetensors \ --gemma-root models/gemma-3-12b \ --lora models/LTX-2.3-22b-IC-LoRA-Pixel-Spatial-Upscaler/<weights>.safetensors 1.0 \ --video-conditioning reference.mp4 1.0 \ --prompt "your prompt here" \ --output-path output.mp4 - Notebooks
- Google Colab
- Kaggle
still crash the faces
not suitable to high motion video
Can you please provide the source asset and the prompt? Which strength did you use?
your workflow ,strength 1. high motion video from bernini 480p, 480*832
There is no need to insist on ltx2.3 i think, which has too much weakness. high motion is impossible in ltx2.3. if ltx3 can catch up with seedance 2.0 mini or fast in 6 months, it will be great
There is no need to insist on ltx2.3 i think, which has too much weakness. high motion is impossible in ltx2.3. if ltx3 can catch up with seedance 2.0 mini or fast in 6 months, it will be great
Your attitude and lack of skills is what makes your goals not achievable. The model is already more than capable if you have the skill and compute.