Instructions to use LiconStudio/LTX-2.3-Multiple-Subject-Reference with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use LiconStudio/LTX-2.3-Multiple-Subject-Reference with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("LiconStudio/LTX-2.3-Multiple-Subject-Reference", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
Injecting frame.
Hello,j you're lora is amazing,and respect pretty good the input images consitency ! thanks for that.
i have a just a little problem, with injecting frames, it gives me error, i tried to edit the workflow, but still getting troubles with it. the problem is the conditionning part. i've tried different variotion or roger but still the same problem.
all errors look like this , ValueError: guide pre_filter_counts (1152) != keyframe grid mask length (3072)
this looks like a shape/grid mismatch between the reference video conditioning and the target latent video. The guide side produces 1152 grid elements, while the keyframe mask expects 3072, so their latent grids are not aligned.