Instructions to use LiconStudio/LTX-2.5-Multiple-Subject-Reference with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use LiconStudio/LTX-2.5-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-2.5-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
Limitations
So i tried an rtx pro 4500 with msr, and it could show me 3 people well, but if i made them move or talk, it would bleed and not work... Is this the limit of MSR or LTX?
So i tried an rtx pro 4500 with msr, and it could show me 3 people well, but if i made them move or talk, it would bleed and not work... Is this the limit of MSR or LTX?
Yes,it is, you need to write ur prompt very careful for 3 or more characters.
is it better to use the other images as object like a pot or pan, or a tv, or a keyboard or an axe?
is it better to use the other images as object like a pot or pan, or a tv, or a keyboard or an axe?
Yes, MSR tends to work fairly reliably with two characters. You can use the remaining reference slots for any props needed in the scene.
that is great news... i will give it a try, i did try iclora but it was worse than msr, msr gets closer to the same character render after render than ic lora does for sure..