Instructions to use Glanty/Capybara with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use Glanty/Capybara with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("Glanty/Capybara", 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
Inability to maintain consistency in character portrayal during video editing tasks
The character consistency is terrible; it can't maintain the consistency of a character's identity. For example, if I give a prompt to change a character's clothes, their face will change to that of a completely different character.
Keep in mind that even a full-body 480p reference image can maintain its ID in WAN2.2 I2V.
This is a problem that shouldn't occur in an I2V video model; hopefully you can fix it.
Trying to add some words like "keep the face of the subject unchanged" or add descriptions of the subject in the editing prompt will help maintain consistency! Detailed prompting strategies to make the editing better can refer to the project page: https://lllydialee.github.io/Capybara-Project-Page/.