import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("Nilaier/Waifu-Diffusers", dtype=torch.bfloat16, device_map="cuda")
prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
image = pipe(prompt).images[0]Repo Overview
A version of Waifu Diffusion v1.4 Float32 Booru 110k model and VAE from Waifu Diffusion v1.4 converted to work with Diffusers library.
^ Quality/Style test
^ Unnatural resolution test
^ Landscape test
^ Extremely long resolution test
Model Description
The current model has been fine-tuned on a Stable Diffusion 2.1 model with 110k anime-styled images using a technique known as aspect ratio bucketing.
That allows Waifu Diffusion v1.4 to handle different resolutions much better than its previous models.
Source
WD v1.4 VAE: https://huggingface.co/hakurei/waifu-diffusion-v1-4/blob/main/vae/kl-f8-anime2.ckpt
- Downloads last month
- 19