Same as model inside the waifu-diffusion-v1-4 repo
#1
by joujiboi - opened
Produces literally the exact same results as waifu-diffusion-v1-4
What is that about? 👀
You are probably confusing between autoencoder/VAE (which is separately provided) and UNet model. Autoencoder model is the part where latent diffusion model produces noise and has a slight effect on smaller parts, details and overall style. You need to load both autoencoder and UNet model.
naclbit changed discussion status to closed