openai/consistency-decoder

Consistency Decoder

This is a decoder that can be used to improve decoding for Stable Diffusion VAEs. To know more, refer to the DALL-E 3 technical report.

To original code repository can be found here.

Usage in 🧨 diffusers

import torch
from diffusers import DiffusionPipeline, ConsistencyDecoderVAE

vae = ConsistencyDecoderVAE.from_pretrained("openai/consistency-decoder", torch_dtype=pipe.torch_dtype)
pipe = StableDiffusionPipeline.from_pretrained(
    "runwayml/stable-diffusion-v1-5", vae=vae, torch_dtype=torch.float16
).to("cuda")

pipe("horse", generator=torch.manual_seed(0)).images

Results

_(Taken from the original code repository)_

Examples

Original Image | GAN Decoder | Consistency Decoder |

:---:|:---:|:---:|

!Original Image | !GAN Image | !VAE Image |

!Original Image | !GAN Image | !VAE Image |

!Original Image | !GAN Image | !VAE Image |