zhangyiss/mlx-examples
1
0
Fork 0
mirror of https://github.com/ml-explore/mlx-examples.git synced 2025-10-24 06:28:07 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Stable Diffusion: Input image downsampling (#276)

Browse Source
This commit is contained in:
LeonEricsson
2024-01-16 22:45:00 +01:00
committed by GitHub
parent 2ba5d3db14
commit b4c20cc7f7
2 changed files with 11 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
stable_diffusion/README.md
View File

@@ -67,7 +67,7 @@ Image 2 Image
There is also the option of generating images based on another image using the
example script `image2image.py`. To do that an image is first encoded using the
autoencoder to get its latent representation and then noise is added according
to the forward diffusion process and the `strength` parameter. A `stregnth` of
to the forward diffusion process and the `strength` parameter. A `strength` of
0.0 means no noise and a `strength` of 1.0 means starting from completely
random noise.
@@ -78,6 +78,7 @@ The command to generate the above images is:
python image2image.py --strength 0.5 original.png 'A lit fireplace'
*Note: `image2image.py` will automatically downsample your input image to guarantee that its dimensions are divisible by 64. If you want full control of this process, resize your image prior to using the script.*
Performance
-----------