I assumed it's mostly your code, so you may be able to help me.
I resized 10 images to 512x512, put them in a folder (I know it finds them as I mistyped the path and got error messages), I then create the embedding and get Done generating embedding message and the embedding is placed in the correct directory. It's done generating super quick but it only generates a 4k .pt file. Tried it with two different training sets of 10 images.
On the txt2img screen, I select a 1.5 variation that I trained with a face, then under Clip Aesthetic I select the aesthetic that I trained, and have varied the steps from 10-max, and changed the weight from .1 to 1 but nothing happens.
I installed the an extension version of your code via extensions in 1111, https://github.com/AUTOMATIC1111/stable-diffusion-webui-aesthetic-gradients
I assumed it's mostly your code, so you may be able to help me.
I resized 10 images to 512x512, put them in a folder (I know it finds them as I mistyped the path and got error messages), I then create the embedding and get Done generating embedding message and the embedding is placed in the correct directory. It's done generating super quick but it only generates a 4k .pt file. Tried it with two different training sets of 10 images.
On the txt2img screen, I select a 1.5 variation that I trained with a face, then under Clip Aesthetic I select the aesthetic that I trained, and have varied the steps from 10-max, and changed the weight from .1 to 1 but nothing happens.
Any ideas?