This project enables users to paint directly in Photoshop, generate an image using the Stable Diffusion server, and view the updated image in a Jupyter Lab window. Change prompts by editing layer names and special options. Knowledge of Stable Diffusion basics is required.
To use this project, you will need to install the following:
(+additional packages required to use .py script)
To use this project, follow these steps:
- Open a Photoshop
.psdfile(use example file)and start painting. Save .psd fileand generate the image using the Stable Diffusion API.- View the updated image in a Jupyter Lab window.
(.psd file contains 3 groups which are necessary for proper operation)
By default, each group is locked (meaning the settings cannot be changed). Each group already has samples that you can edit or delete.
- Unlock any group, set
Activatefor existing layer, or create a new one. - Save
.psdfile, after that, an image will be generated with new settings. - Lock the group if you do not plan to change the settings again.
*(You can leave unlocked, but the computer will process all layers each time you save your image)
This is where you control the prompts that decide how your art will look.
Rename each layer to describe the style, objects, or actions. You can separate each line with a comma or simply create a new layer.
Activate(eye icon) allows to enable or disable prompt
In each group, you can control image generation [WEIGHT]
Opacitycontrols [Weight] by 0.1.Effect layerincreases [Weight] by 1.0.Lockadds [AND] into the prompt (reduces image generation speed).
The main group where you create your art.
You can create, edit, and paint on any layer.
- This group doesn't have any prompt settings.
- You can name layers as you want (they don't influence prompt generation).
This project was inspired by the Unleashing the Creative Power of Stable Diffusion showcase by OpenAI.
Information on how to name layers Some detailed notes.