Config tabs - JasonS09/comfy_sd_krita_plugin GitHub Wiki

SD Common Options tab

This tab will contain the basic common options used for all the modes in this plugin. Here, you can select your model, vae, clip skip, and batch size. If None is selected as VAE value, the workflow will use the VAE output from your model. There are also other options which are more specific to the plugin.

Base size and max size

These come from the original repo this one was forked from (check https://github.com/Interpause/auto-sd-paint-ext#faq).

Setting base size will apply a constraint to your image generation, meaning that all the outputs will have as minimum size, the base size that you set. Max size will constraint the maximum size of your image generation. The image aspect ratio will be defined by your current selection or canvas. If one of the dimensions of the image gets to be bigger than the max size, then the minimum size of the output will be smaller than your base size (hence why it's not called min size).

Setting base and max size will allow you to set an upscaler also. Setting an upscaler will generate a second pass (now optional) in the workflow to upscale the output so it reaches the size of your selection or canvas. This second pass will kick only if needed and if an upscaler is selected. In the dropdown list you will find both upscaler models (if any found) and upscaling method. Picking a method will run a simple latent upscale, picking a model will upscale the image with that specific upscaling model. If no upscaler is selected, output image will simply be resized once it reaches Krita to fill the selection or canvas.

Users can also disable base and max size, so the output will be exactly of the selection or canvas size. In this case, no second pass for upscaling is needed.

SD Plugin Config tab

This tab contains some config options for the plugin itself, as well as the field where you should put your ComfyUI instance url. Note: Some of the options here come from the original source and are probably obsolete (for example, filter NSFW).