About a year ago I was still using WebUI, swapping between A1111, Forge, and ReForge - always frustrated by how it felt like WebUI was constantly playing catch-up with ComfyUI. I decided enough was enough and finally jumped head first into Comfy.
First thing I wanted to do was build an "ultimate workflow" with toggles for conditional branches. It got messy fast. Any time I wanted to add something new or switch things around the workflow got bigger. I had the idea for creating re-usable, re-arrange-able workflow segments I could use to quickly build up any workflow I wanted, bespoke for the piece I had in mind.
But Comfy isn't really built for that. The node interface makes it very powerful but it also makes it tedious when what you want to do is make art instead of manage noodles. Basically, I wanted ComfyUI for when I want to build new workflow segments, but something like WebUI for when I want to actually gen.
So I got to work building SugarSubstitute. It's a desktop app built in PySide6 designed to stay performant even when inference is lighting your GPU on fire and it can connect with remote ComfyUI instances, too. It can even set ComfyUI up for you with an easy to use wizard. You can probably tell it's built to feel native on Windows, but it should work nicely in most places Python runs.
The editor is set up to filter out the noise of a normal Comfy workflow; no noodles, no scatter of nodes, just a wall of the controls you actually need. Some places in the editor get special attention, too. Substitute's model pickers query CivitAI for thumbnails so searching through your archive is beautiful and easy. The multi-line prompt editor is a rich text editor designed for prompt editing, too! Booru tag autocomplete for anime models, rendered decorations for common prompt syntax like emphasis - it even supports scheduling LoRA in the prompt editor itself out of the box, just like WebUI.
I've been doing image gen from just about the very beginning and I understand what kinds of pain points exist in our workflow. Substitute is filled with little details to make things easier for you to get actual work done more quickly and with less friction. Easily compare between different output levels, send an output directly to the canvas of your favorite editor in two clicks (Gimp, Photoshop, Krita are my targets for release), save and re-use your favorite image dimensions from the context menu or even swap them around when you want to go from portrait to landscape. Listing every single little creature comfort of Substitute would have us here all day!
You don't just have to use the graph segments I built, either. You can easily create your own on the Comfy graph and port them into SugarSubstitute. That means if your question is "does this support x or y model" the answer is: If ComfyUI supports it, SugarSubstitute does, too, with the lone caveat that the canvas currently only supports still images - I'll get to video, and eventually 3D and other formats, after the initial release.
When I release it, it'll show up on my Github. I'll be publishing it under the GPL, free for everyone! I'm posting about it here on Reddit for the first time because I wanna know: Is this something you'd be interested in using as a regular ComfyUI user? And what kinds of features would you want to see in an app like this?