>>103635797
I recently looked into comfy because I wanted to try out the new wan stuff. Before getting into the video stuff, I wanted to mess around getting txt2img/img2img/inpainting working so I could understand how to build workflows.

My personal conclusion is that once you understand the idea of when to use nodes and how to connect them and install some utility node packs, it's incredibly powerful and you can choose how finely you want to customize the process. I was able to mostly replicate my txt2img workflow from reforge with wildcards and lora scheduling with better control and it gens much faster. But also things that are a one-click feature in Forge can take several nodes and workflows in comfy. And if you have a large workflow, it can be cumbersome to pan around to find settings to change or organize the most important nodes in a way to make them easier to find. Something like sending an image from txt2img/pnginfo to img2img/inpainting with the prompt and settings and everything is something I took for granted in the web UIs. The way I was able to do it in comfy is more than just one click and feels clunky.

In the end, I still use the web UI to gen and inpaint my images and then toss them into my comfy wan workflow. Take this with a grain of salt because I'm new to comfy and don't know everything there is to know, so there may be easier ways to do the things I struggled with.