Hello there and thanks for checking out this workflow!
—Intro—
Out of curiosity for the PixArt models I built a few workflows to try them out and was really impressed by the prompt adherence as well as by its style, which is a little more raw, more random, natural I'd say compared to what I was used to with only SD models.
Some prompts I've written months ago never really came out in SD generations—nowhere near to how I imagined them anyway—but PixArt got a lot closer.
—Issue—
Unfortunately img2img and upscaling are not really great options at this point, not with PixArt only that is. While the UltimateSDUpscale node was quite capable of producing higher resolutions without the image falling apart, it still looked overly smooth and usually misproportioned.
—Solution—
Simply generating the 1st pass with PixArt and using SD models in further passes to refine the image, similar to the concept of SDXL's refiner in its early days.
SDXL and SD15 both work great as refiners for PixArt and I definitely recommend trying out both.
I've built this workflow with that in mind and facilitated the switch between SD15/SDXL models down to the literal virtual flick of a switch!
—Custom Nodes used—
All of which can be installed through the ComfyUI-Manager
If you encounter any nodes showing up red (failing to load), you can install the corresponding custom node packs through the 'Install Missing Custom Nodes' tab on the ComfyUI Manager as well.
—Thanks—
The workflow would not be possible as is without these custom node packs. If you want to support the custom node creators, give them a ⭐ on their github repos! Thank you!
Feel free to ask any questions, share improvements or suggestions in the comment section!
Also let me know if you encounter any confusing points I can elaborate on and focus on improving for the next update!
Description
v1 — initial release