>>16148
For consistency, I establish the overall framework using 3D and Photoshop.
After importing the image into ComfyUI, I repeatedly apply i2i and inpaint to complete the illustration, then use i2i again to reduce detail for readability.
The model uses multiple LORAs on wainsfwillustrious_v130.
When increasing detail by switching between Photoshop and ComfyUI, I use LORAs trained with 3D and Blender. When simplifying, I use LORAs like flat color and lineart.