A second attempt to make a workflow for automizing comics creation out of a prompt and controlNET images input (when no sketches are avaliable). 2 characters are controlled by the ControlNET (intentionally, openpose and depth, but that's up to user), LoRAs and regional prompting. On first launch: > Load checkpoint; > Specify LoRAs (if you use only one, bypass the second lora loader); > Add global prompt to CLIP global +/-; > Add prompts for characters L and R; > Load ControlNET depth and openpose models (you can use other ones if you wish to) > Load images for controlNET and regional prompting; > Load upscaling model; > Adjust the initial latent resolution and latent upscaling factor (default: x1.25); > Load SAM and Ultralitics for the face detailer; > DO NOT use the GLIGEN (it seems that the conditioning in this workflow is already oversaturated)