A workflow that combines different styles (RGB mask and unmaked black as default condition).
The workflow works just as well if you leave it promptless, as the previews showcase, since the pictures are auto-tagged.

How to use - explanation group by group

Main Loader
Select checkpoint, LoRAs and image size here.

Mask
Upload the RGB mask you want to use. Red goes to the first image, green to the second, blue to the third one. Any unmasked (black) area will use the unmasked image.

Additional Area Prompt
While the workflow demonstrates the results without prompts, you can prompt each area separately as well here. It will be concatenated with the auto tagged prompts taken from the image.

Regional Conditioning
Upload the images you want to use the style of per area here. Unmasked image will be used for the area you didn't mask with RGB colors. Base condition and base negative are the prompts to be used by default, that means it's also used for any unmasked areas. You can play around with different weights for images and prompts for each area; if you don't care about the prompt, only the image style, set that to low weight and vice versa. If more advanced, you can adjust the IPAdapters' schedules and weight type.

Merge
You can adjust the IPAdapter type and combine methods here, but you can leave it as is unless you know what you are doing.

1st and 2nd pass
Adjust the KSampler settings to your liking here, as well as the upscale model and upscale factor.

Requirements

ComfyUI_IPAdapter_plus
ComfyUI-Easy-Use
Comfyroll Studio
ComfyUI-WD14-Tagger
ComfyUI_essentials
tinyterraNodes

You will also need IPAdapter models if the node doesn't install them automatically, you can get them via ComfyUI's model manager (or GitHub, civitai, etc, whichever you prefer)