>>3011
i've used AI to colour some images before. at minimum you'll need controlnets, particularly canny edge and/or hed boundary. they let you turn an outline into an image.
the images you got are fairly complex and cluttered, so helping it by doing rudimentary paint bucket colouring and then tossing that into image to image will help it figure out what's what. you still gotta do some rudimentary prompts to describe what should be in the image.
considering your source image will look terrible with generic anime style, you might also want to see if there's a lora of the artist, a lora of a similar artist, or if a model knows your artist (PonyXL knows a lot of artists, though they are hashed and some are yet to be identified).
if there isn't, a lora can be trained, though that's a bit more advanced, and a lot of artists get angry when you train artificial idiot on their art without their permission. i have, however, also seen artists who trained ai on their own art and started using it to help themselves, particularly kuroni who seems to be using it to colour his own sketches, exactly what you're after here.