A common example is artists who draw on/off art of a character. They'll draw the character naked first, leave the face blank and draw increasing amounts of clothing onto the ghost along with the intended facial expression. If it's a commonly used character they'll have presets they can drag-n-drop onto the ghost or selection of ghosts similar to how Doki Doki Literature Club has numerous poses and expressions. This way the artist doesn't have to keep redrawing the same outline, the same hair strands and so on over and over again. It saves time but it still doesn't mean the process is quick. Compared to the irregularity of AI, it's simply not worth converting to such methods for this use case.
Doesn't even have to be games, animators have been using this sort of trick FOR DECADES. Every wonder why certain animated scenes appear to have the animated parts be a different shade than the background? Like when some rocks crumble off a cliff, the rocks are a different shade than the entire background? It's cuz cel animation. The animated part is drawn on a transparent plastic sheets (ie, cels) separately than the background plate, then combined to make a final composite animation. DBZ's insanely long and overly drawn out battles definitely illustrate this, but honestly animation in general uses this trick; Disney arguably perfected it...
Still, AI image generation advances will
eventually provide an alternative to hand drawing stuff, but the most advanced versions usually also have the most advanced censoring techniques for pretty obvious reasons... money. The same reason Gunsmoke Games keeps their Patreon
scam account running! OpenAI just dropped some new image generation features TODAY, the newer version based on 4o apparently no longer generates images in a purely convolutional manner (ie, all pixels all at once), the newer version is termed "autoregressive." Which should allow for being able to generate images in comic book panel styles.
Or maybe not, just tried to test those features and it's so new, it's actually not showing up yet. Still only giving DALL E image generation option for now...
Edit: As of Saturday, 3/29/25, new feature for autoregressive image generation is up on the free version of ChatGPT, it was only on the paid version until now (release was Tuesday). Interesting, but slower.