What if we had an AI-powered drafting machine for comics layouts?
Let’s say we sit down 300 comics artists in a university exam hall. We give each of them three scripts, and ask them to devise a 12-page comics layout for each. These scripts are standard storytelling, with beginnings, middles, ends. A limited number of characters. The usual action. The usual timeframing.
We specifically ask the artists to do quite rough layouts that only involve architecting the frame size and position, selecting the usual stock camera angles, indicating character positions, and suggesting some basic props and vehicles. No complex backgrounds. Perhaps a different colour could be used for each element, to help the A.I. decide which is which — frames in green, buildings in blue, characters in red, props in orange, balloons in pink.
Could the A.I. then analyze each frame and each page of all these standard-layout comics, ‘seeing’ how they match to the relevant points in the three stories and finding commonalities of approach among the artists?
Could it then devise enough machine-learned rules, so that when it is fed a fourth story, it can automatically ‘draw’ its own rough ‘machine layout’ that fits that new story pretty well? Needing only some adjustment by humans here and there, and perhaps an application of a hypothetical ‘Kirby-ize filter’ to get a little more foreshortening and ‘pop’ on characters and props?
Maybe. Such robo-assistance will surely come, eventually, especially for the more formulaic type of comic. In the meantime, a hypothetical giant “Encyclopaedia of Comic Books Layouts and Camera Angles” would be a handy thing to have. When your comic script demands a “bird’s eye view down into an alleyway”, you’d just type ‘alley’ and set the camera-height slider to ‘high view’, and 24 curated examples from historic comics pop into view to inspire you.