Airpen is a start-up with a great team which is building software to generate images and illustrations from pure text, on demand.
We want our customers be able to create vector compositions based on ordinary speech, that is, in essence, to make speech-to-illustration software.
This is somewhat different from existing image generation algorithms, because our goal is not to create pixel pictures, but to generate on-the-fly combinations of the existing library of primitives (people, objects, backgrounds) and their variations (hair color, poses, mutual arrangement) at the user's request.