I'm looking for a resource for programatically constructing 2d scenes, ideally with an existing library of images including backgrounds, characters in different poses, and objects. something like wordseye maybe, but with a clearer input text -> output image semantics?
If you're interested in procedural generation, Daniel Ritchie (https://dritchie.github.io/ ) has done a lot of work in between probabilistic programming and graphics, e.g. "Neurally-Guided Procedural Models: Amortized Inference for Procedural Graphics Programs using Neural Networks"
cognitive psychology. PhD