Withered Bonnie from FNAF 2 for SD2.1

Note: Development on this model has been halted for the foreseeable future

Quality concern: This model was trained with a low quality, small and narrow dataset on a low learning rate, this model can only produce low quality samples, mostly of Withered Bonnie alone in a beige void

This is a experimental model made to test my training script, image captioning abilities, basic dataset manipulation, etc.

In the future (maybe), better ways of making captions, like training a specialized captioning model, better training infrastructure, like an actual GPU (I trained this on my laptop's CPU, took ~2 days), better image synthesis, transformation and editing, etc.