Further investigation into my main working methodology, this time attempting to force two separate GAN models to dither each others output. This didn't produce what I would consider actual dithering but something interesting..