Testing a theory that dalle2 does worse at generating things the smaller they get, I discovered that it makes really mean eye charts.



@janellecshane iirc at least one other generative net works by generating the image at like 1/16th the final resolution, then upscaling it using text-based neural upscaling. which would explain this kind of behavior!

Sign in to participate in the conversation
inherently digital

a very robotic single-user instance