Re:Ten Years of Image Synthesis
by multimodal on Tuesday November 29, 2022 @08:04PM
(#33795585)
Maybe I'm misreading the tone of your comment - but is that a bad thing? Reframing the approach to a problem, to state-of-the-art results; that strikes me as nothing if not progress. Don't mean to come across as negative here -- I just don't see why anyone would control those variables.
You list reasons why text-to-image is easier than unconditional image gen - but isn't that the point? They're both under the umbrella of image generation, which is ultimately pushed forward.
You list reasons why text-to-image is easier than unconditional image gen - but isn't that the point? They're both under the umbrella of image generation, which is ultimately pushed forward.