My friend recently got his master’s degree in computer science and he has access to the full version of DALL-E (not just DALL-E mini which was released for the public to play with not too long ago). To show a comparison of how DALL-E compares to DALL-E mini, consider the above image, which was an output from DALL-E to the prompt of “a hamster in a cup.” Almost photorealistic! Now take a gander at the BEST of the outputs given by DALL-E mini to the same prompt:
Some of them aren’t TOO bad, but they all have problems (these become more apparent on magnification), and outputs like the unidentifiable pointy-headed creature-or-maybe-object in the upper left corner occur with close to 50% frequency. And things like the flabbergastingly inexplicable tubular hamster in the middle of the top row of the following simply don’t get produced by the full version of DALL-E. Which begs the question - which one is actually more “interesting”? (The answer is both are, depending on what, exactly, you are interested *in* of course.)
