3 ms·
There was a good example somewhere I can’t find, but of a really complex prompt that Dalle could understand, but SD couldn’t. Maybe some of the GPT-3 is being l
by bottlepalm 4y ago
There was a good example somewhere I can’t find, but of a really complex prompt that Dalle could understand, but SD couldn’t. Maybe some of the GPT-3 is being leveraged for parsing.
Anyways I think it’s way too early to start taking sides. I enjoy using all these system.
- bioemerl 4y agoOne of SDs big limitations (understanding from what I had read about it) is positional prompts. dall-e seems to understand x on top of Y, but simple diffusion does not.
- posterboy 4y agoIronically, "the cat is on the mat" is a conventional example sentence in linguistics of metonymy (semantics). I have no examples but imagine things like at the top of his game are immensely problematic, albeit not very visual to begin with.
- adamsmith143 4y agoDoesnt seem to get IN examples either. E.g. a prompt like 'an eagle holding a snake in its beak' ends up generating eagle snake hybrid creatures.
- dogcomplex 4y agoimg2img drawing should take care of that
- bioemerl 4y agoI'm experimenting with the base model right now, I'm going to be very excited to try that one out too.