1,000🏅 Top traders
| # | Trader | Total profit |
|---|---|---|
| 1 | Ṁ140 | |
| 2 | Ṁ91 | |
| 3 | Ṁ74 | |
| 4 | Ṁ74 | |
| 5 | Ṁ18 |
People are also trading
@SG Does it? Stable Diffusion is pretty subpar in my experience, but then again, Dall-E is pretty much subpar too now that it has to be tweaked down for release
@JoyVoid Eh, I guess this is around on par with Dall-e 2.
Prompt: "cyberpunk fortuneteller with cash flying around"
@SG Yeah, that's getting to be about as good as Dall-E, well within my margin. I think most distinctions between them essentially amount to the way they respond to prompts. SD requires a different prompt approach and has a different flavor than Dall-E, but it's really up there once you learn how to use it.
@JoyVoid The margin was 80% in any case, and the differences there mostly have to do with style and I've seen stable diffusion examples that can close the gap once you account for the differences in how they take prompts.
@LarsDoucet That makes sense, thanks for restating the resolution criteria. My experience with playing with stable diffusion is that it's very hard to get anything good out of it, while dall-e needs very little direction (though it might just be that I haven't found the right way to prompt it correctly yet).
But I understand that if your question only compares peak performance, then this might not matter. I was only confused about SD being over 100% the bar, I don't have much opinion on 80% either way
@JoyVoid Cool, thanks for understanding! I expect that we'll get to 100% Dall-E quality pretty soon with an open model anyways, but ... that's for another market!