Which prompt for Dall-E 2 will I be most impressed by?
11
1.1kṀ146resolved Jun 24
40%1.0%
Sunrise on Mars, oil painting by Thomas Moran
30%8%
A ray of light piercing through a dense forest, as though it were God's touch, hyperrealistic digital art
20%9%
Niagara Falls at Sundown (1830), Oil Painting on Canvas by Thomas Cole
10%1.4%
asmodeus funko pop
30%Other
1.2%
pillowfort city
0.8%
waterfall on venus
9%
an abandoned overgrown server room, cinematic
8%
A stunning impressionist painting of a space shuttle taking off from a beautiful green meadow in the middle of an old sequoia forest
0.0%
claymation octopus holding a trident
1.4%
shadow of a tree against a wall, photograph
0.0%
mouse solving a maze shaped like a paper crane
1.7%
a stained glass window in an old gothic cathedral
3%
Oil painting of a young shepherd, Caravaggio 1608
7%
Emulated minds
3%
Oil painting of two cats staring at a goldfish
0.9%
The most beautiful woman in the world, a photograph by Helmut Newton
1.0%
“Horseshoe Bend”, photograph by Ansel Adams
1.9%
Jade carving of a phoenix, Qing Dynasty Chinese art
0.9%
Fire at Sea, shipwreck painted by J.M.W. Turner (1840)
I have recently received access to Dall-E 2. I will enter each response to this market as a prompt, and resolve with the prompt which generated the coolest/most impressive/most beautiful images.
Jun 24, 4:00am: An unbiased judge (my twin brother) chose this resolution. Those were great prompts!
This question is managed and resolved by Manifold.
Get 
1,000 to start trading!
1,000🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ19 |
Sort by:
@JoyVoid It didn't seem to improve at all, all of them still had face and hand deformities. https://labs.openai.com/s/T2Ty2Nxhnw1Q0WfuTm49YwGO
@Angela Too bad we can't tip answers yet :( ... If you comment here, I'll tip you for this prompt :D
https://labs.openai.com/s/wuOE2oXYWebDjZQeEKPZYnhu
Frankly, they almost all had some wild face deformities. Dall-E 2 isn't very good with faces
@SneakySly Yeah, and pretty much all other images had mostly text in them and nothing which looked like the prompt. I think that longer descriptions give the model less wild creativity, and that tends to lead to better outputs