Spaces:
Running
Gallery of prompts
Share your input text and any keyword tips (such as: detail, illustration, style of, depth of field, etc.) and other grammatic tips and tricks, along with some example outputs.
"Illustration. Detail of Basset Artesian Normand face. Style is cross-hatching with ink-pen."
(The dalle-mini app delivers image results to your browser as a base64 data stream. You can inspect the result page and cut-n-paste img tag(s) here. It is adviseable to remove the class attribute from the img tag).
"Illustration of the planet Saturn in the style of a bacon lettuce and tomato sandwich"
StarWars full cast group photo oil painting
madoka and sonic ice-skating under a snowstorm waterpainting
@funkylchor why remove class attribute?
Hatsune Miku painted in the style of Zdzis艂aw Beksi艅ski
This model adopted Beksi艅ski's style surprisingly well, maybe it has something to do with how their art is already looks uncanny. (i forgot to make this remark in my earlier comment sorry)
BTW... we are working on an upload image feature inside comments... Hoping to ship that in the next 2 days!
synchronized swimming cats
madoka floating over ancient city in the style of starwars
Madoka isn't anywhere to be seen in most images, but somehow the prompt has made the cities look much more impressive than [floating] ancient city in the style of starwars
I tried various characters floating over the city, it was very interesting how de different charcters impart personality to the city. However this often causes the charcter to appear on the scene, wich is not desirable. I'm trying to figure out a prompt to do this more generally, without causing the character to appear on the scene.
A representation of DALL-E by hitself
A self-portrait of DALL路E. In the style of a mural.
(In response to ANDevereaux)
"A self-portrait of DALL路E. In the style of a mural." gave these kind of results:
Thanks, Dubreek. I couldn't figure out how to post the images, but they were really cool :)
Dall-e seems to understand emojis well. To the point that (at least with my limited testing) you can put in a few to get an output.
For example I put in 馃憠馃Ж馃巼 at one point and it actually outputted images of a hand holding a sparkler
Here are some of my results, enlarged with Bigjpg. Prompts are in the title, liks to Flickr: