D&D General DALL·E 3 does amazing D&D art

[Zaukrie]

four armed demons all seem to have 6 arms.....

 

[schneeland]

I think many of them are excellent, other than the ridiculous ears.

The ears are definitely strange, but there's also something about the faces that's off-putting to me. Probably a bit of an uncanny valley effect, because they look almost human and just a few details are weird.

 

[Zaukrie]

white dragon flying over village in style of dungeons and dragons......too bad it looks like a mediocre dragon on a green screen, because these are close:

 

[Parmandur]

I think many of them are excellent, other than the ridiculous ears.

The ears are definitely strange, but there's also something about the faces that's off-putting to me. Probably a bit of an uncanny valley effect, because they look almost human and just a few details are weird.

Interestingly, the default Elf seems to be anime influenced.

 

[Zaukrie]

viking village at the end of a fjord being attacked by giants fantasy art

 

[Parmandur]

pig-faced green-skinned rainbow-haired strong agile lithe graceful beautiful robust Orc lady wearing rainbow leather jacket over silk blouses with a fashionable purse and gawdy jewelry playing a concertina and singing bawdily, winking at the camera, tavern stage, high quality digital painting

 

[Zaukrie]

a viking warrior stands on a snow covered cliff fighting a troll.....ok, I'm in love with one of these. They are all decent or better. Why do trolls wear clothes in one spot?

 

[Parmandur]

a viking warrior stands on a snow covered cliff fighting a troll.....ok, I'm in love with one of these. They are all decent or better. Why do trolls wear clothes in one spot?

View attachment 300067
View attachment 300066
View attachment 300065
View attachment 300064

Bijg is a family service.

 

[Jfdlsjfd]

Yeah, having more than one character will confuse the AI about which adjective goes where: problematic if you want anyone of two different species in the same shot.

Yes, "concept bleeding" is something that is yet to be perfected. Other models have implemented regional prompting, where you create an image based on several partitions of the final image and the AI stitches them together in the end result. It is more involved than just prompted. I am half-tempted to spend 20 bucks on chat-gpt pro to see if the integration with chat-gpt allow such a workflow through speaking with the interface to get the desired changes... but I fear it will not and I'll have lost 20 bucks.

 

[Zaukrie]

I think I could use this as a visualization for an adventure location that I then write about.....