overgeeked
Dragonbane
Maybe. I can get it to spit out all kinds of dark-skinned women. It might be the Frazetta. Most of his women would spoil the generator. Nope. Pulled that reference out too and it was still blocked.
D&D General DALL·E 3 does amazing D&D art
- Thread starter M.T. Black
- Start date
EzekielRaiden
Follower of the Way
I recommend dropping the word "muscular", and probably "powerfully built", in addition to the tattoos, unless you specify that the tattoos are on her face. Including "muscular"/"powerfully built" means that the thing is going to want a way to show that she's muscled, and that is typically done via arms. Hence why every part of her is heavily armored except the arms, where the flex-able bicep muscles are. Likewise, most tattoos are either on the arms or torso, so those are going to encourage it to have her be "fully armored" while uncovering those areas that have the tattoos.
Scribe
Legend
Amusingly Frazetta as a prompt key works best at giving men a righteous chiseled jaw line.
EzekielRaiden
Follower of the Way
Alright. Spending a couple of attempts. First one only generated a single image, so that's not ideal.
Prompt: "Fantasy watercolor dark age comic book art. Gritty, ethereal, aggressive mood, soft focus. A tall, powerful female stands on a rocky outcropping in a volcanic setting, wearing full-body black armor. She is powerfully built, with short windblown fiery hair, onyx marble skin, and spiral red tattoos on her face. In the style of Frank Frazetta."
Second attempt...also generated only one image, but I had to take four attempts just to generate anything. Slightly rephrased various bits. The tattoos seem to have migrated from her body to her armor, which is...not what was intended, but it works, I guess?
Prompt: "Fantasy watercolor dark age comic book art. A tall, powerful female warrior with onyx marble skin, short windblown fire hair, and spiral red marking tattoos, stands on a rocky outcropping in a volcanic setting, wearing full-body black armor. In the style of Frank Frazetta. Gritty, ethereal, aggressive mood, soft focus."
(Semi-)Final attempt:
This time, I tried using the NightCafe version rather than the Microsoft one. I think it turned out...somewhat better? But still not quite what you're looking for. Same prompt as the previous, and again it only generated one image. I kind of get a slight Sabriel/Lirael/Abhorsen vibe from this one.
Edit: Correction, the above was not using DALL-E 3, but rather "Flux", whatever that is. This is using the actual DALL-E 3 model from NightCafe.
True Final Attempt:
Same prompt as before. Actually, not half bad, albeit still not quite what you're looking for. Not really sure why it's suddenly in an ocean area rather than a volcanic lanscape, but it's among the better results thus far.
Second attempt...also generated only one image, but I had to take four attempts just to generate anything. Slightly rephrased various bits. The tattoos seem to have migrated from her body to her armor, which is...not what was intended, but it works, I guess?
(Semi-)Final attempt:
Edit: Correction, the above was not using DALL-E 3, but rather "Flux", whatever that is. This is using the actual DALL-E 3 model from NightCafe.
True Final Attempt:
overgeeked
Dragonbane
All that’s one reason I’ve gone to a rougher, more concept art style with most of the “AI” art. Keep things vague and less detailed. It will surprise me and I can enjoy it instead of sweating the details and getting frustrated over not getting exactly what I want. More interesting output and fewer wasted prompts.Amusingly Frazetta as a prompt key works best at giving men a righteous chiseled jaw line.
Scribe
Legend
I'll have to check I would think 'boob' would just kill the prompt dead.
I have not scrolled through this thread in a while and here is my current take (which some of you may notice has evolved over time):
AI generated art (including writing and music as well as visual art) "works" insofar as most people do not examine art deeply. They glance, maybe even occasionally linger, but rarely examine (or read deeply, or listen fully). There are lots of pieces that are great at a glance in this thread, but none that stand up to scrutiny. The reason for that is that generative AI algorithms are not thinking things. They aren't making decisions. And they certainly aren't, in and of themselves, creating art. They are generating pixels (or words or beats) within a probably expected set of parameters.
This is a good metaphor for art and culture in the modern age, i think. AI art will take people's jobs not because some people want to use AI to generate art -- I think the promise of AI art actually makes art inclusive in its own way -- but because as a culture our consumption of art is so shallow and so untrained in eye, ear and mind that the "slop" looks like art at first glance and very few of us care to examine it further.
AI generated art (including writing and music as well as visual art) "works" insofar as most people do not examine art deeply. They glance, maybe even occasionally linger, but rarely examine (or read deeply, or listen fully). There are lots of pieces that are great at a glance in this thread, but none that stand up to scrutiny. The reason for that is that generative AI algorithms are not thinking things. They aren't making decisions. And they certainly aren't, in and of themselves, creating art. They are generating pixels (or words or beats) within a probably expected set of parameters.
This is a good metaphor for art and culture in the modern age, i think. AI art will take people's jobs not because some people want to use AI to generate art -- I think the promise of AI art actually makes art inclusive in its own way -- but because as a culture our consumption of art is so shallow and so untrained in eye, ear and mind that the "slop" looks like art at first glance and very few of us care to examine it further.
Yaarel
🇮🇱 🇺🇦 He-Mage
Generally, yes. Except when one does need the AI to do something specific.
