Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
When there's a full body shot and you zoom in on the face, it's awful.
Midjourney doesn't allow lingerie in prompts but the images don't come out very well here. The bedroom background looks decent at first sight but again when you zoom in, so many flaws. MJ lets you correct wonky sections. There has to be something to this that is better than Midjourney or why? If we could use our own products and describe in detail how to pose and light the products and make them look less "rendery" and either photorealistic or stylized fantasy art, that would be great but right now I don't see an advantage.
Couple more that I liked. The first made me think of Marion Zimmer Bradley's Darkover novels. The second brought up memories of Esther Friesner's Chicks in Chainmail anthology series. Neither had anything to do with the prompts I used.
Remember it's only Beta
BTW the top of the bed is two levels .....
I suspect this beta does not have another very necessary feature yet, and that is the ability to do Face Restoration. Which is a thing I never render without when using Stable Diffusion, that way I mostly avoid facial disasters. Probably get it when Upscaling comes to Daz AI since the two are related. But you can probably find some free online sites that will improve your images while waiting for the next Beta update.
Example: smaller (1024 x 1024) image is right out of Daz AI Studio. Not bad but the eye that is not in shadow is a bit wonky. So I ran it thru a facial restore/upscaler (google ARC or tencent arc) and now he looks much better.
make art button
It's still winter round here.
I agree.. with all the fiddling you'd have to do to 'fix' this AI it would be quicker to set up in Daz and be right the first time... lol.
prompt:
a man in a white flight suit with a red visor running towards a white fighter aircraft in the hangar of a space ship, view from behind, red neon lights, rim lighting, depth of field, science fiction,
###
prompt variations:
a close-up portrait of a man from switzerland wearing a white pilot suit, cheerful, in the landing bay of a space ship, red neon lights, rim lighting, depth of field, science fiction,
###
a man in a white flight suit with a red visor, a white fighter aircraft in the hangar of a space ship, red neon lights, rim lighting, depth of field, science fiction,
###
I thought I'd give this a try out of sheer curiousity. I've never used AI to generate images before, and while I prefer creating images using Daz Studio, Blender and GIMP, I do feel AI could be a useful tool. Perhaps as a way to tweak final renders.
These are my first attempts. I tried to get the AI to generate an image I have in my mind that I plan to create in Daz Studio. 'Child hugging fox with forest background' didn't give me anything at first. The generator lagged out and crashed on me. So I tried the prompts 'Girl hugging fox with forest background' and 'Little girl hugging fox with forest background' I later added extra prompts such as 'autumn' because the autumn/fall colors match the fox's fur and girl's outfit better.
However, one thing I have noticed is the posing is a bit random. Despite writing 'hugging' 'or 'hugs' for the majority of my attempts, often the characters were standing. So I guess there's other prompts I have to add to fine tune the posing and whether or not the characters face the camera?
I also noticed the generator doesn't seem to like the word 'child' in the prompts. I don't really know why that is, since I love rendering scenes of children with fluffy animals for the cuteness overload. Is that because there's not enough child characters in the AI training algorithm?
A group of goblins, causing mischief and chaos wherever they go
The dark skin is beautiful; the dress on this picture is sensational. The hairstyle and the headdress are also well done. The programme just needs to work on the faces and hands... I have a mermaid with a human hand instead of a fish tail, lol. Too bad, because the rest of the picture is good. That's why I keep it and edit the spot in Gimp or Affinity. It's much better for close up portraits.
'Monsoon rain in an old Indian town. A woman in a saree laughs as she hops over large puddles.'
It looks like the road is a bit flooded and has more than just a few puddles, but apart from the hands, I like the picture of the young woman.
"Beautiful sphinx lounging on a divan in a palace, holding a fan in her hand, dark brown hair, Egyptian, Greco-Roman, 8k, photorealistic,"
Two of the results were good, and conveyed the tone I was going for, but they were just regular women. (I plan to edit these later.) Then, there were these two that also conveyed the tone but that "pet sphinx" is funny to me, and IDK what kind of animal that second one has (Looks like a baboon.)
Royal
Beautiful Cinderella fairytale vibes. Plus we have the ability if we choose, to correct the incorrect hands, with a little 3d magic called DazStudio.
mature woman birthday party and her husband kissing her
These came from "golden griffon on a mountaintop at sunset". I don't know what prompted the last one.
It sure doesn't know what a black hole is. These are the prompts I tried:
"massive cruise ship in space being sucked into a black hole"
"massive cruise ship in space being drawn into a black hole"
Even tried breaking the word "into" in two words - no dice:
"massive cruise ship in space being pulled in to a black hole"
Finally gave up after it stopped creating round ships but still failed to produce a black hole with this:
"massive cruise ship in space being pulled in to a vortex"
The last image is the closest it came to producing what I wanted to see.
Sure seems to know what a 'stargate' is though... LOL. Even though you didn't ask for a stargate!
Prompt - Woman getting drenched in a heavy windy rain storm in New York City
Prompt - Black Neon Fire dire wolf in a volcanic landscape with embers flying in the air and burnt trees. Make the atmosphere filled with smoke.
These came out amazing! Wish it would let me make large versions for print.
Prompt - Woman in a bikini getting drenched in a heavy rain storm on a California beach.
Thought I'd try a drenched in a rain storm image as well. And so far I'm impressed that when I say "riding a horse" the figure is actually riding a horse... despite the mangled baby hands.
It seems to have a problem with hands and eyes. Several of my images I had to scrap because the hands or eyes or both were messed up.
Close, but no cigar. It flipped the characters in this prompt: "man in shadows in background looking at scared woman in foreground on fog-covered pier."
I am new to AI image generation ... so I have been testing it in many genres and concepts. Most had short to moderate length prompts - a few I re-prompted a few times to fix. So here, in a few posts, are some results. Generally, I find that it is hit or miss - it either knnows a topic and does well, or it does not know a topic and does poorly.
In this post:
) perhaps my best and most re-worked image with a gardener
) my drenched woman scene - located at a Bus Station [Daz has so many you know?]
Now for some of the successful images [ successful being defined as main content is correct, even if minor details like fingers are wrong ]:{listed are topics, not prompts]
) steam train
) middle eastern woman in a garden
) city destroyed by earthquake
) preacher close up
) astronaut with camera
) lost city of atlantis
Now for some images that are 'not quite right':
) Sumo boxing
) not sure what type of submarine this is ... but I do not want to fight it ... the octopus might take it down though
) I have heard of missing leg day at the gym ... but only training one arm !?
) Not sure what the game is ... but is the dragon with the broken wing for the winner or the loser?
) well ... you usually don't have much choice where you land