Adding to Cart…
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.You currently have no notifications.
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
I tried an experiment. Generated in DAZ Ai and then sent to Kling for animation.


Some Star Wars stuff.
What I do find odd is that there are so many words that it times out on. I found that you can't use the word "voluptuous" but "curvy" is okay.
Nice! I keep meaning to try Kling and Runway to do some animations. Imagine how much work it would be to do that in DS animation, LOL
Try using FLUX at Krea, it is even more restrictive.
Unusual tree...
A suitable level of feline contempt being shown in the image below. How does the AI know..?
Regards
Richard
I like, how we can create different type of statues.
Using a modified prompt from the other thread, I got:
Dangerous Ride
Somebody has a degree of courage... Prompt 'Seated Woman flying through futuristic skyscrapers with similar flying vehicles.'
And, wee, space battles:
Regards,
Richard
Love the idea of taking your office chair to work.
Sometimes the dynamism of the AI produced images can be a surprise. These are from prompts that are variations of 'Futuristic armoured soldier running with incoming rockets and outgoing missiles'. I tried quite a few variations and can't remember the exact ones used for these two images.
As it happens I've been re-reading David Drake's 'Redliners' and the initial chapter or two could be illustrated with images like these.
Regards,
Richard
The mid stride run, is particularly dynamic and airborne.
When there are no hands to get wrong, the result is quite good:
You can imagine the overwhelming roar and almost feel the radiant heat on your face as the surface of the runway ablates.
Regards,
Richard
Things not to do at a beach: Hit Dragonling with flaming branch when protective parents are around.
Not quite sure how that came out of the tropical beach & dragon breathing fire prompt I gave.
Regards,
Richard
I think Stable Diffusion's user base have skewed the training towards bikinis and beaches. Its sort of inevitable given the male demographic of users.
I have been trying to get an AI image of a Flying Fortress. The fortress in question that I want to fly is Beaumaris Castle in Wales, a mediaeval fortress. I want to have it at an altitude of 10,000ft, flying in the way bricks usually don't. I don't seem to have the imagination to find the correct term to trigger the AI into doing it. The image below is the best I can get, with the castle firmly rooted to the ground. Can anyone else do better?
This is the type of Flying Fortress it wants to show:
And while it's mad, it's nowhere near the order of madness I want to portray.
OK, another try 'FLYING FORTIFICATION' . This feels as if the cliff is very high
And 'Dont fly through clouds, they're full of rocks' from this 'FORTIFICATION FLYING THROUGH THE SKY'
Still not there, though, is it?
All I want is of a stone mediaeval fortress flying through the air as if it's meant to be there. Not that hard, is it?
Regards,
Richard
I did a few tries, but it wouldn't leave the ground. Apparently the AI has never seen a castle without ground, so it "thinks" the ground is part of the castle.
prompt:
For research purposes.
Source image from wikipedia:
Castell_Beaumaris_Castle,_Ynys_Mon_(Anglesey),_Wales_45.png (3840×2160)
This source image was used as input in flux-1-canny-dev
You can find the workflow as template in ComfyUI - just download the models needed from there...
Prompt:
Output image:
Now you could try to repeat that process with the output image as new source image...
Or you could take the output to photoshop or another GenerativeAi software to selectively edit the background further...
One challenge may be that the tag "flying fortress" is already used by something else...
I would like to live in that flying castle though:
And if you add some "spaceship engines" it may actually reach warp speed...
The Boeing B-17 Flying Fortress Bomber has an 80-90 year history. Hence why it prompts the planes.
True. But Mediaeval fortresses have a considerably longer history.
It is a skill getting ones head around the possible word soup necessary to prompt the type of image you desire.
This prompt: "flying castle,view from above, planet orbit, star, mediaeval"
Then: "flying castle,view from above, planet orbit, star, science fiction"
Finally "flying castle,view from above, planet orbit, stars and moons with space ships, science fiction"
Still not a mediaeval castle made from shaped stones with crenalations etc flying high, but closer
Regards,
Richard
try floating in the sky
Maybe, just maybe:
Prompt: "flying castle made from stones with crenalations,view from above, planet orbit, stars and moons with space ships, science fiction"
Prompt: "flying castle made from stones with crenalations and turrets and flage,view from above, planet orbit, stars and moons with space ships, science fiction"
Well. That was more difficult than I expected, I must admit. Shows I need to exercise a particular type of imagination to make it easier.
Regards,
Richard
my result
You seem to have got there Wendy.
Regards,
Richard
These are wonderful! Here are some good ones I did back in May. I lost some files on a different hard drive after formatting C:, so I don't have the prompts any more. So sad your partner got greedy and stopped providing the AI. Could you consider putting an AI on your local machine and training it on your data? There are plenty on GitHub and I've downloaded several. AI tells me how and writes code, but my higher-end machine isn't enough to run an image generator yet, though there is a version of Stable Diffusion that connects online with daily tokens. I've been having AI write code for me because I never learned programming except Access 2003, and I've been away from that for quite a few years. It writes Python or Javascript and I test. So far I've had good results with trying to embed many science fiction story metadata in ChromaDB, a vector database. I started with AnythingLLM which requires no code and is great for beginners. Now, I'm trying to learn LangChain. Fun stuff!
Anyone know of an AI project like DAZ that gets permission to train on its artwork? Right now I'm in GIMP massively altering some cartoons I generated with CoPilot in my drawing style, so I won't infringe on the poor artist's rights before I make my video for YouTube.
Meanwhile, I found these in Amazon Photos so the quality isn't as good as original. Hope you enjoy them. BTW, you can't copyright AI-generated art if you don't alter it, so have fun with these. They are public domain [US Copyright Law]. Enjoy!
Series 1 I tried to get the girl pulling a full hand truck up the stairs but got these. The picture I Frankensteined together in GIMP from old photos and the file boxes in the last one is on this blog post https://www.tumblr.com/danniswrites/775584343217373184/deadly-job-136-file-boxes-by-myself?source=share I do love the stair pictures I got! And Victoria is awesome, if only the AI knew what a hand truck looks like. It's hard to find pictures on the Web, too, because they come in many different kinds. Sorry to send you to my blog but I can't find the final photo.
Series 2 is Victoria and Michael in a train station. These came out great!