HBalckl posted at 1:05 AM Mon, 12 February 2024 - #4481493

Some examples from my 3d to AI work:

Mosaik - Runkelserie - Suleika

https://www.artstation.com/artwork/XgLPN0

Mosaik - Runkelserie - Adelaide von Möhrenfeld

https://www.artstation.com/artwork/Ny0KDd

Mosaik - Runkelserie - Runkels Minnesang Szene animation

https://www.youtube.com/watch?v=X0LhpjrF_cY

https://www.artstation.com/artwork/6N0oXO

Mosaik - Runkel - Hannes Hegen - 3d Reconstruction

https://www.artstation.com/artwork/NyJJE5

Myrmidon Achilleus 1190BC

https://www.artstation.com/artwork/WBJ51v

Trojan Hector 1190BC

https://www.artstation.com/artwork/lDO42a

Gaius Julius Caesar - Pharsalos 48BC

https://www.artstation.com/artwork/1xBqXo

Cleopatra VII Filopator - Queen of Egypt

https://www.artstation.com/artwork/rJ0n6O

WW2 Generals

https://www.artstation.com/artwork/Alv4Kz

https://www.artstation.com/artwork/04Ol0G

Thanks

OK thanks

what AI ? How did you produce the images ? Just trying to get up to speed :)

FVerbaas posted at 11:01 AM Sat, 27 January 2024 - #4480815

Nevertrumper posted at 4:07 AM Sat, 27 January 2024 - #4480808

AI still has difficulties with hands. The model on the right has a left hand on her right arm. At least all hands are messed up.

Yes. Hands and feet.

Sometimes 3 legs. 

Artificial Ignorance 

Indeed, but still extremely useful as I've said. If you turn down the AI strength you are less likely to get an extra leg. We are of course talking about a program that is in beta. Who knows where it will be in 6 or 12 months from now. AI development is lightening fast. 

Lotharen posted at 9:01 PM Fri, 26 January 2024 - #4480805

serene posted at 3:38 PM Mon, 22 January 2024 - #4480677

Crikey blimey. Enhancement indeed. One of my Poser 12 test renders:

What I love about this one is how it looks like it was painted in Clip Studio or Photoshop. If I could get it to just do that for my images I would be happy. Never could figure out how to replicate that in any art program.

My guess is that this was done at a low AI setting. With the app you can change the image gradually by upping the level of AI. There aren't too many AI quirks with the lower settings, but the pictures you see are with AI at 50% or greater to get that wow factor. Most people are picking photo real output, but you could generate digital art as that is an option. I have a ten year old computer, all the rendering is done their end so its all very simple. Why don't people go on and try it ? 

Shadow^Mist posted at 6:22 PM Tue, 23 January 2024 - #4480719

Qaz, I don't mind at all. Please go ahead and post on-going improvements. As a side note, I found that getting a single figure to look like the original is much easier than getting a scene of several figures with a background to look like the figures. I'd appreciate any insight you discover re that. 

OK so from working with this for a while, it is clear that the AI works great to an extent but it has limitations. Its not good with hands. I'm told that you can overcome this with in depth negative tags, but not found that yet. If there are small intricate things in the picture such as faces the AI struggles. I think since this part of the program is trying to enhance what is already there it struggles when it has low level of pixels to work with. Additionally the data set it has is going to consist mainly of people looking forward towards the camera, and struggles with unusual angles. A scene where stuff is happening is going to be hit and miss. 

So this is a render I did 10 years ago in firefly and I was happy with it at the time, but the Superfly versions would have been better. I did this for Halloween.

And this is what I eventually came up with:

Now it was originally messed up. So basically I cheated. I thought what if you cut up the individuals into separate pictures and enhanced them. Much better. As you have seen large pictures of faces come out pretty well. I had problems initially with the Megan Fox like vampire as she was on her side, so I thought what if I turned her 90 degrees so she is upright and pressed enhance. Yep that worked. Cut and paste them into the same picture. Make a few other corrections such as two of the hands and voila ! And to think this AI is just getting better and better.

Awesome isn't it. And that's the default. Hope you don't mind I used your picture upped the AI and added additional description to try to get it closer to the original.

Another problem with enhancer is that it tries to enhance everything. Out of focus things in the background it will try to 'focus' those.

Those soldiers were really fuzzy ! When you create with Poser you would only export the figures you want enhanced. If you are dealing with old pictures you can use another AI tool like Clipdrop to remove the background.

serene posted at 3:38 PM Mon, 22 January 2024 - #4480677

Crikey blimey. Enhancement indeed. One of my Poser 12 test renders:

I have been having great fun with this, with the only problem being only limited to 50 renders a day ! What I have found is that you have to add additional description if the AI doesn't get parts of the image right. For instance in the above, did you want the eyes green ? If not, in the description say "her eyes are light brown green". I pick photo-realistic as the image type. If you want the AI to do its own thing increase AI strength and lower resemblance strength. To get closer to the original lower AI to 25% and up the resemblance to max.

Razor42 posted at 7:31 PM Wed, 17 January 2024 - #4480489

These look great, typical AI quirkiness aside.

To me the process seems not so far from the origins of Poser tbh. 

"Born from Larry Weinberg's desire to be an illustrator, Poser was created to be a better mannequin for artists to use as a reference.

Weinberg started with a small wooden mannequin from an art store, but it wasn't quite right for what he wanted to do. Then he asked himself a question that would change the course of computer graphics and digital art: Why not make a digital mannequin?

So, he turned the experience and knowledge he had gained at his day job as a programmer and animator in Hollywood to build the first version of Poser.

The first version was released in 1995 as a simple visualization tool for artists to quickly and easily create poses (hence the name Poser) to use as a human simulation."

Excerpt https://www.posersoftware.com/article/419/a-brief-history-of-poser-with-larry-weinberg

Yes getting back to what Poser was all about. Can't tell you how much fun I'm having !

hornet3d posted at 2:04 PM Wed, 17 January 2024 - #4480470

qaz posted at 11:42 AM Wed, 17 January 2024 - #4480467

But where does the art begin ? I go to a great deal of trouble to create personality morphs, but I didn't create the figure, just manipulated what the computer generated. The great thing about this software is it doesn't invent a scene based on some text input. You invent the scene in Poser. In the same way a photographer  produces art by framing a shot and choosing the right light conditions, we are doing the same in Poser. All this is doing is taking what you have created and enhancing it a bit.

This is a scene I created in Poser  a few years back

Again impressive and it may be nit picking but she is now no longer holding the wand as before and the thumb seems to be doing contortions not possible by your average human.

Yes, AI and hands, what can I tell you. Of course because it is in exactly the same position as the Poser model, it would only take a couple of minutes to cut and paste them so sorted. Its an enhancement not a replacement. The program is still Beta and it undoubtedly will keep improving.

But where does the art begin ? I go to a great deal of trouble to create personality morphs, but I didn't create the figure, just manipulated what the computer generated. The great thing about this software is it doesn't invent a scene based on some text input. You invent the scene in Poser. In the same way a photographer  produces art by framing a shot and choosing the right light conditions, we are doing the same in Poser. All this is doing is taking what you have created and enhancing it a bit.

This is a scene I created in Poser  a few years back

Now I played around a bit. Double enhanced, and a bit of photoshop. The rays from the wand came from the AI influence.

Low Res to High Res

The program I used is still beta and it does have a few issues. The great thing here is that you set the scene in Poser. You create the poses. The AI puts everything exactly where you put it, so its more picture enhancement than an AI imagination based on text prompts. As you say it probably doesn't need advanced render to work on. Doesn't even need decent clothes, just the general idea. The first picture is 11 years old and the outfit was a freebie. It was set at the default 50% AI. I had a go at 100% and it improved the figure and suit - better buckle !

Here's another one.

I was quite happy with this render I did a few years ago:

Ah that's good says the AI, but let's see what I can do ! Cue the magic

OMG ! What the F---- !  I think I've started to dribble. Look at the neck, the hair, the face. Now it has made a mess of the golden Lasso and the sword handle. I could try to fix in description, but it's a quick job in Photoshop.

Now the AI does change the image with descriptive text. This is the descriptive text for the above:

"A 3D rendering of a woman in a blue and red superhero costume with a cape, yellow belt, and blue shorts. She has blue eyes, blonde hair, and is wearing a blue top with a red \"S\" on her chest. She appears to be flying and has her arms outstretched."

The AI came up with that description. Genius. I couldn't describe it that well. You use Poser for - yes you guessed it POSING figures and the AI says, are you happy with those skin textures, those clothes, let me see what I can do. 

What would be good is if you could create a scene in Poser and then use Stable Diffusion to make it look more realistic / atmospheric. You would obviously not need a lot of assets as the AI will invent those. AI is only just beginning, so it will be interesting how it develops in the next two years.