Algovincian 2024: New Year - New Junk
Haven't been posting much here lately, so I figured it was time for a new thread for the new year and upload a couple of recent images in an effort to start posting again.
Arthur bumped into this cute little voodoo princess on a recent ride down to the bayou:
The image makes use of several recent releases from Strangefate and a doll from Thorne:
https://www.daz3d.com/stf-tribal-armor-for-genesis-9-and-genesis-8-female
https://www.daz3d.com/stf-tribal-witch-doctor-undies-and-body-paint-for-genesis-9-and-genesis-8-female
https://www.daz3d.com/stf-kadis--legendary-armor-for-genesis-9-and-genesis-8-female
https://www.daz3d.com/akubisa-for-genesis-8-female
I was so happy G8 was still supported in Strangefate's recent releases - woot! Thank you.
Arthur had been sleeping rough for several days before finally making it back to camp, and probably should have grabbed a bath & shave before having this picture taken:
I love exploring and being an in-game photographer, which is where the background in the first image came from. The second image is also an in-game screencap processed to be more photorealistic using AI, and then one of my standard post-processing scripts for pShop to make it look like an aged photograph.
- Greg
Comments
Looks great. I especially like black & white version.
Thanks @Artini. It's so cool how just a screen capture of actual game play (the color image of Arthur) can be used to create a more photo-realistic black & white image.
Here's another example - an image of a busy city street market in Saint Denis circa 1900 using the same workflow:
- Greg
It is also great. Looks like a black and white photo.
I probably should have been more clear about the fact that the last image of Saint Denis also started as a screen capture during gameplay.
Of course, in-game screencaps aren't the only source of input images that can be used. DS renders also work great as input to the processing, which is why I'm posting here . . . duh! lol
Here's 2 more examples (Streets of Venice and Wild West Town) of environments rendered in DS and processed using the same workflow. The workflow is still a work in progress and a few tweaks were made since the Saint Denis image, but the basic idea remains the same:
https://www.daz3d.com/the-streets-of-venice
https://www.daz3d.com/wild-west-town
- Greg
Love the new images, Greg! The realism with the black and white images are incredible! Looks like historic photographs! SO COOL!
Thanks, @3Diva. The workflow is really fun to play around with. The AI didn't do so good with the writing on the signs in the old west town, though . . . doh!
- Greg
Yeah, writing and hands are my two biggest frustration with AI created content. But I do know it's only a mater of time before the models improve to the point where we no longer have a problem with writing and hands.
Wanted to upload some subtle expression tests for a workflow I've been playing with to check them out on different devices. Not perfect, but not bad considering each of the 4 variations for each character were generated from the same initial image:
- Greg
ETA: Here's a quick video making it easier to see the change in expression for each character:
This traditional-looking manga style you've been able to achieve is awesome! It looks fantastic! Like Bill & Ted would say, Most Excellent!
Good stuff, but you might want to spellcheck the title of the thread.
Doh! Thanks, @Gordig.
Thank you, @3Diva. This workflow is better on tight face shots and not so good for full head-to-toe shots or groups. The previous algos were the exact oposite - better on wider framed shots and no good on close-ups. I can't win! lol
- Greg