Loading...
> >

Remixing your art with AI

1…1617192122

Comments

  • ArtiniArtini Posts: 9,513

    Thanks for the comment, Diomede, and for showing your results.

    I have so many interests in 3D graphics and only a certain time to play with it,

    so I choose something to bring nice results quickly.

     

  • FirstBastionFirstBastion Posts: 7,777
    edited February 2

    It is true, we do get results quickly.

    .

    cliffs-conceptart-1stb.jpg
    320 x 237 - 36K
    Post edited by FirstBastion on
  • charlescharles Posts: 848
    edited February 3

    link removed

    original

    https://ibb.co/g9ByCjK

    Not actually embedding the image because the mods apparently don't like it posted here.

     

    Post edited by Richard Haseltine on
  • Richard HaseltineRichard Haseltine Posts: 101,507

    charles said:

    link removed

    original

    https://ibb.co/g9ByCjK

    Not actually embedding the image because the mods apparently don't like it posted here.

    Regardless of whether it is linked or embedded, the visible edge or an areola is not permitted under the ToS.

  • charlescharles Posts: 848

    Richard Haseltine said:

    charles said:

    link removed

    original

    https://ibb.co/g9ByCjK

    Not actually embedding the image because the mods apparently don't like it posted here.

    Regardless of whether it is linked or embedded, the visible edge or an areola is not permitted under the ToS.

    hah, it's not because the areloa wouldn't be at that location of the boob. But I see what you are talking about.

     

  • ArtiniArtini Posts: 9,513
    edited February 6

    The another nice feature to test is image restoration and upscalling.

    After AI treatment:

    Post edited by Artini on
  • ArtiniArtini Posts: 9,513

    The original image I have created in Daz Studio from Face Transfer

    https://www.daz3d.com/forums/discussion/672386/facetransfer-1-and-genesis-8-1#latest

    Enh01.jpg
    1770 x 1766 - 459K
  • ArtiniArtini Posts: 9,513
    edited February 6

    Slightly better AI treatment of fabrics on the clothes.

    Enh04.jpg
    1770 x 1766 - 660K
    Post edited by Artini on
  • ArtiniArtini Posts: 9,513
    edited February 7

    I think AI enhancement works quite nicely for characters generated with Face Transfer 2

    My image before enhancement is in:

    https://www.daz3d.com/forums/discussion/661746/face-transfer-2/p11

    Woman06enh01.jpg
    1500 x 1500 - 629K
    Post edited by Artini on
  • ArtiniArtini Posts: 9,513
    edited February 6

    Another enhancement of the previous image.

    Woman06enh02.jpg
    2048 x 2048 - 1M
    Post edited by Artini on
  • DiomedeDiomede Posts: 15,188

    Wow, very nice enhancement.  

  • DiomedeDiomede Posts: 15,188
    edited February 6

    Here is another quickie using the ancient Poser 4 figures (P4 Dork and P4 Boy) in the ancient Carrara program, combining with primitives and ancient low res objects, and generating a depth pass.  

    I used Carrara's native 'toon' filter to make it easier for the 'Canny' processor in AI to find the edges.  And I used Carrara's native depth pass (inverted) without using the AI processor.

    NOTE - I did not use image2image.  The original image is only used for Controlnet to find edges in Canny and for the depth map.

    Original image.

    Processed with Stable Diffusion text2image with the renders in Controlnet Depth (weight middle) and Canny (weight very low)

    Post edited by Diomede on
  • ArtiniArtini Posts: 9,513

    Thanks for the comments, Diomede.

    Your results are excellent.

    I think, I need to experiment more with AI enhancements.

    The results are encouraging, so far.

     

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,312

    not remixed, just straight AI but all done on my PC

  • ArtiniArtini Posts: 9,513

    Nice, Wendy. I have seen your other video, which I like very much.

     

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,312

    that one I actually used Leonardo AI motion

    the advantage is higher resolution than I can do with SEINE in Visiions of Chaos but of course costs 25 credits and I only get 150 a day free

    my PC just costs electricity but I also get to use a positive and negative prompt too

    Leonardo you never know what the motion will be

  • DiomedeDiomede Posts: 15,188

    Impressive videos!

  • FirstBastionFirstBastion Posts: 7,777

    Artini said:

    Nice, Wendy. I have seen your other video, which I like very much.

     

     Wendy did a great job on this.  Visually very compelling.  Would have gone with a different Narrator's voice though.  This one sounds like a dull deacon. 

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,312
    edited February 15

    it's a text to speech voice generated from a 10 second wav file of someone I know's voice I had

    no, it doesn't sound like them

     

    the Llama generated screenplay

    [The camera pans over a sweeping landscape of rolling hills, with a hint of mist in the distance. The sun casts a warm glow on the scene, illuminating a magical book lying on a pedestal in the center of a grand library.]
    Narrator (in an epic, heroic tone): "In a world beyond our own, where magic reigns supreme... lies a book of untold power. A tome that holds the key to a realm of wonder and awe, where fantastical creatures roam free and magnificent cities rise from the earth like giants."
    [The camera zooms in on the book, as the pages begin to glow with an otherworldly light. The edges of the page turn into golden trimming, and the words themselves seem to come alive, shimmering like diamonds.]
    Narrator: "This book is no ordinary tale. It's a portal to a realm of unimaginable beauty and danger, where mythical beasts soar through the skies, and ancient magic pulses through every stone and tree."
    [The camera cuts to a sweeping shot of a vast green valley, filled with towering spires of mushrooms that seem to stretch up to the clouds. A procession of fantastical creatures marches through the valley, including faeries flitting about like sparks of light, dragons that breathe fire and smoke, and griffins soaring overhead.]
    Narrator: "In this enchanted land of the book, you'll find castles that float on clouds, cities made of crystal and gold, and forests alive with magic. Where giants rule the land, and wizards conjure spells that shake the earth."
    [The camera cuts to a shot of an adventurer, armed with a sword and shield, standing atop a castle turret. The sun sets in the distance, casting a golden glow on the landscape below.]
    Narrator: "But beware, brave traveler... for this realm is not without its dangers. Dark forces lurk in the shadows, waiting to strike. And the book itself holds secrets and mysteries that can only be unlocked by those who are pure of heart and brave of spirit."
    [The camera cuts to a shot of the adventurer, as they embark on a perilous journey through treacherous lands filled with danger at every turn. They encounter fierce battles and narrow escapes, but never lose hope or courage.]
    Narrator: "Will you answer the call of this magical book? Will you embark on a quest to unlock its secrets and save this enchanted realm from destruction?"
    [The camera cuts back to the adventurer, as they stand before the book, their face filled with determination and resolve.]
    Narrator: "The fate of this world rests in your hands. The choice is yours... to unlock the magic, or let it remain lost forever."
    [The camera zooms out one final time, as the adventurer opens the book and steps through the portal, disappearing into the fantastical world beyond.]
    Narrator (in a final, dramatic flourish): "Step into the magic... and discover a world beyond your wildest dreams!"

     

    screenshot, yes generating a poem, on my PC will take a while

    uhm no it stuffed up, I need to uninstall Jan by looks and redownload

    Untitled.png
    1905 x 1044 - 169K
    Post edited by WendyLuvsCatz on
  • ArtiniArtini Posts: 9,513
    edited February 16

    Custom Genesis 9 character created with Face Transfer 2 with AI textures applied to some elements of the scene.

    g9F121.jpg
    920 x 1200 - 541K
    Post edited by Artini on
  • DiomedeDiomede Posts: 15,188

    @Artini.  Love it.  Very creative way to integrate into a workflow.  

  • DiomedeDiomede Posts: 15,188
    edited February 24

    Daz Studio Users, if you are interested in a comparable composition workflow with Daz Studio that I have experimented above with Carrara's multipasses, try Studio's advanced render tab for canvasses.  I recommend use of the depth pass, but I am sure there is much more to explore. The AI probably has an algorythm to try to detect depth from your still image, but you can have a lot more control and it might be faster if you use your own depth pass instead of preprocessing in the AI program.  In Stable Diffusion, have to remember to set preprocessing to none in the Controlnet setting in which you upload the depth map.  For canvasses more generally, this video explains how to get something useful from Daz Studio's exr file generated by the depth canvas.

    EDIT: Note that you will typically want to invert your custom depth map for use in Controlnet.

    Post edited by Diomede on
  • ArtiniArtini Posts: 9,513

    Thanks a lot, Diomede.

    Great tips on canvasses - worth to explore more of them.

     

  • DiomedeDiomede Posts: 15,188

    A 3D to AI example of a fantasy art image.

  • DiomedeDiomede Posts: 15,188
    edited March 7

    A 3D to AI landscape illustration

    Post edited by Diomede on
  • DiomedeDiomede Posts: 15,188
    edited March 7

    One more example of a final image and the 3D renders used.  This is an attempt at a vintage pulp cover was made using 'text2image,' not 'image2image.'  The 3D renders were used in Controlnet (a) 'canny' to detect outlines (the toon filtered render) and (b) depth (inverted the depth canvas/multipass).  It can be problematic to rely on text prompts to direct the composition of multiple people.  In this case, I did not use the main render (beauty canvas) for anything so I am not including it.

    Pulp Castaways - "I told you to check the map for shoals."

    Here is a side by side of the final image with the 3D renders used.  I am using Stable Diffusion with Controlnet while I am learning.  

    Pulp Castaways.png
    1024 x 1536 - 2M
    Side by Side Pulp Castaways.png
    1024 x 768 - 719K
    Post edited by Diomede on
  • ArtiniArtini Posts: 9,513

    Interesting compositions, Diomede. So many possibilities...

     

  • 3Diva3Diva Posts: 11,630
    edited March 8

    A subtle enhancement with Victoria 9. Before and after:

     

     

    I think Daz Studio was already able to get it pretty close to "realism", the slight push with Stable Diffusion turned out cool, IMO.

    2024-03-08_15-31-02_7292.jpg
    1800 x 2160 - 622K
    2024-03-08_15-33-49_3189.jpg
    896 x 1080 - 165K
    Post edited by 3Diva on
  • FirstBastionFirstBastion Posts: 7,777

    .I'm seeing a young Lindsay Lohan.

     

  • FirstBastionFirstBastion Posts: 7,777

    Diomede said:

    One more example of a final image and the 3D renders used.  This is an attempt at a vintage pulp cover was made using 'text2image,' not 'image2image.'  The 3D renders were used in Controlnet (a) 'canny' to detect outlines (the toon filtered render) and (b) depth (inverted the depth canvas/multipass).  It can be problematic to rely on text prompts to direct the composition of multiple people.  In this case, I did not use the main render (beauty canvas) for anything so I am not including it.

    Pulp Castaways - "I told you to check the map for shoals."

     

    Here is a side by side of the final image with the 3D renders used.  I am using Stable Diffusion with Controlnet while I am learning.  

     Showing the steps in the process is helpful. 

Sign In or Register to comment.