More Daz to AI rendering 😉

Home Page Forums Art Showcase More Daz to AI rendering 😉

Viewing 15 posts - 1 through 15 (of 24 total)
  • Author
    Posts
  • #2036374
    Legolas18
    Participant
    Rank: Rank 7

    More Daz to AI rendering 🙂

    Lali's Casual Style Bundle 4.0 (AI)

    Lali's Casual Style Bundle 4.0 (DAZ)

    AI Render:

    Daz Render:

    dForce Lali's Winter Love Outfit for Genesis 8 and 8.1 Females (AI)

    dForce Lali's Winter Love Outfit for Genesis 8 and 8.1 Females (DAZ)

    dForce Lali's Winter Love Outfit for Genesis 8 and 8.1 Females (AI)

    dForce Lali's Winter Love Outfit for Genesis 8 and 8.1 Females (DAZ)

    #2036386
    Frank21
    Participant
    Rank: Rank 5

    Faces and clothing looks so much more realistic. What denoising strength are you using in img2img or do you adjust it for each image?

    #2036387
    Legolas18
    Participant
    Rank: Rank 7

    @frank22 I typically go from 0.45 to 0.65 depending on the image (more complex images require a lower number), but I am also using SDXL (RunDiffusion) which is also pretty good as a model. 🙂

    #2036388
    Frank21
    Participant
    Rank: Rank 5

    Thanks.

    #2036400
    gaver
    Participant
    Rank: Rank 5

    I am still puzzeled.....

    #2036407
    Frank21
    Participant
    Rank: Rank 5

    graver - Putting the DS image into SD ing2ing will "re-render" it preserving much of the original appearance, but SD can "re-do" aspects of the image according to the prompts you use. So in this case prompting a character's face (with a LoRa) can change the face, and prompting "fur" and describing the snowsuit can make it more realistic looking. Compare the fur and snowsuit in the first two images.

    #2036412
    gaver
    Participant
    Rank: Rank 5

    I git that far...been playing around with it. Understand what a lora does etc, it is just not enough knowledge yet. What lora to us, what promts to use, why does the face go wonky... I get the idea of the amount you can apply ( As long as it is a button or slider one can just try and see what happens.
    Then the are al the other things beside loras ( hypernetworks,checkpoints)still have to read up about those.
    IF there is a "basic setup" for just upgrading images I made is DS I would love to hear.
    I understand the power of Ai in generating TextToImage, and for that I still think it is not much more than a funny toy, or at best something to make nice backgrounds/HDRI's to use in DS.
    So, just gonna use it to improve my renders. IF anybody has a walk trough for that, I would be a happy gaver 🙂

    #2036413
    Frank21
    Participant
    Rank: Rank 5

    Jesus fucking wept! Went to start SD and now I get the message below. Would having it installed to C:\Users\Owner\ be a problem rather than just straight to the C drive? Everything worked fine a hour ago.
    Stable Diffusion LOL. It should be called completely fucking unstable diffusion!

    1

    #2036415
    gaver
    Participant
    Rank: Rank 5

    Ah nice.....
    It should work, but sorry sir the helpdesk does not have a solution for that. best advice is to wait for the update...
    IF they make something and put it out in the world...do it friking right! 🙁

    #2036419
    Frank21
    Participant
    Rank: Rank 5

    Did a complete fresh install, updated pip, and still the same message.
    Might have to find a new hobby 😒.

    #2036429
    gaver
    Participant
    Rank: Rank 5

    I know of a prog called DAZ Studio,...give it a try...:D

    #2036434
    Frank21
    Participant
    Rank: Rank 5

    After 6 hours of stuffing around I gave up trying to get SD to run. But not all doom and gloom... I picked up a portable version of Adobe After Effects today after my old one stopped working. As a bonus it exports to .mp4 so I don't have to re-encode massive 30GB .avi files for 4 minutes of video, yay 😊.

    #2036445
    Spats
    Participant
    Rank: Rank 4

    @gaver "I know of a prog called DAZ Studio,...give it a try...:D"

    Laughed Out Loud.


    @frank22
    "Unstable Diffusion" unstably funny. (Not to belittle your install pains. Feel like suffering with you.)

    Get stable you people. And keep at it a don't give up.
    The un-AI-ified world needs your works.
    (I should preach to the self as much.)


    @legolas18
    wow
    Holy crap.

    #2036453
    Legolas18
    Participant
    Rank: Rank 7

    @gaver I am not using any Loras in these images. I use Loras only to get very specific things.

    As for hypernetworks & textual inversions, just ignore them, they are an older tech and tend to be used more in the older gen of SD (1.5).

    Just concentrate on the checkpoints (which are the main models), Loras/Lycoris (essentially the same thing that add some item/hair/clothing/style). Also, learn about 2 essential extensions:

    1) Adetailer (which automatically can enhance faces/eyes/bodies (depending on what you select).
    2) ControlNet (which allows you to pose character but also to recreate certain images).
    3) Wildcards: Allows you to create lists of things so you don't always have to come up for them yourself. For example, if you call a list of hairstyles, the extension just picks a random item from the list 🙂


    @frank22
    Since you just got a new graphics card, you should uninstall the old graphics drivers and then re-install the latest ones. this did fix any problems I had in the past. I hope it works 🙂


    @spats
    I am glad you liked it 🙂

    #2036558
    Paul (Fuzzy3000)
    Participant
    Rank: Rank 5

    @frank22

    Never give in .... Never surrender!

Viewing 15 posts - 1 through 15 (of 24 total)
  • You must be logged in to reply to this topic.

 

Post You Might Like