Loading...
> >

Daz AI Studio - Image Share

24

Comments

  • MelanieLMelanieL Posts: 7,424

    FirstBastion said:

    MelanieL said:

    I guess it can't count too well - my "Couple" turned into a love triangle!

     It does offer some rather complicated story possibilities,  tension,  and  dare we say,  paths to explore.

    Tee-Hee! laugh 

  • ArtiniArtini Posts: 9,513

    Great images, everyone.

    I created a separated thread to show my approach with Daz AI.

    https://www.daz3d.com/forums/discussion/681031/artini-s-aiteracy#latest

    I like results very much so far.

     

  • csaacsaa Posts: 824

    Ivy, 

    Delightful surprises ... Thanks for sharing.

    Cheers! 

    Ivy said:

    washington crossing the potomac river

    Sexy Male Underwear Model (Whoa Thats a man?)

  • EthinEthin Posts: 1,143
    edited April 5

    dazai_846bb5ac-f25c-11ee-8f4f-8de68b29fa88.png
    1024 x 1024 - 1M
    Post edited by Ethin on
  • richardandtracyrichardandtracy Posts: 5,741

    Err.

    What's the probability of the AI learning off your images on the DAZ forum/Gallery Ethin? Almost looks too similar to be chance.

    Regards,

    Richard

  • 31415926543141592654 Posts: 975
    edited April 5

    I had not yet tried to use AI generated images. I am not opposed to it as an art medium, but not excited about it either. Since it is here (and free to try) I have done a few images. I was surprised with how good the first few images did ... then I tried this: image prompt 'male photographer in a studio with a large format camera' ... similar to odd numbers of fingers, I am not sure how this bipod is supposed to work.

    EDIT: after a few more image tests, the system seems wildly hit or miss ... the hits are really good and the misses are ridiculous.

    dazai_272d1572-f327-11ee-9b2c-7d4410e2f198.png
    1024 x 1024 - 1M
    Post edited by 3141592654 on
  • GordigGordig Posts: 10,107

    Love that non-Euclidean bipod.

  • Richard HaseltineRichard Haseltine Posts: 101,507

    richardandtracy said:

    Err.

    What's the probability of the AI learning off your images on the DAZ forum/Gallery Ethin? Almost looks too similar to be chance.

    Regards,

    Richard

    The Daz add-on is trained only on Daz-owned images, not user or (currently) PA images.

  • HavosHavos Posts: 5,376
    edited April 5

    I was impressed with Ivy's "washington crossing the potomac river" image, as it is rare that you see AI images with multiple figures in them. As any kind of many people images are often the most time consuming to create in DS, I tried to see if I could get something like that using the DAZ AI. First I tried: 

    "battle of helms deep" and I got this image:

    Not what I had in mind. Yes, there is a castle, but there is no battle. Also given the holes in the walls, it is not a castle I would want to defend.

    Then I tried: "charging cavalry squadron" and I got this image:

    I was hoping for a line of charging horse and riders, and I got some tanks crossing a river, a tank squadron I guess.

     

     

    dazai_f2f675c8-f34e-11ee-8f52-8de68b29fa88.png
    1024 x 1024 - 2M
    dazai_376e4aa0-f34f-11ee-9b2c-7d4410e2f198.png
    1024 x 1024 - 2M
    Post edited by Havos on
  • junkjunk Posts: 1,341

    Stezza said:

    nicely dressed lady using Carrara on a hotted up computer in the kitchen laugh

    OUCH!  I guess she's going to be there for a long while. 

  • junkjunk Posts: 1,341
    edited April 5

    Richard Haseltine said:

    richardandtracy said:

    Err.

    What's the probability of the AI learning off your images on the DAZ forum/Gallery Ethin? Almost looks too similar to be chance.

    Regards,

    Richard

    The Daz add-on is trained only on Daz-owned images, not user or (currently) PA images.

    That's good to know. 

    I wonder if the images I see in the overwhelming ads could also be thrown into the training.  For example I see an ad showing six different women taking up approximately 20% of my screen.  They're pretty and since the advertisement is on the Daz website perhaps they could be added into the mix.

    Post edited by junk on
  • FirstBastionFirstBastion Posts: 7,777
    edited April 5

    cyberlighting

    1st-cyberjack8.jpg
    1024 x 799 - 110K
    Post edited by FirstBastion on
  • ArtiniArtini Posts: 9,513
    edited April 5

    I have tried a similar prompt:

    nicely dressed lady using Bryce program on a hotted up computer in the kitchen

    Lady02.jpg
    1024 x 1024 - 372K
    Post edited by Artini on
  • ArtiniArtini Posts: 9,513
    edited April 5

    Another similar prompt from this thread:

    female photographer in a studio with a small format camera

    Photographer01.jpg
    1024 x 1024 - 323K
    Post edited by Artini on
  • 31415926543141592654 Posts: 975
    edited April 5

    Artini said:

    Another similar prompt from this thread:

    female photographer in a studio with a small format camera

    Well ... as strange as the camera looks .. it is still better than the ones I got in my attempt ... maybe we should try a photographer out in the field ... 

     

    Post edited by 3141592654 on
  • 31415926543141592654 Posts: 975

    Oh ... an photographer on the moon ... what is he taking a picture of ? ? ? ? ? 

     

    dazai_803773d0-f375-11ee-8f52-8de68b29fa88.png
    1024 x 1024 - 2M
  • ArtiniArtini Posts: 9,513

    Yes, I was also surprised by the construction of the camera.

     

  • ArtiniArtini Posts: 9,513
    edited April 5

    I have tried:

    a female photographer out in the field with a bunch of different cameras

    Photographer02.jpg
    1024 x 1024 - 438K
    Post edited by Artini on
  • 31415926543141592654 Posts: 975
    edited April 5

    @ Artini ... yours is getting better.  I tried to get a trail camera taking a picture of bigfoot ... I cannot get a camera in the image ... but I would give space to bigfoot in this situation also.

     

    dazai_e51b4c98-f377-11ee-8f52-8de68b29fa88.png
    1024 x 1024 - 2M
    Post edited by 3141592654 on
  • ArtiniArtini Posts: 9,513

    Yes, bigfoot is still bigger than a photographer.

     

  • ArtiniArtini Posts: 9,513
    edited April 5

    I have tried:

    photographer room full of cameras

    to see what kind of cameras AI will create.

    PhotographerRoom01.jpg
    1024 x 1024 - 499K
    Post edited by Artini on
  • ArtiniArtini Posts: 9,513
    edited April 5

    Did not want to scare a photographer, so have tried:

    a lady photographer taking a photo with a modern camera of small bigfoot on the ground out in the woods

    but the bigfoot looks a bit strange.

    Bigfoot03.jpg
    1024 x 1024 - 440K
    Post edited by Artini on
  • DiomedeDiomede Posts: 15,188
    edited April 5

    Havos said:

    I was impressed with Ivy's "washington crossing the potomac river" image, as it is rare that you see AI images with multiple figures in them. As any kind of many people images are often the most time consuming to create in DS, I tried to see if I could get something like that using the DAZ AI. First I tried: 

    "battle of helms deep" and I got this image:

     

    Not what I had in mind. Yes, there is a castle, but there is no battle. Also given the holes in the walls, it is not a castle I would want to defend.

    Then I tried: "charging cavalry squadron" and I got this image:

     

    I was hoping for a line of charging horse and riders, and I got some tanks crossing a river, a tank squadron I guess.

    Another attempt at multifigure... - seem to be short a few mummies, and the moon is in wrong phase.  But, there is a campsite, and there is a moon.

    "an explorer defending a desert campsite from an attack by ten mummies at night under a sickle moon"

     

    dazai_76c6fc62-f384-11ee-8f52-8de68b29fa88.png
    1024 x 1024 - 1M
    Post edited by Diomede on
  • EthinEthin Posts: 1,143
    edited April 5

    Richard Haseltine said:

    richardandtracy said:

    Err.

    What's the probability of the AI learning off your images on the DAZ forum/Gallery Ethin? Almost looks too similar to be chance.

    Regards,

    Richard

    The Daz add-on is trained only on Daz-owned images, not user or (currently) PA images.

     

    Even if it was I wouldn't get upset about it, I'd have to get mad at myself 'cause our brains do the same thing. Input data, rearrange it, spit it back out. Inspiration, training, same thing.

    The similarities are due to the fact that we're both fallowing the "rules of cute" and "using" Daz assets.

    I'm definitely drawing "inspiration" from the images I had it generate.

     

    dazai_d72bef00-f266-11ee-9b2a-7d4410e2f198.png
    1024 x 1024 - 1M
    Post edited by Ethin on
  • ArtiniArtini Posts: 9,513
    edited April 5

    I got a camera and bigfoot right, but...

    a lady photographer taking a photo with a modern camera of bigfoot out in the woods

    Bigfoot04.jpg
    1024 x 1024 - 473K
    Post edited by Artini on
  • ArtiniArtini Posts: 9,513
    edited April 5

    At least she points a camera in right direction.

    a lady photographer taking a photo with a modern camera of bigfoot in front of her out in the woods

    Bigfoot05.jpg
    1024 x 1024 - 470K
    Post edited by Artini on
  • 31415926543141592654 Posts: 975

    Ooooh ... this may be one of the best yet. I simply asked for a 'steampunk camera' and was not disappointed.

     

    dazai_27aeecb8-f387-11ee-8f52-8de68b29fa88.png
    1024 x 1024 - 1M
  • ArtiniArtini Posts: 9,513

    Yes, great camera. Well done.

     

  • ArtiniArtini Posts: 9,513
    edited April 5

    My attempt at multifigure...

    bigfoot family climbing a tree out in the woods at sunshine

    Bigfoot06.jpg
    1024 x 1024 - 547K
    Post edited by Artini on
  • 31415926543141592654 Posts: 975

    I am not sure that this is a cryptozoology thread ... but this Loch Ness Monster is a little suspect:

     

    dazai_27a2cf7e-f38b-11ee-9b2e-7d4410e2f198.png
    1024 x 1024 - 2M
Sign In or Register to comment.