Loading...
> >

Remixing your art with AI

1…5681011…21

Comments

  • johnjohn808johnjohn808 Posts: 159
    edited January 2023

    Worlds_Edge said:

    johnjohn808 said:

    It aint happening with me. Ever.

    I get paid to make art in my style, and that style has to be consistant.

    If I go the AI route, then it`s no longer my style. 

    It would be the style of an algorythm. This would be insulting to me.

    I could not call it my own, and my followers/clients would know the differenece.

    What if you had to duplicate it?

    Just my take on it.

    You could train it on only your art/art style, then see if it helps speed up your process or is useful as a reference piece, etc... That would be an interesting experiment. 

    You`re right. I could train it, but I still feel that it still wouldn`t be my art. Plus I dont have the time to train even my dogs LOL

    I dont use references at all.

    To be honest, after looking a examples all over the web, nothing has impressed me. (kind of like G9 LOL)

    And I always have new ideas which I get done pretty quick im told. 

    AI might be for some, but not for me.

    But no knock on those who use or will use it in the future.

    Post edited by johnjohn808 on
  • ArtiniArtini Posts: 9,331

    WendyLuvsCatz said:

    Artini said:

    Thanks for the explanation, Wendy.

    Hope to try it at some point. It seems unavoidable...

     

    more or less my reasoning

    I don't do commercial work, 3D is just a hobby for me...

    It is the same for me.

    3D and the other imaging is just a hobby for me, as well.

    It keeps my brain busy and it is just the thing I like to do in my spare time after work.

     

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,056

    another DAZ render given the ai treatment (was Bijou in Eowyn dress)

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,056

    a video

  • GoggerGogger Posts: 2,375

    WendyLuvsCatz said:

    another DAZ render given the ai treatment (was Bijou in Eowyn dress)

    I like this! Something about the old-tyme animation jumpiness, the arm movement, and of course the flowing fabric. Nice!
  • WendyLuvsCatzWendyLuvsCatz Posts: 38,056

    Gogger said:

    WendyLuvsCatz said:

    another DAZ render given the ai treatment (was Bijou in Eowyn dress)

    I like this! Something about the old-tyme animation jumpiness, the arm movement, and of course the flowing fabric. Nice!

    well it is dforce

    (original on sides in video above ) 

  • GordigGordig Posts: 9,903

    WendyLuvsCatz said:

    another DAZ render given the ai treatment (was Bijou in Eowyn dress)

  • wsterdanwsterdan Posts: 2,340

    wolf359 said:

    jdavison67 said:

    Ai messes up fine details like zippers, buttons, and fingers on hands....so it will be a while until it really is a threat.

     

    JD
     

     

    A vey short while IMHO.
    My interest is how animation will be assisted by AI

    This Video is entirely created by AI, from the script to the talking head and all of the other images.

    Thanks very much for posting your samples, very, very impressive.

    How hard would it be to have it do the same video, but only using 2D art and animation? I would think it would just mean adding "2D" as part of the request(s), but I haven't tinkered the AI artwork generators at all yet.

    -- Walt Sterdan 

  • wolf359wolf359 Posts: 3,821

    @wsterdan

    For the speaking character you need a fairly “Photorealistic ,front facing headshot, with nothing obscuring the face.

    You get a two week free trial, mine has expired

    https://www.d-id.com/creative-reality-studio/

  • wsterdanwsterdan Posts: 2,340

    wolf359 said:

    @wsterdan

    For the speaking character you need a fairly “Photorealistic ,front facing headshot, with nothing obscuring the face.

    You get a two week free trial, mine has expired

    https://www.d-id.com/creative-reality-studio/

    Thanks! very much for the link! I'll give it a shot once I finish my current animation.

    I'm still super-impressed by what you were able to do with it. LIke all the examples of AI creative work that I've seen so far, images or text, a great part of successful execution still depends on the artist using the tools, and I don't suspect that will change too soon. Like any computer art tools I've used since the 80s, I suspect AI will allow bad artists to produce bad art faster and more efficiently. So too, mediocre artists like me. wink

    -- Walt Sterdan 

  • generalgameplayinggeneralgameplaying Posts: 517
    edited January 2023

    wsterdan said:

    I suspect AI will allow bad artists to produce bad art faster and more efficiently.

    -- Walt Sterdan 

    With scaling to millions of users instantly, this already is a big problem. But it'll get much worse, unless regulated in a sustainable way.

    E.g. the "bad ai-traininer" also is a problem, unless you give them absolution a priori, in which case you get bad market, bad opportunities for jobs, bad artists moods, bad training data... it doesn't end there. Further they needn't be bad, stupid or careless would be enough for millions over millions of lookalike pictures destroying prospects of anything that reaches any attention at all. Worse than just worse, following that will be arbitrary rules enforced by platforms on their own behalf. Especially when you train on the artists pictures without their consent, they will be hunted by the new user crowd forever, without any chances to escape, but to change the field of application, or to cease to publish (...).There will not be relief without actively shaping the rules, resulting in a sustainable construction. Or you walk blindly into it, over bodies then.

    (Should a sustainable construction not be possible on the long run, even with explicit opt-in-only for images, you will still have the question of how to shape the transition.)

    (Not mentioning the cloud service scenario with filtering, data processing of prompt, results and whatever similar they find on the net, in addition to what they find about people on the net, censorship, shaping what most/many people see or think is creative. In the current uncertain form, it is not just a tool. I'd call it a plain threat to society, as unimportant as artwork may be in general / to some people. Don't get me wrong, many things are both threat and opportunity in the raw form, but it's not as simple as "denying the bad guy access" or similar.)

    Post edited by generalgameplaying on
  • WendyLuvsCatzWendyLuvsCatz Posts: 38,056
    edited January 2023

    hasn't stopped anyone in the past flooding the internet with bad photography, 3D renders, (click load expressionless) traced screenshots

    Post edited by WendyLuvsCatz on
  • generalgameplayinggeneralgameplaying Posts: 517
    edited January 2023

    WendyLuvsCatz said:

    hasn't stopped anyone in the past flooding the internet with bad photography, 3D renders, (click load expressionless) traced screenshots

    You do see the difference?

    1. Previously we had some kind of copyright. So taking down and filtering works to some extent.

    2. Now imitating anything will be very simple, especially if you train on the to-imitate artist's images.

    3. It's not only easier, it works on a buttonpress for everyone, and in addition scales to millions of users within no time.

    4. Competition wants to monetize their machines, while artists rather get demonetized.

    5. (Ruinous competition, for whose data you need...)

     

    This for rethinking approaches of "fairness", in particular. Sustainability would also be a keyword here.

    (For a fun scenario: to make an artists "impossible" you previously needed at least someone savvy with anything like a post-processing application, and that someone might want to get paid for it and could later remember whom they worked for, or at least some details about it, and they could still fail at it. But now you just need to use the crowd, someone will engineer enough "fun stuff", based on "inspiration". Making it "fair use" will certainly have strong effects within short term.)

    (With cloud services all sorts of "irrigations", censorship and data processing become less transparent, but more telling about people and what they do on the internet. So the crowd gives itself away as well.)

    (For the "it's inspired by art" fans: It's not. It trains to mangle stuff such into training data, explicitly such, that it can pretty much reconstruct it. While not 100% accurate, it's the actual function of the thing. It then combines things in an encyclopedic way. This is literally one of the abilities of neural networks, sometimes misunderstood: They can actually "remember" and expose training data in more or less indirect ways. Some uses of machine learning do not seem like they would be outputting something, but under special querying they might reveal bits of training data too, effectively. The image generators are expressly crafted towards as good as possible reproduction. If it had inspiration, it would change dramatically what it paints for a given prompt, the next moment. They reconstructed part of memory and visual processing of brains, if you want to lean that far out of the window. But not inspiration. Calling it inspiration is like selling a photo for a video or a video for reality. This is a medium-wild claim/estimation by me, of course...)

    Post edited by generalgameplaying on
  • WendyLuvsCatzWendyLuvsCatz Posts: 38,056

    generalgameplaying said:

    WendyLuvsCatz said:

    hasn't stopped anyone in the past flooding the internet with bad photography, 3D renders, (click load expressionless) traced screenshots

    You do see the difference?

    1. Previously we had some kind of copyright. So taking down and filtering works to some extent.

    2. Now imitating anything will be very simple, especially if you train on the to-imitate artist's images.

    3. It's not only easier, it works on a buttonpress for everyone, and in addition scales to millions of users within no time.

    4. Competition wants to monetize their machines, while artists rather get demonetized.

    5. (Ruinous competition, for whose data you need...)

     

    This for rethinking approaches of "fairness", in particular.

    and as I said in the other thread, existing laws cover this

    trace Mickey Mouse and try and sell it commercially and see

    ai art can get users selling it fined for copyright  if it contravenes it just like applying a Photoshop filter to a rightclicked downloaded artwork and passing it off as your own

    it's not a magic avoid consequences tool

    just has the additional disadvantage of not being able to get a copyright on the results yourself 

    as should be with the current dataset 

  • generalgameplayinggeneralgameplaying Posts: 517
    edited January 2023

    WendyLuvsCatz said:

    and as I said in the other thread, existing laws cover this

    trace Mickey Mouse and try and sell it commercially and see

    ai art can get users selling it fined for copyright  if it contravenes it just like applying a Photoshop filter to a rightclicked downloaded artwork and passing it off as your own

    it's not a magic avoid consequences tool

    just has the additional disadvantage of not being able to get a copyright on the results yourself 

    as should be with the current dataset 

    You know the results of the lawsuits already? Especially with "fair use" this will certainly go haywire as described.

    Apart from that, it's not the same, it will not be obvious infringing anymore, but lookalikes, style reproduction. It will not be easy to trace for all cases, and with easy imitation and easy obfuscation, there will not be an easy detection anymore. They'll address issues, likely with post-production inpaining/modification for hands/eyes or similar. And then the artist gets demonetized for the crowd that "follows" them... and then it's not just the more proficient people doing some fan fiction, with the occasional rogue one, but the full crowd with the cloud-based ai image generator. And that's what i wrote above too.. who will filter and why? Platforms will, to prevent "kinds of haywire". People will fight over stuff, and artists may be on the short end. At least it's no fun arguing with a platform like Youtube, why they demonetized one of your videos (other people's stories), e.g. resulting in the platform demonetizing literally all your work :). Art platforms may be smaller and different (yet) for most, but i wouldn't count on sanity in general. Could become a "funny phase of transition", with artists publishing on mutliple platforms, each of which will have entirely different sets of filtering rules in place, in addition to platforms being located in different countries with different legislations concerning both of copyright and use of ai.

     All i wrote are actual possibilities (to be ruled out rather). Existing law may prove to be a dinosaur.

    Let's hope for the artists, that i am just stuck on the "everything will be flooded" page, and that it never really gets read by reality...

    Post edited by generalgameplaying on
  • manekiNekomanekiNeko Posts: 1,405
    edited January 2023

    loved the title of this thread - because it's exactly what i hoped for. i mean, enhancing my renders, since i can't manage to give them the extra oomph that makes renders interesting, 95% of mine look kinda always "unfinished", bland, somethng is missing - probably the "art" part, lol.

    so, i really love playing with AI, let myself be surprised by whatever the AI creates, it's so fun, and i also try to create focused/themed results, not always with success, it depends, maybe my prompt ideas are too weird.

    and i only have access to wombo, which is totally free, and nightcafé when i earn enough credits. i do occasionally use stablecog/deepAI but they kick me out after a few times asking me to pay/subscribe, which i can't - others ask me to register/login, which i don't want (yet). stablediffusionweb is kinda ok but no presets and no image import.
    so wombo it is, trying to enhance renders with nightcafé burns thru my credits too fast.

    i have no access to good AI like midjourney etc. which might explain why my attempts at enhancing my own DS renders with a bit of extra style are utterly a fail. i tried nightcafé's restyle tool, the one where you upload an image and then pick a style to modify your source - but honestly their styles are too extreme, i don't want my renders painted by van gogh or the likes, just enhanced.  results are meh, my image in worse.

    when i try nightcafés prompt creation with one of my own renders the results are totally another image, just the colors remain, and this with a source image and 100% fidelity to source...

    i tried wombo with a prompt and source image, it was the best yet - and still either horrible or nice/interesting results.. JUST NOT MY IMAGE, more like if i had just typed in a prompt.
    note: in this sense, DS renders are great if you want a fancy AI image but the AI doesn't get your prompt, this really helps.

    SO WHAT CAN I DO TO REALLY ENHANCE MY RENDERS BUT STILL VERY RESEMBLING, LIKE THE OP OF THIS THREAD DID???
    is this impossible with a free/no-login AI generator?? or is there somewhere an "enhance" AI generator?

    here examples - about a really bad render i made just for fun to test sth, i would never uplad that horrible render on DA or anywhere... *shame*
    the original (made long format for wombo) + the wombo result (looks cool, but it's not MY render in better...) + the grid of the nightcafé results (they're horrible)
    i didn't include the nightcafé restyle because it was just my image in even worse.

    highAliens01_Original-womboFormat.png
    900 x 1568 - 3M
    wombo_HDRstyle_strong.jpg
    960 x 1568 - 210K
    nightcafes results with prompt and source.jpg
    768 x 1280 - 132K
    Post edited by manekiNeko on
  • manekiNekomanekiNeko Posts: 1,405

    WendyLuvsCatz said:

    another DAZ render given the ai treatment (was Bijou in Eowyn dress)

    this is absolutely GORGEOUS! ^^ 

  • manekiNekomanekiNeko Posts: 1,405

    WendyLuvsCatz said:

    meanwhile... 

    back to 

    Remixing your art with ai






    some postwork in Gimp mostly erasing backgrounds of the ai ones with the DAZ one added back

    interesting! but is the 1st your original or they're all AI? and what generator did you use? 

  • manekiNekomanekiNeko Posts: 1,405

    wolf359 said:

    Back on the subject of "remixing you art with AI"

     

    A basic Daz content render "Disney Pixarized" with one click 

    lol.. the femme fatale/sexy borg effect is a tiny tad lost in the pixared version XDD - but she's endearing

  • manekiNekomanekiNeko Posts: 1,405

    McGyver said:

    Well, one thing that's helpful about AI art generators is that it's gonna make for more interesting Christmas cards this year...

    lmao.. this is what happens to those who don't believe in santa XDDD - but you're right, best xmas card ever! or maybe the pic for a xmas episode of the walking dead? 

  • manekiNekomanekiNeko Posts: 1,405

    bytescapes said:

    One of the interesting features of AI image generators like Stable Diffusion, Midjourney and DALL-E is that they can take other images as inputs, and generate new images that are more or less based on those input images. I've been playing with Stable Diffusion to 'remix' some of my 3D art, and the results are quite intriguing. Attached are some examples: the original of the armored soldier was made with Carrara, the woman with DAZ Studio, and the city with Vue.

    One amusing thing is that Stable Diffusion is quite careful not to create anything risqué: notice how the Stable Diffusion woman has modestly covered up, compared to her DAZ Studio original.

    Share your own remixes?

    yup, that's exactly what i expect from AI enhancement/remixing from own renders! i like the extra oomph, especially the space soldier, tho the desert city too.
    it's what i'd wish for my renders - but you say midjourney? outta my reach crying
    anyways, the remixes are cool ^^ 

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,056
    edited January 2023

    I use Visions of Chaos with the machine learning addons on my own computer 

    it downloads stuff I need for me after setting up a few dependancies

    which includes various forms of Stable Diffusion including the Automatic 1111 webui

    you can manually install them yourself too

    I just find this easier 

    is free need at least 8GB VRAM I have 11GB

    Post edited by WendyLuvsCatz on
  • wolf359wolf359 Posts: 3,821

    SO WHAT CAN I DO TO REALLY ENHANCE MY RENDERS BUT STILL VERY RESEMBLING, LIKE THE OP OF THIS THREAD DID???
    is this impossible with a free/no-login AI generator?? or is there somewhere an "enhance" AI generator?
    @manekiNeko

    Take a look at playground.ai
    You get 1000 free images per week
    and it has a new option to “re-imagine any uploaded image 
    you provide.

  • manekiNekomanekiNeko Posts: 1,405

    WendyLuvsCatz said:

    I use Visions of Chaos with the machine learning addons on my own computer 

    it downloads stuff I need for me after setting up a few dependancies

    which includes various forms of Stable Diffusion including the Automatic 1111 webui

    you can manually install them yourself too

    I just find this easier 

    is free need at least 8GB VRAM I have 11GB

    ok, interesting, i'll have to digest that and look up all of this. not sure if my laptop makes it tho i think i do have 1gb from nvidia gforce gt540m (vram ??) and 8gb ddr3 memory (guess it's the normal ram...) - this piece of garbage laptop is from 2012 or sth. but interesting to know that you can install art AI at home!

  • manekiNekomanekiNeko Posts: 1,405
    edited January 2023

    wolf359 said:

    SO WHAT CAN I DO TO REALLY ENHANCE MY RENDERS BUT STILL VERY RESEMBLING, LIKE THE OP OF THIS THREAD DID???
    is this impossible with a free/no-login AI generator?? or is there somewhere an "enhance" AI generator?
    @manekiNeko

    Take a look at playground.ai
    You get 1000 free images per week
    and it has a new option to “re-imagine any uploaded image 
    you provide.

    thanks for the info! i think that you meant playgroundai.com - playground.ai is about a company.. right?

    little issue tho.. Sign in to create your first image :( - i don't want to login/register anywhere yet, pity those 1000 images are really tempting, annoying they don't even let you try a few times before you register... and it only offers signing in with google, i don't have/want any google account (else than my phone and i never used/will use it for another purpose after setting up the phone). IF, and only IF i ever sign/register on an AI site, it will be with an account/login directly there - not thru google/insta/fb or whatever...

    but it's good to know, thx ^^

    Post edited by manekiNeko on
  • WendyLuvsCatzWendyLuvsCatz Posts: 38,056

    yeah manekiNeko  you don't have the hardware but there are many online options 

    https://pharmapsychotic.com/tools.html

    https://www.futurepedia.io

  • wsterdanwsterdan Posts: 2,340

    wolf359 said:

    SO WHAT CAN I DO TO REALLY ENHANCE MY RENDERS BUT STILL VERY RESEMBLING, LIKE THE OP OF THIS THREAD DID???
    is this impossible with a free/no-login AI generator?? or is there somewhere an "enhance" AI generator?
    @manekiNeko

    Take a look at playground.ai
    You get 1000 free images per week
    and it has a new option to “re-imagine any uploaded image 
    you provide.

    I think that should be https://playgroundai.com 

    -- Walt Sterdan

  • N-RArtsN-RArts Posts: 1,457
    edited January 2023

    @manekiNeko : I started out on Playground.ai, but I moved to another site (Mage.Space) because I wanted to do NSFW works. 

    Oh, and it's 1000 images PER DAY, not per week ^^ (Although I haven't managed to make 1000 images in one day... Yet :P )

     

     

     

    Post edited by N-RArts on
  • manekiNekomanekiNeko Posts: 1,405

    WendyLuvsCatz said:

    yeah manekiNeko  you don't have the hardware but there are many online options 

    https://pharmapsychotic.com/tools.html

    https://www.futurepedia.io

    wow cool listings! thanks! also, pharmapsychotic? wtf XDDDD - this is.. insane ^^

  • manekiNekomanekiNeko Posts: 1,405

    N-RArts said:

    @manekiNeko : I started out on Playground.ai, but I moved to another site (Mage.Space) because I wanted to do NSFW works. 

    Oh, and it's 1000 images PER DAY, not per week ^^ (Although I haven't managed to make 1000 images in one day... Yet :P )

     

     

     

    i had a peek at mage space, looks cool! thank you,, one more ^^

    ah and 1000 pix be it per week or day, it's way more than i'd manage to create anyways, lol - even in a frenzy.

Sign In or Register to comment.