Loading...
> >

Thoughts on Leonardo.AI and other AI

TY3DArtTY3DArt Posts: 154
in The Commons

I am looking at all this art produced by AI sites like Leonardo.AI and others.  Have we come to the point where 3D art, photoshop art will come to an end? I am looking at this as more of an enhancement to my art.  But I am looking at all this art that people generate just by putting in a discription in prompts and boom - art done.  No need for Daz, Poser, Blender, Photoshop or any other artist software. I think it is kind of sad where AI is going, it is taking all the real work and creativity out of the artist and putting it all on AI generated.  I been using DazStudio for over 10 years. Buying a lot of product and artist software like Photoshop.  But now there is AI, no need to buy anything. Will this be the end of creative artist?  IDK just my random thoughts on this whole AI art. I kind of think it's a little sad.  Like to know your thoughts.

Comments

  • SnowSultanSnowSultan Posts: 3,596

    It's not the end of traditional art because you still have far less control of the results than many people think. There is no true consistency of characters or environments between generations, hands and any sort of action poses are still a nightmare to get right, and animation always looks like something that came out of an 80s music video. I've experimented heavily with Stable Diffusion and Midjourney, and while both are excellent additions to add to an artist's toolkit, the only thing they might be replacing any time soon is stock photography.

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,220

    well I use both

    I like meat and vegan food too

    you don't have to choose

    that said there are a lot of strong feelings about this topic and threads on it get locked

  • TY3DArtTY3DArt Posts: 154
    edited August 2023

    LOL yeah - I feel the same

    WendyLuvsCatz said:

    well I use both

    I like meat and vegan food too

    you don't have to choose

    that said there are a lot of strong feelings about this topic and threads on it get locked

    Post edited by Richard Haseltine on
  • TY3DArtTY3DArt Posts: 154
    edited August 2023

    I agree with your take, but this AI is still in it's infancy.  I agree right now to me it's just a tool.  But the way things are progressing, IDK. It kind of kills the whole artist vibe, and creativity.  I don't really like that. 

    SnowSultan said:

    It's not the end of traditional art because you still have far less control of the results than many people think. There is no true consistency of characters or environments between generations, hands and any sort of action poses are still a nightmare to get right, and animation always looks like something that came out of an 80s music video. I've experimented heavily with Stable Diffusion and Midjourney, and while both are excellent additions to add to an artist's toolkit, the only thing they might be replacing any time soon is stock photography.

    Post edited by Richard Haseltine on
  • RobotHeadArtRobotHeadArt Posts: 917

    SnowSultan said:

    It's not the end of traditional art because you still have far less control of the results than many people think. There is no true consistency of characters or environments between generations, hands and any sort of action poses are still a nightmare to get right, and animation always looks like something that came out of an 80s music video. I've experimented heavily with Stable Diffusion and Midjourney, and while both are excellent additions to add to an artist's toolkit, the only thing they might be replacing any time soon is stock photography.

    If you train a LORA/Dreambooth/Textual Inversion/whatever else the newest thing is this week, you absolutely can get consistent characters.  It may be a complicated and GPU intensive task, but it is totally possible as I have made my own characters trained on as few as 12 images.  Another approach to getting consistent characters outside of training is to use the Reference Controlnet preprocesor, us the picture of the character, give a general physical description of the character for the prompt, and the use the roop or faceswaplab addons with the character.  It won't be nearly as flexible as training but it will give a pretty good likeness.

  • Richard HaseltineRichard Haseltine Posts: 100,948

    As Wendy said, there have already been several threads discussing this topic at length - and turning itno arguments that had to be locked.

This discussion has been closed.