Generative AI Art

[Pentarctagon]

The eyes and hands are definitely not fixed. The hand has some weird things going on still, the left eye doesn't appear to be an eye (?), and the pupil of the right eye is looking so far to the left it looks like he's cross eyed.

[gnombat]

Once again, I fail to see this Soulless quality in AI artwork, especially in newer engines.

Well, I'm not an artist, so it's hard for me to say exactly what's wrong with these images, but they still feel "off" to me. A real artist could probably give you a better explanation of what's technically wrong with them.

Looking at the first image:



As I said before, this is not bad for an AI-generated image (it's definitely better than most of the awful images in the previous topic). But it feels like the perspective and distances are wrong somehow. The elf, the second person in the distance, and the tree don't really fit together. If you look at the top half of the image, the tree looks fairly close behind the elf and not very big, but if you look at the bottom half of the image, the tree must be truly enormous and some distance away.

It feels like multiple different unrelated images stitched together by a robot (which is perhaps what happened here).

I'm just not really sure why people would want something like that in their game. It seems like it would be more a distraction than an enhancement. (This is without even considering the legal/ethical issues involved.)

[bssarkar]

Well, I'm not an artist, so it's hard for me to say exactly what's wrong with these images, but they still feel "off" to me. A real artist could probably give you a better explanation of what's technically wrong with them.

Looking at the first image:



As I said before, this is not bad for an AI-generated image (it's definitely better than most of the awful images in the previous topic). But it feels like the perspective and distances are wrong somehow. The elf, the second person in the distance, and the tree don't really fit together. If you look at the top half of the image, the tree looks fairly close behind the elf and not very big, but if you look at the bottom half of the image, the tree must be truly enormous and some distance away.

It feels like multiple different unrelated images stitched together by a robot (which is perhaps what happened here).

I'm just not really sure why people would want something like that in their game. It seems like it would be more a distraction than an enhancement. (This is without even considering the legal/ethical issues involved.)

Thanks for the detailed criticism. But we should look at the strong points too.
1. It was just one or two iterations (AI artists often refine things hundreds of times, and I'm not a prompt engineering or AI expert)
2. The lighting, the anatomy (I get it AI usually has trouble with hands and eyes, seen it enough times), the details of scene, the 2.5D shading, the color choice: how many years in art school and industry would it take before somebody creates something at this level? Also are human artists perfect? I can probably nitpick about stuff created by humans too, at least on the intermediate level. (I don't have the skills to criticize a AAA tier game artist, no)
3. AI art is cheap. I'm using Free Tier services here. How much time and effort would it take for an artist? Can they do it in half a minute? I can go thru hundreds of images in an hour.
4. It's also free (except for the legally dubious nature, we really need some definitive legislation). On the other hand, can a human artist afford to give up his hard-worked artwork for free? I'd always prefer human artwork, provided it does meet the criterion I said above (aka the lighting and anatomy at least, and that I'm not going to pay them, sorry. I'd rather use my frankensteined images instead if AI art is not available. You can say I'm stinky.)
4. I have said it already in my previous post, but I'm not saying to take AI images AS-IS. They should be reviewed and adjusted if possible. (I'd probably say the same about any AI generated stuff: it should really be checked.)

Anyway, I give up at this point. Thanks for debating it with me, appreciated. Perhaps I'll reopen this once we have some definite laws/regulations in place. Have a good day.

[Kapiork]

1. What if someone uses an AI generated image as a reference image and draws the image themselves?

I feel like that's not too different from using non-AI art as a reference, with the only difference being that you don't know what the "true" original reference is (that being the images that were used to train the AI), thus not being able to credit the artist.