←back to thread

Ask HN: Why are AI generated images so shiny/glossy?

67 points arduinomancer | 6 comments | 16 Aug 24 02:20 UTC | HN request time: 0.715s | source | bottom

I’ve noticed a lot of the time you can tell an image is AI generated because it has a shiny/glossy lighting look to it.

Has anyone figured out why this is the case?

Show context
txnf ◴[16 Aug 24 03:25 UTC] No.41262876[source]▶
there is an "aesthetics" model

https://github.com/LAION-AI/laion-datasets/blob/main/laion-a...

obviously, it reflects the mass preference for glosslop

secondarily it is likely due to a desire to ensure that ai images have a distinct look

replies(2): >>41262936 #>>41263247 #
1. kingkongjaffa ◴[16 Aug 24 03:36 UTC] No.41262936[source]▶
> secondarily it is likely due to a desire to ensure that ai images have a distinct look

This does not make sense and seems like conjecture, do you have a source?

replies(3): >>41263041 #>>41263431 #>>41277710 #
2. cen4 ◴[16 Aug 24 03:58 UTC] No.41263041[source]▶
Once the herd starts stampeding in one direction, we get runaway processes - https://www.racked.com/2017/2/1/14441128/local-news-anchor-i...
replies(1): >>41274983 #
3. HeatrayEnjoyer ◴[16 Aug 24 05:33 UTC] No.41263431[source]▶
It's industry standard procedure to tune your model to output a consistent distinct style, to prevent malicious actors from abusing it and presenting fabricated (but very convincing) images as real.
replies(1): >>41274980 #
4. kingkongjaffa ◴[17 Aug 24 14:52 UTC] No.41274980[source]▶
Credible source?
5. kingkongjaffa ◴[17 Aug 24 14:52 UTC] No.41274983[source]▶
Your link has nothing to do with genAI
6. txnf ◴[17 Aug 24 20:29 UTC] No.41277710[source]▶
there is no source it is pure conjecture, but I would say that there are many many fine tunes available of various image generation models so it is clearly possible to make many styles. thus it must be a conscious choice on the part of a API provider to render by default to a distinct style. there are many plausible reasons they would want to do this. Surely, but I don't have actual evidence from the internal management processes of these organizations that they were doing this for one reason or another.