| Author | Thread |
|
|
02/05/2026 10:36:36 AM · #1 |
Ran across an interesting article this morning that discusses a wide range of analysis and theory on photography, beauty, historical art reference, etc. I found some references to DPChallenge images and community datasets. It's interesting what the datasets provide, and or indicate, along with other discussion of art in general.
The proper citation for the article is:
Impett, L. (2024) 'Computation and Beauty', The Photographers’ Gallery: Unthinking Photography. Available at: https://unthinking.photography/articles/computation-and-beauty
Direct link ==> Computation and Beauty |
|
|
|
02/05/2026 12:38:54 PM · #2 |
That's actually a very interesting article.
Leonardo Impett is assistant professor of Computer Science at Durham University. He works in the digital humanities, at the intersection of computer vision and art history. He was previously Scientist at the Bibliotheca Hertziana – Max Planck Institute for Art History, Digital Humanities Fellow at Villa I Tatti – the Harvard University Center for Italian Renaissance Studies, and PhD Candidate at the École Polytechnique Fédérale de Lausanne. In trying to bring ‘Distant Reading’ to art history and visual studies, his current research focuses on unveiling the implicit image-theories of computer vision and constructing new computer vision systems based on early modern philosophies of vision. He is an Associate of Cambridge University Digital Humanities, an Associate Fellow of the Zurich Center for Digital Visual Studies, and an Associate Research at the Orpheus Institute for Artistic Research in Music. |
|
|
|
02/05/2026 04:38:02 PM · #3 |
I am struggling to figure out which image I want to enter into Hidden Corners. Can I upload a few candidates and have it tell me which is "prettiest"?
It looks like it was trained on voting data through 2012. Our taste is much more refined now. |
|
|
|
02/05/2026 05:06:12 PM · #4 |
| Fascinating stuff. Just the sort of thing for a guy like me to nerd out on. Thanks for sharing, Barry! |
|
|
|
02/05/2026 05:20:43 PM · #5 |
Originally posted by markwiley:
I am struggling to figure out which image I want to enter into Hidden Corners. Can I upload a few candidates and have it tell me which is "prettiest"?
It looks like it was trained on voting data through 2012. Our taste is much more refined now. |
It appears that an update, although I don't quite comprehend all of it yet, was done in 2022?
"But LAION has also given the AVA/DPChallenge dataset a kind of second life. In 2022, a prettiness estimation algorithm trained on AVA (and a few similar datasets) was used to filter through LAION-5B to find only the prettiest images — leading to a dataset called LAION-Aesthetics. I think we can recognise something of the visual taste of DPChallenge users — middle-aged men in the US — in LAION-Aesthetics’ highly-ranked images." |
|
|
|
03/03/2026 04:08:57 AM · #6 |
Hi all - Leo Impett here, the author of that article! So glad you\'ve found it interesting.
I was always a little worried that my paper would come off as snobby to the \'taste\' of DPChallenge. That was never my point - users never asked to be the taste-machine for the internet, and there is no moral requirement for an internet forum to represent the average opinion of the world, if such a thing were even possible. My point was simply that pre-2012 DPChallenge seems to like some styles of images more than others, and that computer science has interpreted this as some kind of objective measure, even though other people might plainly like different kinds of images.
That said, I\'d be fascinated to hear how you think the judgements have changed since then. Or even - if AI images (like Stable Diffusion) are now effectively being designed to be liked by pre-2012 DP challenge users (there are some caveats here - there are other datasets used like SAC), then maybe the community of digital photographers like yourselves starts to pivot to the sorts of styles that AI tends not to make? I\'m thinking here of things like very formal composition (the crazy symmetry of AI photos), the excessive detail and HDRiness, and so on.
By the way - as well as AI stuff, I\'m pretty sure that at least some smartphone cameras will also have been trained in part on DPChallenge. That\'s also an interesting loop....
|
|
|
|
03/03/2026 08:29:50 AM · #7 |
| Hi Leo - Thanks for chiming in. I look forward to community feedback on your response. Thanks! |
|
|
|
03/03/2026 12:33:03 PM · #8 |
I don't think the taste of DPC has changed much at all in the past 14 years, though voters do seem more tolerant of editing flaws.
The average taste is not artistic, by the way. It prefers polished, familiar, professional images. |
|
|
|
03/07/2026 11:43:46 AM · #9 |
Consider the judges at DPC as well. About half of them are our own members and participants with varied interest (or studies) in any art other than what we see here. If AI is learning from us, that could be a problem. As shown in the article our overall winners are pretty consistent in content and design. But in my opinion many of our losers are far from ugly.
As for what has happened here in the last 14 years? Clearly we aren\'t as robust in number but the collective tastes haven\'t seemed to change much. I think digital processing may play a larger part now but the end results are about the same, at least in terms of \"prettiness\". |
|
|
|
03/08/2026 10:08:25 AM · #10 |
One of the things I've been thinking about a lot since coming (back) to DPC last year, anecdotally, is how differently images seem to perform in voting here compared to images "in the wild" when rated by people who have very different cultural immersions. Without going into too much detail, let's just say that I'm routinely astounded by how poorly certain kinds of stereotypically "pretty" images perform in terms of average score here compared with how frequently downloaded similar images are at places like Pexels and Pixabay. In particular things like sunsets, silhouettes, and botanicals.
I've often wondered how ratings of images here at DPChallenge would change if, for instance, the platform was frequented by (and open to voting by) people who didn't have to register an account as a contributor. I think it erects an arbitrary "participation wall" that makes it difficult for people who don't have an a priori interest in (and presumably a culturally constructed concept of) photography and/or visual artistry to weigh in and make their visual preferences known.
Mind you, I understand 100% why this is impractical and technically unfeasible her at DPChallenge. I wouldn't advocate for changing it, because that's outside the scope of what this platform exists to do. I just wonder. Because if AI is being trained on what makes a "pretty" image by datasets that are evaluated by the image-makers themselves rather than by a larger sample of the population which would include, for lack of a better word, "image consumers" who may or may not have the skills or aesthetic inclinations to construct their own images but still have the prerogative to "like" what they like, then it's possible that there's a functional bias in the data set to begin with, right?
Message edited by author 2026-03-08 10:18:17. |
|
|
|
03/08/2026 12:37:41 PM · #11 |
| I usually like "pretty" images, does it make me a middle aged American man? ;) |
|
|
|
03/08/2026 11:35:07 PM · #12 |
Originally posted by MargaretNet:
I usually like "pretty" images, does it make me a middle aged American man? ;) |
No. THIS would:  MarvinNet 🤪😂 |
|
|
|
03/09/2026 09:13:09 AM · #13 |
Originally posted by kanaj:
it's possible that there's a functional bias in the data set to begin with, right? |
of course there's a bias, but there's also the two and a half kids problem. No one has the average taste. |
|
|
|
03/09/2026 12:17:53 PM · #14 |
Thanks, Art Rolfmao, I am Marvin now ;)
DPC is a competition site so the participants may be inclined to enter images that rate highest very often in order to win. That would skew the result of analysis a bit. |
|
|
|
03/09/2026 01:36:35 PM · #15 |
Originally posted by MargaretNet:
DPC is a competition site so the participants may be inclined to enter images that rate highest very often in order to win. That would skew the result of analysis a bit. |
Digital Photography CHALLENGE versus Digital Photography COMPETITION
"A challenge is primarily a personal test of skill, endurance, or ability, focusing on self-improvement or overcoming a difficult task, whereas a competition is a formal, structured rivalry against others to determine a winner, often for a prize"
|
|
|
|
03/09/2026 03:13:44 PM · #16 |
Originally posted by mariuca:
...structured rivalry against others to determine a winner, often for a prize" |
Yes, Ribbons |
|
|
|
03/09/2026 07:15:09 PM · #17 |
Originally posted by MargaretNet:
Originally posted by mariuca:
...structured rivalry against others to determine a winner, often for a prize" |
Yes, Ribbons |
Right. If you participate in the Digital Photography Competition (not Challenge) |
|
|
|
03/11/2026 10:26:00 AM · #18 |
Originally posted by posthumous:
Originally posted by kanaj:
it's possible that there's a functional bias in the data set to begin with, right? |
of course there's a bias, but there's also the two and a half kids problem. No one has the average taste. |
Hey I am not sure about analogies throughout the history of visual arts, but DPC is torn between "hey look what I can do with cutting edge equipment and Camera techniques v. a definition of "artsy" that eschews traditional notions of "beauty- or at least images based solely on how pretty they are"
the very nature of visual arts is based upon the works quality based upon being "beautiful or pretty" but that can take on different even skewed meanings-in that mundane or wretched things can also be beautiful? |
|
|
|
03/11/2026 11:20:48 AM · #19 |
Originally posted by blindjustice:
Originally posted by posthumous:
Originally posted by kanaj:
it's possible that there's a functional bias in the data set to begin with, right? |
of course there's a bias, but there's also the two and a half kids problem. No one has the average taste. |
Hey I am not sure about analogies throughout the history of visual arts, but DPC is torn between "hey look what I can do with cutting edge equipment and Camera techniques v. a definition of "artsy" that eschews traditional notions of "beauty- or at least images based solely on how pretty they are"
the very nature of visual arts is based upon the works quality based upon being "beautiful or pretty" but that can take on different even skewed meanings-in that mundane or wretched things can also be beautiful? |
I feel like we're running into a terminology problem. If artistic quality is based on being "beautiful" and "mundane or wretched things" can be "beautiful" then "beauty" just means "art". The words are no longer different from each other.
Message edited by author 2026-03-11 11:21:11. |
|
|
|
03/11/2026 02:21:38 PM · #20 |
Originally posted by leoimpett:
Hi all - Leo Impett here, the author of that article! So glad you\'ve found it interesting.
I was always a little worried that my paper would come off as snobby to the \'taste\' of DPChallenge. That was never my point - users never asked to be the taste-machine for the internet, and there is no moral requirement for an internet forum to represent the average opinion of the world, if such a thing were even possible. My point was simply that pre-2012 DPChallenge seems to like some styles of images more than others, and that computer science has interpreted this as some kind of objective measure, even though other people might plainly like different kinds of images.
That said, I\'d be fascinated to hear how you think the judgements have changed since then. Or even - if AI images (like Stable Diffusion) are now effectively being designed to be liked by pre-2012 DP challenge users (there are some caveats here - there are other datasets used like SAC), then maybe the community of digital photographers like yourselves starts to pivot to the sorts of styles that AI tends not to make? I\'m thinking here of things like very formal composition (the crazy symmetry of AI photos), the excessive detail and HDRiness, and so on.
By the way - as well as AI stuff, I\'m pretty sure that at least some smartphone cameras will also have been trained in part on DPChallenge. That\'s also an interesting loop.... |
I am a long time outlier but member of the community such that lots know me as such.
I learned virtually everything I know through my experiences and interactions at/on DP Challenge. For many years it was tremendously active......weekly challenges had hundreds of entries on a regular Tuesday challenge.
The most valuable part of DP Challenge was that with so many people, with so many different perspectives, styles, taklents and creative influences wre *ALL* on display and for the most part readily discussed and shared amongst each other.
Years ago I belonged to two local, quite active photography clubs in my area.
Because I knew *NO* limitations in my desire to learn whatever I could not to mention that the ol' "More tha oe way to skin a cat" enabled me to learn and try multiple ways of accomplishing the same result through different styles or techniques. I ditched both clubs after a while because without the wildly diverse ways of doing things being ever present, their results were constricted in comparison.
The event that did it for me that had me drift away from both clubs? Had a contest that comprised a B&W composition. Entered & voted. I used a techniqueI learnt here that was somewhat unorthodox, nt really intended as a B&W editing tool but it worked really well.
I won the contest. Ugly won. I accumulated more points than the second and third place images added together.
There was discussion afterwards...The Professional photographers in the club that were judges agreed to share their insights with us to improve our work. The head judge led the discussion ad asked us how we went about B&W conversions. I submitted that I had used the RAW converter after desaturating it 'cause if you play with it you get some really cool tonal range effects. i forget exactly who brought that to DPC but we all had fun and added another tool to our kit here.
So this guy listened to me describe this and said, "Well, I guess that could work but it sounds kind of convoluted and could produce dubious results.".
So I pointed to my first place image and said, "That's how I did that image right there.". That was a real perspective pointer for me, and was pretty much the end of my involvement with the local clubs.
IMNSHO, social media and general laziness were the causes of the fairly harsh diminishment of DPC.
It's much easier to post on any one of the generic pages and get the whole, "Oh, your pics are amazing!" from a steady stream of people who just work off of instant appeal with no examining the value of the offering. And most folks do NOT want critique. I get that..... For me it's "Don't tell me how to fix it.....I posted it so it's as I want it. Tell me how you feel, good or bad, that's what I want, your impressions as rendered. But one doesn't get to choose reactions, just review them for reference.....
There simply isn't the same community and interaction there was when there were so many more people.
But you can bet there are a bunch of darn fine photographers out there that are who they are as photographers because of much influence and resources as what they learnt here.
My GF/SO and I have a gallery. I have and sell work people readily like and take home and it's generally known that I have a unique perspective in much of what I do.
That it still remains at almost a quarter of a century is pretty freakin' special.
Pretty sure there are a bunch of us out here that know this place was what unleashed our abilities and vision.
It's no stretch for me to think that DPC has influenced thousands of photogs over the last quarter century in a most positive way.
Anyway.... My $0.02 USD.....
|
|
|
|
03/11/2026 02:31:50 PM · #21 |
Originally posted by blindjustice:
Hey I am not sure about analogies throughout the history of visual arts, but DPC is torn between "hey look what I can do with cutting edge equipment and Camera techniques v. a definition of "artsy" that eschews traditional notions of "beauty- or at least images based solely on how pretty they are"
the very nature of visual arts is based upon the works quality based upon being "beautiful or pretty" but that can take on different even skewed meanings-in that mundane or wretched things can also be beautiful? |
Originally posted by posthumous:
I feel like we're running into a terminology problem. If artistic quality is based on being "beautiful" and "mundane or wretched things" can be "beautiful" then "beauty" just means "art". The words are no longer different from each other. |
I have to laugh at this 'cause *SO* much of that is absolutely on the viewer.
Which one is more beautiful?
 
To me, they are both beautiful......and always will be.
They represent my world to a significant degree.
And DPC taught me how to show it to you.
|
|
|
|
03/11/2026 02:35:48 PM · #22 |
Originally posted by posthumous:
I don't think the taste of DPC has changed much at all in the past 14 years, though voters do seem more tolerant of editing flaws.
The average taste is not artistic, by the way. It prefers polished, familiar, professional images. |
Ding! Ding! Ding! Winner, winner, chicked dinner!
I knew I was becoming a more serious photographer when I started noticing that I scored more poorly yet I statred getting comments from photogs whom I admired and felt that they had a "thing" that I liked and/or wanted.
At DPC you evolve better when you no longer care much about score, IMO.
|
|
|
|
03/11/2026 08:38:58 PM · #23 |
Originally posted by posthumous:
Originally posted by blindjustice:
Originally posted by posthumous:
Originally posted by kanaj:
it's possible that there's a functional bias in the data set to begin with, right? |
of course there's a bias, but there's also the two and a half kids problem. No one has the average taste. |
Hey I am not sure about analogies throughout the history of visual arts, but DPC is torn between "hey look what I can do with cutting edge equipment and Camera techniques v. a definition of "artsy" that eschews traditional notions of "beauty- or at least images based solely on how pretty they are"
the very nature of visual arts is based upon the works quality based upon being "beautiful or pretty" but that can take on different even skewed meanings-in that mundane or wretched things can also be beautiful? |
I feel like we're running into a terminology problem. If artistic quality is based on being "beautiful" and "mundane or wretched things" can be "beautiful" then "beauty" just means "art". The words are no longer different from each other.
|
Can't argue with that. |
|
Home -
Challenges -
Community -
League -
Photos -
Cameras -
Lenses -
Learn -
Help -
Terms of Use -
Privacy -
Top ^
DPChallenge, and website content and design, Copyright © 2001-2026 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 03/31/2026 11:48:30 AM EDT.
