Commons:Deletion requests/File:PESO PLUMA 2023.jpg
Jump to navigation
Jump to search
Dubious educational usability, there's no need for hosting generative software phantasies when a "good enough" model exists under File:Peso Pluma w. Escorpión Dorado (2023).png. Grand-Duc (talk) 02:46, 10 December 2024 (UTC)
.png){kind=link}
- @Grand-Duc
Keep The photo is a retouch of the other, since the original is in low quality.On Wikimedia Commons It is forbidden to overwrite the image, that is why it was uploaded separately. So the reason for its removal is not valid. Aurelio Sandoval (Mensajes aquí please) 03:44, 10 December 2024 (UTC) - @Grand-Duc That photo and others will be used to illustrate the singer's articles because the quality of the originals is low. Aurelio Sandoval (Mensajes aquí please) 03:45, 10 December 2024 (UTC)
- "The quality of the originals is low" - and so is, technologically unavoidable, the quality of this candidate for deletion. You have to be aware of the characteristics of raster graphics: any information that is not present in your starting point of image editing cannot be added afterwards. It simply does not exist (anymore), that's why any photographer or digital artist has to be aware of Non-destructive editing. So, here, we do have a starting point (File:Peso Pluma w. Escorpión Dorado (2023).png). There's a claim of AI upscaling. This means that a software took a pixel of this base image, got an order to add pixels, parsed through its database of statistically likely colour values for each to be added neighbouring pixel and wrote that into the output file. That is pure fiction, guided by the programmers of the statistic model of the upscaler, not educationally useful for depicting the current likeness of a person. Upscaling of raster graphics never enhances the quality of any image, as no information gets added; only the file size grows. See also: Commons:AI-generated media#Are AI-generated media within the Commons project scope?. Regards, Grand-Duc (talk) 05:05, 10 December 2024 (UTC)
- Keep as COM:INUSE. It's breaking the English Wikipedia's MOS guidelines on the use of AI-upscaled images so I'll remove it from those pages, but if other projects don't mind this kind of stuff, Commons won't stand in their way. In this case the AI has given the subject a mild and questionable makeover (removing all of his minor skin blemishes, making his lips more of a traditional cupid's bow, making his nose a little narrower, making his eyes a little more alert, guessing what his eyelashes might look like), so he doesn't actually look like this. It's also done something needlessly weird to the logo on his hat. Belbury (talk) 09:42, 10 December 2024 (UTC)
- Aurelio Sandoval spammed this thing across several WP editions, after I nominated the file for deletion and approximately around the same time I explained how it's not of a better quality (in my second posting here). So, COM:INUSE shouldn't be an obstacle for deletion, I think. Regards, Grand-Duc (talk) 10:59, 11 December 2024 (UTC)
- @Grand-Duc So it is forbidden to improve images with artificial intelligence? Aurelio Sandoval (Mensajes aquí please) 00:47, 28 December 2024 (UTC)
- It is not forbidden. But it's impossible, as using an automated software especially for upscaling, but also for inventing image parts that were not there beforehand is never an improvement. In fact, the border between acceptable modifications aiming at improvements and falsification IMHO lays at most around the usage of functions like Photoshop's "Content aware filling" (example: modification vs. basic motif). There, a human has still a large leverage on the output, a level of control that goes away when using some AI tool. I think that you should inform yourself about technical intricacies of computer graphics. Saying that eye candy is an improvement (for Wikipedia), is an exhibition of behaviour fit for social networking and Instagram filtering, but not for encyclopaedic work. Regards, Grand-Duc (talk) 01:30, 28 December 2024 (UTC)
- @Grand-Duc Thank you for your help and information. Aurelio Sandoval (Mensajes aquí please) 15:58, 28 December 2024 (UTC)
- It is not forbidden. But it's impossible, as using an automated software especially for upscaling, but also for inventing image parts that were not there beforehand is never an improvement. In fact, the border between acceptable modifications aiming at improvements and falsification IMHO lays at most around the usage of functions like Photoshop's "Content aware filling" (example: modification vs. basic motif). There, a human has still a large leverage on the output, a level of control that goes away when using some AI tool. I think that you should inform yourself about technical intricacies of computer graphics. Saying that eye candy is an improvement (for Wikipedia), is an exhibition of behaviour fit for social networking and Instagram filtering, but not for encyclopaedic work. Regards, Grand-Duc (talk) 01:30, 28 December 2024 (UTC)
- @Grand-Duc So it is forbidden to improve images with artificial intelligence? Aurelio Sandoval (Mensajes aquí please) 00:47, 28 December 2024 (UTC)
- Aurelio Sandoval spammed this thing across several WP editions, after I nominated the file for deletion and approximately around the same time I explained how it's not of a better quality (in my second posting here). So, COM:INUSE shouldn't be an obstacle for deletion, I think. Regards, Grand-Duc (talk) 10:59, 11 December 2024 (UTC)
