Commons:Quality images candidates/Archives December 15 2023

Consensual review edit

File:2022_July_-_JawaharKalaKendra_Jaipur_12_(cropped).jpg edit

 

  • Nomination JawaharKalaKendra, Jaipur --Chainwit. 19:56, 10 December 2023 (UTC)
  • Decline
  •   Support Good quality, interesting shadow -- Spurzem 22:09, 10 December 2023 (UTC)
  •   Oppose Very small for a picture of stationary object, below 2MP --Jakubhal 19:45, 11 December 2023 (UTC)
  •   Oppose Less than 2 MP. --Plozessor 08:14, 12 December 2023 (UTC)
  •   Oppose less than 2 MP --Sandro Halank 14:57, 12 December 2023 (UTC)
Total: 1 support (excluding the nominator), 3 oppose →   Declined   --Robert Flogaus-Faust 21:06, 14 December 2023 (UTC)

File:Rosenbühler_Weg_20231201_HOF02793_RAW-Export.png edit

 

  • Nomination Rosenbühl path in snow in Hof, Germany. --PantheraLeo1359531 18:43, 1 December 2023 (UTC)
  • Promotion
  •   Comment Is it possible for you to use jpg format? 88 MB for a picture of this size is way over the top. --Imehling 10:13, 8 December 2023 (UTC)
  •   Info I decided to go to PNG because it allows the best opportunity to have a good editing scope. PNG offers better color gradients in image than the JPEG compression for example --PantheraLeo1359531 19:16, 8 December 2023 (UTC)
  •   Comment Interesting question. I would like to have a discussion about that. --Imehling 08:58, 9 December 2023 (UTC)
  •   Comment I'm pretty sure that there is no rule on Commons that would ban PNGs. Recently someone uploaded a giant TIFF file, and even that is allowed. My personal opinion is that PNG is good for the photographer's personal archive, but JPG in quality 11 (of 12) is good enough for export / publication. --Plozessor 09:01, 9 December 2023 (UTC)
  •   Comment Funfact: With irfanview converted to a JPG in the best quality setting and subsampling switched off, the photo is still about 55 MB in size. I then superimposed the two variants and offset them against each other, the result is... almost black. Only if you brighten this result drastically in GIMP shadowy differences become recognisable. The real mistake lies in saving PNG with only 8 bit colour depth per channel, thus combining the worst of both worlds: poor compression of PNG in photos with poor colour depth of JPG. If you really want to take advantage of PNG, you must use 16 bit colour depth per channel. Which of course makes the files considerably larger... --Smial 12:38, 10 December 2023 (UTC)
When I switched to PNG, I took several points into account. I upload some files with 16 bits per channel for best representation in color gradients and upload a smaller file as new version in 8bit. PNG files are saved 1:1 as edited. With JPEG compression, the image is changed in appearance. Yes, "best quality" compression produces artifacts, that are barely visible. But they are there, especially in areas where the human eye does not see details easily. JPEG groups in 8x8 pixel fields and adds patterns to the image. PNG without these changes offers best editing preconditions. PNG as lossless compressed file indicates that the image is the original. Image sensors generate no perfect images, but tend to leave behind and add "sensor patterns" to images, better identifiable with exaggerating color curves afterwards. As they are not visible with typical settings to the visible eye, they may be changed after JPEG compression. The PNG 1:1 leaves as is and can prove authenticity and show that this image is real (especially in times of AI). Of course, some arguments can be questioned by other people, but I think PNG is better here, especially for historic purposes. There also may be differences in chroma subsampling and some more --PantheraLeo1359531 17:28, 10 December 2023 (UTC)
And you can switch from this PNG to JPG at any time, if you wish. So you can choose between an original version and a smaller version :) --PantheraLeo1359531 17:30, 10 December 2023 (UTC)
I did a lossless compression from a PNG to WebP (from 66.6 MiB to 43.7 MiB). This would be a fair deal. But WebP is not really popular --PantheraLeo1359531 17:34, 10 December 2023 (UTC)
  •   Comment Why don't you use 16-bit as a standard to achieve a really useful improvement over JPG? When it comes to authenticity, PNG does not have the slightest advantage over JPG, both are formats derived from the RAW format of the camera used, and the algorithms of the raw developers have considerably more influence, especially when interpolating the Bayer matrix, than the small difference between lossless PNG and "JPG in best quality" would make up. What the fighting term "AI" is supposed to prove is unclear to me, if I save artificially generated images as PNG, they are just as "authentic" as your PNG photos. Please don't misunderstand me, I accept your goodwill in wanting to offer a workaround for the weaknesses of the outdated 8-bit JPG format, I'm just trying to make it clear that 8-bit PNG only offers recognisable advantages over JPG if you approach it with forensic methods. Especially when it comes to contrast or brightness manipulation in digital post-processing, 8-bit PNG is just as shitty as 8-bit JPG. At best, RAW data would be truly authentic. --Smial 19:59, 10 December 2023 (UTC)
  •    Support Reviewing the actual photo, (ahem ;-P ) I think it's a QI. --Peulle 13:08, 11 December 2023 (UTC)
  •   Support ja, sicher. --Smial 15:52, 11 December 2023 (UTC)
  •   Support Schon gut, aber ich sehe eigentlich keinen Grund dafür, die Datei so aufzublasen ohne wirklichen Qualitätsgewinn. --Imehling 16:28, 11 December 2023 (UTC)
Total: 3 support (excluding the nominator), 0 oppose →   Promoted   --Robert Flogaus-Faust 21:01, 14 December 2023 (UTC)