Commons talk:Quality images candidates/Archive 26

From Wikimedia Commons, the free media repository
Jump to navigation Jump to search
Archive 20 ← Archive 24 Archive 25 Archive 26 Archive 27 Archive 28 → Archive 30

Picture of the Year 2020

Just a small reminder, the contest has started --Andrei (talk) 19:20, 21 September 2021 (UTC)[reply]

The misleading "Discuss" status again

Here's yet another example of a user quite reasonably thinking that any time they start a discussion about a nomination, they should change the status of the nomination to "Discuss" - the problem, of course, being that in the overwhelming majority of the time, there's no good reason to send the nomination to Consensual Review when there's been no vote, pro or con. Explain to me again why we shouldn't change this status to something clearer like "More votes"? Pure inertia? Can we please do something to make this site work better? -- Ikan Kekek (talk) 12:40, 23 September 2021 (UTC)[reply]

I actually don't care much how this parameter exactly is named, if you feel better with a "More votes" it's fine with me. IMO it's not as important that it would need consensus, more like "be bold (and edit)", however probably the difficulty is, that it needs to modify the bot and the review gadget as well. I'm not an expert on both, so I cannot say anything on the procedure. I think we also have sometimes the problem, that the bots (QIC+FPC) are poorly maintained and no one feels responsible. I would say, at least it's useful to ask the botmaster. And whoever created the gadget. Regards --A.Savin 18:31, 23 September 2021 (UTC)[reply]
You're right, A.Savin. Do you know who the botmaster is, or is there an easy way to find out who they are? Regards, Ikan Kekek (talk) 11:02, 25 September 2021 (UTC)[reply]
QIC bot is owned by Dschwen, the advanced QIC gadget wasn't it by Wilfredor? and the basic QIC gadget ("QIVoter helper") I don't know who is the creator (but the page Help:Gadget-QIvoter was started by Elisardojm). Regards --A.Savin 12:43, 25 September 2021 (UTC)[reply]
Thanks very much. I know that pinging doesn't work on QIC; does it work here on this talk page? If so, User:Dschwen, User:Wilfredor and User:Elisardojm, would you like to assist in changing the confusing "Discuss" parameter to "More votes"? -- Ikan Kekek (talk) 04:39, 26 September 2021 (UTC)[reply]
Ikan Kekek, A.Savin Sorry, I only made a simple help page, I did not participate in its development. Elisardojm (talk) 08:46, 27 September 2021 (UTC)[reply]
Understood. Thanks for checking in. Is there a simple way to edit the help page if "Discuss" is changed to "More votes"? Where can we access the code? -- Ikan Kekek (talk) 17:57, 27 September 2021 (UTC)[reply]
Nevermind, the help page is not at issue, only the script. -- Ikan Kekek (talk) 18:01, 27 September 2021 (UTC)[reply]

Let us raise the QI bar, please!

Coming back from a very long break, I was negatively impressed by how low the QI bar has dropped. Pictures are promoted that fail the most basic criteria of framing, composition, lighting, sharpness, etc., and few seem to care. Particularly disturbing is the fact that some editors seem to automatically promote several images in a row and never use the oppose option, as if they were engaged in some popularity competition. Today, I went through some of those images and tried to explain why I didn't consider them good enough to exhibit the quality seal. But will stop now because I want to avoid any kind of conflict. Come on, guys! -- Alvesgaspar (talk) 14:15, 17 October 2021 (UTC)[reply]

Since QI is mostly decided on the basis of a single reviewer's opinion, we can't have too many variations in the standards of reviewers, or it would be unfair to nominators. As a result, a lot of subjective factors like composition, lighting, etc. are not valid reasons to oppose IMO unless everyone would pretty much agree that it is bad. -- King of ♥ ♦ ♣ ♠ 15:52, 17 October 2021 (UTC)[reply]
What you say have disturbing implications and I wonder if you really mean it. Discarding fundamental quality criteria such as composition and lighting as "subjective", and only accepting them when everyone agrees, means that the requirements of our QI will drop to the lowest possible level: the one of the less experienced and knowledgeable creators and reviewers. Other perverse effects, which I very much suspect are already being felt, are the increase of images needing review and a meaningless competition for having the largest number of promoted pictures. Is this really what we want for QIC? I very much believe that talented and competent photographers like yourself should have a larger share of responsibility on these implications. Alvesgaspar (talk) 19:35, 17 October 2021 (UTC)[reply]
As Commons:Quality images says: "they need not be extraordinary or outstanding, but merely well-composed and generally well-executed." Therefore, I don't view an overcast sky as a valid reason to oppose a QIC when a blue sky may be more appropriate for the subject, so long as the sky is not actually blown out. I've also made various oppose votes over composition at FPC, such as Commons:Featured picture candidates/File:Parco fluviale alta Val d'Elsa 10.jpg, on images that I would easily accept at QIC, because the dividing line here should not be between "great" and "mediocre", but rather between "acceptable" and "bad". -- King of ♥ ♦ ♣ ♠ 21:53, 17 October 2021 (UTC)[reply]
How about QI: Great photo, quite possibly boring and FP: Not boring, perhaps not a great photo. The goal at the end of the 'aughts was for making really good documentation photographs in QI, for infoboxes and other documentation purposes. You would be able to get a great photograph of a person, place or thing from here. FP was for front page stuffs, a photograph of a jet taking off is "Wow" but not so good for an infobox. Not so much "great" and "mediocore", more "wow" and "boring". FP being "great" and QI being "acceptable" is kind of sad for the photographers that participate here, don't you think?--RaboKarbakian (talk) 04:19, 18 October 2021 (UTC)[reply]
QI need not be suitable for use in an infobox; that's what VI is for. It could be documenting some very minor feature of a national park that nobody will ever write even one sentence in a Wikipedia article about. Let's not try to make QI into something it's not; it has always been for photos which meet some minimum bar but there are no requirements beyond that. -- King of ♥ ♦ ♣ ♠ 07:25, 18 October 2021 (UTC)[reply]
General remark: I am totally with King of Hearts here (thank you for your statements!). We should not turn QI into some kind of a “poor man’s FP” and also not confuse it with VI. QI photos do not need to be great, they just must offer solid quality; and solid means, just as the guidelines state, “merely well-composed and generally well-executed”. And this is important and useful, of course. We now have more than 77 millions of media files on Wikimedia Commons, the waste majority of them (> 90%) is of bad quality, just as on Flickr etc., and so QI is very useful if it just marks the files which are of solid quality. --Aristeas (talk) 10:19, 18 October 2021 (UTC)[reply]
However, I would welcome it when we would update some of the more formal requirements. The min. file size (2 Mpx) could easily be increased to 4 or even 8 Mpx today; only for some very special subjects (which are very hard to photograph, e.g. some rare wildlife) smaller files should be allowed. And we could also require a plausible location (geocoding) and basic EXIF data for almost all files (with the sole exception of files where these information would seriously violate privacy).
On the other hand we should take the resolution of the sensor and of the photos into account. With the current practise of judging all images by pixel peeping at 1:1 regardless of their resolution, photographers which downscale their photos are rewarded and photographers which provide the full resolution are punished. This is nonsense, therefore we must take the resolution of the sensor and of the photos into account. It is as easy as 1, 2, 3 to get a photo with only 2 Mpx sharp; it is still easy for most subjects with 12 Mpx; but it is much more difficult when the sensor resolves 40 or 50 and more Mpx: the lenses must be much better, bigger and more expensive, the diffraction becomes visible earlier, you must focus much more carefully, etc. The same problem arises with the requirement that there should be no visible noise. This should be specified more exactly, too. The higher the resolution of a camera sensor gets, the more so-called shot noise (photon noise) you will get, this is not a fault of the photographer or of the camera, but just physics. Therefore while a photo from some low-resolution sensor (or a downscaled photo) should be more or less free of irritating noise, a photo of a high-resolution sensor will always show a bit more noise, and we should accept this, because the current general fear of noise only leads to exaggerated application of denoising which often damages details and creates mushy images. If we take both things together, noise and sharpness, we can easily formulate better rules for QI: In photos near the minimum size the complete subject should be tack sharp and the whole image should be noise-free; in photos about 12 Mpx, the complete subject should be sharp and there should be only a very low level of noise; in high-resultion photos (> 36 Mpx or so), the subject should still be reasonably sharp and the noise should not be irritating. Or, if you want to formulate it in another way: When the image is downscaled to some base size (e.g. 12 Mpx), the subject must be entirely sharp and the photo almost noise-free (a bit more noise may be allowed for very difficult, i.e. dark subjects, especially when the use of a tripod etc. is not possible because of rapid movement or because tripods are prohibited etc.).--Aristeas (talk) 10:51, 18 October 2021 (UTC)[reply]
  • I quite agree with Aristeas in that QIC should not be a "poor man's FP" or be confused with VI. I also strongly support the idea that all QIs should offer a solid quality. But how can we guarantee a solid quality by discarding such important components as framing, composition and lighting because they are not entirely objective? Even more objective parameters, such as sharpness, are often devalued because they are the inevitable result of perspective manipulation. This is deeply disturbing, as those are exactly the quality parameters that come from non-digital photography and are now being discarded because can't be expressed in numbers. By passing this message to newcomers you are jeopardizing one of the noblest goals of QIC, the educational one. The almost absence of critique from our most talented photographers, who seem more interested in enlarging their portfolio of QIs, much adds to the situation. Browsing through today's list of candidates, I counted less than 15 rejected images in an universe of more than 400. The numbers are eloquent and surely do not express the solid quality of the candidates. Alvesgaspar (talk) 12:28, 18 October 2021 (UTC)[reply]
  • If the problem is too low QI bar for today's standards, then yes the obvious solution is to raise the minimum requirements, e.g. 2Mpix->8Mpix. If the problem is that many participants do not respect even the current requirements and misuse QI for gameplaying while the others are unwilling and/or unable to do something about it, then the obvious solution is to ignore QI and spend your time for more useful projects. Sounds harsh, but that's reality. Regards --A.Savin 12:49, 18 October 2021 (UTC)[reply]
  • Agree with A.Savin, I don't see the point in QI. The restriction on user-generated content doesn't help our users. The obsession with pixel peeping and rejection of basic criteria for what makes a good image (in any medium). It harms other projects too, where folk learn to pixel peep at QI and bring those ridiculous standards to FP. QI seems to be all about the quality of the pixels, which is of diminishing concern as you increase image resolution. And QI doesn't seem to be concerned at all about the quality of the image. See my comments at User:Colin/PixelPeeping and User:Colin/ShortestSide (which has an interactive chart you can look at dates).
While I sympathise with Aristeas about high resolution images being pixel peeped and the pressure to downsize to avoid this, I disagree with their solution to judge camera sensor size. For a start, this isn't actually related directly to noise. What counts is the sensor pixel size and us APS-C camera photographers have been walking around with the FF equivalent of a 50MP sensor, cropped to APS-C. Nobody ever gave us any breaks about noise. Traditionally QI is rather obsessed with noise, and fails to take into account conditions such as indoor event photography where it is inevitable. We see comments about noise but also about detail lost through NR and both in fact are likely irrelevant because they are only visible if you are looking 100%, which makes the image about 2 metres wide, observed from 50cm.
A better solution would be to set a standard like 2MP shortest side, and then have a link (using MediaWiki thumbnailer) to an already downsized image, and review at that level all photos. Then we are all e.g. reviewing 2:3 6MP images. Our users do not care what camera you have and I don't think we should care if the 6MP image came from a 18MP smartphone or a 24MP APS-C or 56MP FF camera. If the quality level is X then judge at X. But importantly, we really should be judging "Is this a useful image of sufficient quality" and not "Are these pixels free of noise, CA and sharp"? Those with 4K+ screens will know that once the pixels become invisible then they cease to matter. And examining images at that level is a fools game today when we have AI raw converters that invent feather detail and hair detail that simply wasn't captured by the lens. -- Colin (talk) 13:27, 18 October 2021 (UTC)[reply]
If we require geolocation, we should only make that requirement for images of objects where geolocation is a useful information. Users should not have to disclose the location where they took pictures of common household items, staionary or food or any other objects that are not specific to any location. --Kritzolina (talk) 14:13, 18 October 2021 (UTC)[reply]
I agree with Colin regarding the use of a standardised image for QI purposes. In that way we will be comparing like with like. ALso we will be setting a target use for QI images (compare with the VI target use of "looking good in a computer screen", where the ideal size if 400 x 600 pixels.
  • I'll add my two cents as an outsider that regularly uploads to Commons but doesn't really see the point of QI:
1. Colin's suggestion of automatically opening a scaled version of the image seems very actionable (maybe also for FP and VI) and would ensure an even playing field for reviewing images taken at different resolutions.
Colin, by the way, I don't think Aristeas ever mentioned a relationship between sensor size and noise, but rather between sensor resolution and noise (which is accurate, if one assumes a constant sensor size and hence a direct correlation between resolution and pixel density). On this topic, and if ever further proof was necessary that pixel level comparisons are pointless, Richard Butler has a wonderful article where he demonstrates why higher resolution sensors have more pixel level noise but produce cleaner scaled images.
2. It doesn't seem like an insurmountable task to define clear and transparent rules on composition and lighting. Perhaps it would decrease the number of reviewers to a pool of more experienced photographers/enthusiasts of photography, which doesn't seem like a bad thing either.
3. In the long-term though, there doesn't seem to be much point to the sort of review done in QIC, as the same results will be achievable with algorithms (automatic photo culling based on machine learning is evolving rapidly). But perhaps that is the sort of scenario that really makes sense for QI: imagine having an algorithm that automatically judges the QI merits of all uploaded images, and not only the minority that is nominated!
Julesvernex2 (talk) 17:55, 18 October 2021 (UTC)[reply]
Wrt sensor-size vs pixel-size, I fully agree with you, though I think Aristeas was trying to adjust pixel peeping to compensate for reducing sensor pixel size (and resolution is only a proxy for that if you keep the sensor size the same). Whereas our point and Richard Butler's point is that we should judge a consistently scaled image. If we stopped pixel peeping 100% images, none of this would matter, and photographers would be free to upload whatever resolution they feel is appropriate for the quality of the image they captured.
QI has a problem that really there are a lot of good quality images on Commons and a lot of bad ones. At the moment, QI has put blinkers on to only select photos taken by Commoners and to only accept photos that pass pixel peeping, and not to care if the actual image is good and at a level of quality to be useful for a variety of purposes. I think that is silly and a waste of everyone's time. I think it would be trickier than you suggest to agree on "rules" about composition and lighting. We unfortunately have lots of photographers who haven't progressed in composition beyond the "Rule of Thirds" that their camera manual or beginners book mentioned. A lot of "useful" images are perfectly fine with a "stick the subject in the middle of the frame" approach.
Maybe the process needs to be made lighter weight, where folk can just award an Amazon style rating to images, and they collect "reviews" like that. Stop trying to get a binary result from this and accept not everyone will agree -- Colin (talk) 10:01, 19 October 2021 (UTC)[reply]
It’s always difficult to formulate the relation between sensor size, sensor absolute/relative resolution, noise etc. in a few words without giving room for misunderstandings; and I am sorry if I have caused confusion ;–). To avoid all these misunderstandings, I fully agree that the best solution would be to judge photos at a consistently scaled size, just as Richard Butler’s article suggest (my wording was based on some older article by the same Richard Butler, but I agree that the new article which views the same relation from the other side offers a much more practical description). This would also be more similar to the approach websites like DXOmark, DPReview, ‘Photons To Photos’ etc. use for describing the ‘noisiness’ and dynamic range of a sensor – they scale down the test images to some common constant size, e.g. 8 or 12 Mpx. AFAIK the Mediawiki software is already able to scale photos on request to some given width. Maybe this could be used to provide under each QI candidate an auto-generated link labelled “Show image at normalized size” which would just link to the scaled image; when we have the original size (in x * y pixels), it’s easy to calculate the width value which is necessary to get an 8 Mpx or 12 Mpx image. Of course all of this applies to photos only, not to computer-generated PNGs etc. --Aristeas (talk) 09:29, 20 October 2021 (UTC)[reply]
The thumbnailer is pretty good with normal sized images, but often gives an error if you try to scale a large image or request a very large result. I think it is deliberately capped at a certain amount of memory usage, to prevent it causing performance problems for the site. The results are cached, so if we all view the same resized variant then that won't require the code to re-render the image every time. It even adds a small amount of sharpening to the result, to compensate for the softening that might normally occur. This really needs a javascript solution for the voting page and the community to agree on a standard from which to judge. One reason I suggested a shortest side rule rather than MP is that it copes better with different aspect ratios, and is linear so easier to relate to. The thumbnailer always takes width as a parameter. The MediaWiki api allows one to query an image dimensions, so it is possible to do the maths whichever method we pick (MP or shortest-side). -- Colin (talk) 09:05, 21 October 2021 (UTC)[reply]
  • I would also like to make a small contribution. The requirements for a QI picture should definitely be raised. 8 megapixels for example would be good. But: no rule without exception. There are pictures that do not meet one or the other rule and are still QI in my opinion. For example, vertical lines are not always a good choice, and another choice can still be good. The pictures don't necessarily have to follow a Wikipedia style, they have to be good photographically. --XRay 💬 04:28, 19 October 2021 (UTC)[reply]

I wonder what people here think of the photos at User:Colin/Hertfordshire Way. I took them last year and only had my mobile phone on me, not my DSLR. The were processed in Lightroom from raw. But they were all taken with a tiny smartphone sensor. I think most of them should pass what'd I'd think of as a "quality image" test: they are good but not amazing photos of their subject, with high enough resolution and technical quality to be useful for a variety of purposes. But I suspect most would be pixel peep rejected, and I'd get told I need a DSLR for QI. -- Colin (talk) 10:01, 19 October 2021 (UTC)[reply]

I think smartphones tend to do the best on medium detail like File:Flax in field near Great Gaddesden 2020-07-23.jpg, which you have to squint at to tell it wasn't taken with a DSLR. Cellphone quality is most apparent on large areas of no detail (e.g. empty sky) due to noise, or on very fine detail (e.g. distant vegetation) as it fails to be resolved. FWIW I think most of your images there can pass QI; development from RAW can really do wonders compared to default camera JPEG processing. -- King of ♥ ♦ ♣ ♠ 14:28, 19 October 2021 (UTC)[reply]
  • I am against a change of minimum size rule as large as 2Mpix to 8Mpix. I never downsize an image to please reviewers, but I often cut out an interesting shot from a larger frame, and many good shots would have to be discarded after that change. Perhaps the time has come to raise the size, but in my opinion, a change to 4Mpix would be enough for now. -- Jakubhal 14:35, 2 November 2021 (UTC)[reply]

[Reply]?

Tracked in Phabricator
Task T249293

What's this supposed to do? It seems to appear at the end of any nomination or comment and just takes you back to the nominations page. It should go. Rodhullandemu (talk) 23:24, 20 November 2021 (UTC)[reply]

It’s the reply tool, which became enabled by default on Commmons on Wednesday. It makes replying to talk page comments much easier, but it’s useless (and doesn’t even work) on the QI candidatures. We could hide it with TemplateStyles right now, but maybe it’s better to wait for an official solution, which is tracked in phab:T249293. —Tacsipacsi (talk) 23:44, 20 November 2021 (UTC)[reply]
Can (and should) be disabled via Special:Preferences -> Editing -> Discussion pages. Regards --A.Savin 03:00, 21 November 2021 (UTC)[reply]
No, it should not. Preferences is the right place if someone wants to disable the tool only for themselves, on all pages on Commons. Here we want to disable the tool for everyone (including logged-out users), only on this particular page. This is a completely different use case, and requires a completely different solution. —Tacsipacsi (talk) 12:55, 21 November 2021 (UTC)[reply]

Multiple versions of the same image

Alvesgaspar and I were discussing his nomination of File:Colombo Lisboa November 2021-21.jpg and File:Colombo Lisboa November 2021-12x.jpg, which are two versions of the same raw image, one with perspective correction and one without. Based on my understanding of the rules, editing the same image in slightly different ways should not entitle both versions to become QI. Hoping to hear what others think. -- King of ♥ ♦ ♣ ♠ 20:46, 25 November 2021 (UTC)[reply]

Definitely, no derivative works of a QI can become QI. Poco a poco (talk) 21:25, 25 November 2021 (UTC)[reply]
Same rule for B@&W versions, we have always been consistent with that. Poco a poco (talk)
I wonder where Poco is getting that idea from, I cannot find anything like that in the rules. And it does not make much sense to me either. In fact, I'd ask the opposite question: If the differences are only slight, how can they not both be QI (or both not be QI)? This is not VI where there can only be one image picked per scope, or FP where we try to find the best of the best. This is QI, where a quality image gets the badge regardless of whether there are five more similar shots from the same series. And whether that series consists of a bunch of shots taken in burst mode or a set of different edits does not matter much. And with two alternate edits of the same raw file, how do you even determine which one is the "original" and which is the "derivative"? Each file needs to be assessed on its own, that's the only way this whole project can make any sense. --El Grafo (talk) 12:41, 26 November 2021 (UTC)[reply]
First of all, the fact that this or that user managed to pass some derivative works and got the QI award fro them along with the original files is not the proof that it's a common and accepted practice (in fact, the image with the trees I wouldn't promote, too noisy).
We have had in this discussion page several times complaints from reviewers that saw a bunch of very similar images, taken maybe a few seconds later or from one step further. And I agree, what is the point of doing that? what is the point of uploading variants with different crops or B&W and promote them as QI? What is the benefit for the final user, we should rather have a set of good images instead of flooding the site. Each cell phone today can easily do all kind of simple editing tasks like cropping, converting to B&W and so on in a few seconds, as nobody can previously know what the final user needs.
Btw, derivative works are versions of an original file and that should be documented as such, in the examples above provided by Alvesgaspar there is no information about that.
Supporting this practice looks to me rather like the box of Pandora, we shouldn't do that and I'd like to hear other opinions. Poco a poco (talk) 20:28, 26 November 2021 (UTC)[reply]
  •  Comment It should be clear by now that the purpose of nominating those photos of mine was not to subvert the spirit of QIC, even less to flood the system with similar images. All I intended to do was to draw the attention of both creators and reviewers to the abuse of geometrical manipulations. Yes, that should be considered as making a point, which is sinful in Wikimedia. But using my own photos for the purpose was the least of the evils because I did not hurt the feelings of other creators. As for nominating variations of the same base image, I don’t have strong feelings about the matter or even consider the discussion useful. We all know that very similar images of the same subject are daily nominated and promoted. -- Alvesgaspar (talk) 00:42, 27 November 2021 (UTC)[reply]

Incorrectly promoted QI

I was looking at some very old QI archives and noticed File:Peacock mite, Tuckerella sp.jpg, which can't be a QI because it was photographed by someone from the U.S. Department of Agriculture. I'm not sure how to remove the designation, so I'm posting here. For the record, I would otherwise still consider it a quality photo today. -- Ikan Kekek (talk) 01:40, 7 December 2021 (UTC)[reply]

Delisted Ezarateesteban 23:38, 8 December 2021 (UTC)[reply]

Template/category needed?

I believe File:Central Library, Seattle (2014) - 03.JPG was promoted, but did not get the QI template or category added. Perhaps this is still in a queue, but in case not, can these be added manually? Just noticing how this file seems to have been treated differently than my other recent promotions.

Thanks! -Another Believer (talk) 14:29, 10 December 2021 (UTC)[reply]

Let us raise the QI bar, Part 2

Most QI reviewers are now aware of my discomfort regarding the evaluation criteria and the way images are presently assessed in QIC. As already expressed in another discussion, my feeling is that the quality of the images promoted to QI has dropped to a subminimal level, as a result of an increasing permissiveness. The reasons for this tendency are probably complex but certain (new?) phenomena, such as not being selective in the nomination of own images or avoiding criticizing other reviewers/creators, may have a considerable share of responsibility. Browsing through the many active entries of QIC, one realizes that the red colour is seldom used, as most reviewers prefer to either ignore some nominations or to draw the attention to some of their flaws, rather than just opposing them. If the red colour were used more often, perhaps some creators were more careful when proposing their own images.

Concerning the quality of the reviews, here are some general problems I encountered when browsing the QIC page:

  • Common sense and subjective image assessment, which is mostly based on experience and photographic knowledge, is being partially replaced by objective quantitative criteria that anyone can apply. Some extreme examples involve detailed pixel-peeping examinations to detect flaws like chromatic aberration, dust spots, spherical aberration, or the lack of parallelism between vertical lines.
  • Fundamental photographic criteria of image quality, such as composition, framing, colour and exposure, are often discarded or minimized because they are considered subjective by influential reviewers. The result is that many poorly composed, exposed, or contrasted images are being automatically promoted as examples of quality. Another perverse effect is the degradation of the aesthetical value of our QI galleries.
  • The idea that unfavorable conditions, such as bad lighting, the presence of obstacles or the impossibility of a better shooting angle, justify and mitigate poor image quality. In fact, doing the best possible under adverse conditions should not be a valid reason for promotion.
  • Gross geometrical distortions resulting from the blind application of quantitative criteria or the misuse of panorama projections, are tolerated - when not encouraged. A good example of the first issue is the deformation caused on photographs of buildings, when vertical lines are unnaturally forced to become parallel. A typical example of the second is the deceiving geometry of natural and artificial features, when a projective process is wrongly applied to the construction of a panorama. Apart from aesthetic (or scientific) reasons, I don’t see a valid reason for an illustrative photo of an object to be grossly distinct from what our eyes perceive.

Avoiding (for now) the major problems and starting with easy steps, maybe we can improve the reviewing process by adjusting the content of the Image Guidelines, which are intended at providing assistance to both image creators and reviewers. In my opinion, some indications offered there need to be revised, because they can be detrimental of what is usually understood as a good image. That is the case of the sections about Editing, Distortions and Panoramas, whose content is sometimes invoked as a justification for tolerating some of the problems described above. But before proposing any alterations I would like to know the opinion of as many regular reviewers and creators as possible. Maybe most will reject my diagnosis and consider that nothing is wrong. In that case, I will conclude that my idea of what QIC should be is not shared, and all I must do is to discreetly close this discussion. -- Alvesgaspar (talk) 23:45, 9 November 2021 (UTC)[reply]


Introductory discussion

  •  Comment A quick reply addressing just one of the points listed: the red frame thing has been bothering me for a very long time. These "discussions" above in the nominations section are completely counterproductive. They delay the whole QIC process enormously and, just as you say, obviously tempt people to nominate half-baked images. You don't work out your work carefully before uploading, but rely on a reviewer to find overlooked errors and then make corrections. Or you hope that errors will be overlooked and collect as many badges as possible with little effort... -- Smial (talk) 11:56, 10 November 2021 (UTC) Translated with www.DeepL.com/Translator (free version)[reply]
  •  Agree I completely support your analysis and would like to motivate you in achieving improved standards. I'd like to see more quality. And I don't have any problem with longer QIC process time, if we can raise the level of quality. --Augustgeyler (talk) 15:20, 10 November 2021 (UTC)[reply]
  •  Disagree My understanding of the QI mark is that it is a rather technically correct picture. Common sense and subjective image assessment is a domain of Featured Pictures. The change as stated here would change the QIC page to just a poorer version of those Featured Candidates. I am open to discussing if we should pay more attention to composition and framing (in my opinion yes) and let go of a perspective correction if it doesn't make sense (also yes), but I am strongly against promoting pictures with strong chromatic aberrations or noise, because according to someone's common sense the picture looks special, even if you called it "pixel-peeping". This is another problem I have with that proposition. I do not believe in the one standard of common sense. I think that may be very subjective. My problem with the QIC page is rather that reviews are very uneven. Sometimes, one day, from candidates of very similar borderline quality, one gets promoted and one is discarded. The difference is just another reviewer, because of different times of day, or something as random. I think more common sense would only change it for the worst. In the case of Featured Pictures, it works because there must be several votes. Here very much depends on a single reviewer. -- Jakubhal 15:52, 10 November 2021 (UTC)[reply]
  • I am with you even if we just agree on paying more attention to composition. Additionally for me light is an essential part of a quality image without trying to install a poor version of Featured Candidates, just by saying a QI shows its subject in a suitable light for example. --Augustgeyler (talk) 19:21, 10 November 2021 (UTC)[reply]
  • agree with this assessment. I think there is value in QI and FPC having differing goals, with QI being more focused on technical aspects and FPC on artistic ones. I totally agree about some extremes here, though, like the perfectly parallel verticals that lead to completely distorted church towers and things like those. There is certainly a lot of room for improvement on the guidelines here, which at the same time may raise the bar reasonably (like going from 2 MP to 6 MP as suggested in another comment). I do think, though, that QI should be something that is easily achievable under normal conditions for an experienced photographer. In other words, it is valuable to weed out phone pictures, pictures without any perspective correction / strong tilt or bad lighting. But I think it should not become like FPC where creators are encouraged to only upload the best of their images. If I go out and take pictures of some building and those pictures are fine (lighting, sharpness, perspective), I don't think we should discourage nominating multiple pictures of the same day and building for QI, and asking for selecting only "the best" one(s), at least within reason (e.g. not nominating five pictures that essentially are the same). --Domob (talk) 12:22, 11 November 2021 (UTC)[reply]
  • Regarding technical quality, I think it always makes sense to evaluate at the minimum resolution (currently 2 MP, but I'd be open to something like 4 MP). So if you have CA which is pretty visible on a 36 MP image but nearly disappears when reduced to 2 MP, then it can become a QI without fixing the CA. Same with sharpness and noise. Regarding aesthetic aspects, the problem is that these things are inherently subjective, so I fear that we will the number of CRs will increase drastically if we simple ask people to evaluate whether a composition is "good but not necessarily FP level". I wonder if there is a way to define when perspective correction is and isn't required. My rule of thumb is that any pitch less than 15 degrees requires correction, any pitch greater than 30 degrees should never be corrected, and anything in between is up to the photographer. -- King of ♥ ♦ ♣ ♠ 16:25, 10 November 2021 (UTC)[reply]
  • That would make it too difficult for some subjects, like small birds. QI standards should not be higher than FP standards, as a general rule; of course we may, on an exceptional basis, promote FPs which wouldn't pass QI, but we shouldn't make a habit of doing that. -- King of ♥ ♦ ♣ ♠ 19:27, 10 November 2021 (UTC)[reply]
  • It would be good to raise the QI level. On the one hand there is the technical progress in cameras and lenses and on the other hand the further development of Wikimedia Commons as a media archive. It is not for nothing that many rules are already formulated today with "should" and not "must". The minimum resolution should certainly be raised from 2 MPix to 6 MPix or so. With the "should" images, for example from sports or the animal world, are also taken into account. Or verticals in nature should also be verticals - if they are really vertical in nature and the composition requires verticals. I remember calling for verticals on historic buildings that have no real verticals. Or measuring the sharpness at pixel level in images with 20 or more megapixels, especially since even a comparatively expensive lens cannot do this, especially with high resolutions. It would be idle to always have to declare these exceptions.
It is important to me that a photograph is really good. The focus is not exclusively on the Wikipedia view, which is represented here by some. For a QI image, large parts of it can be blurred if this is used for the composition of the image. It can also be deliberately underexposed or overexposed if this is deliberately used as an element of the image composition. In this respect, one should also take into account the photographic development of all of our photographers. Many of these really good pictures are often rejected at QIC, unfortunately there is also a similar development at FPC. Despite the good but rejected pictures, this ultimately leads to frustration among the photographers and, from the perspective of Wikimedia Commons, to the loss of good, active photographers. --XRay 💬 10:40, 11 November 2021 (UTC)[reply]
  • I think the Commons:Image guidelines applied as QI standards are good and well defined. I would only change the minimum resolution to something higher (which then makes sense as a reviewing resolution), with of course a mention that exceptions can always be made for exceptionally hard to capture subjects.
I strongly disagree with adding more of a composition constraint; that is entirely subjective and does not make for technical quality. We currently reject pictures where the subject is cropped, obstructed, not the main focus, and when there are perspective distortions. I think that is plenty.
Likewise I believe there is no wrong time to take a picture as long as the equipment (software and hardware), skills (right camera settings and software development), and scene (eg acceptable traffic, unobstructed view) allow it. Sometimes that's not possible and it yields poor results such as too much noise, blown highlights, unsharp pictures, and obstructed scenes, and those are already valid criteria to reject an image.
In fact I think QI is a great exercise to test how and how far one can push their software and hardware to produce good quality images outside of ideal conditions, I wouldn't want to deprive follow photographers of this unique opportunity to improve their skills while providing valuable images.
If there are any concrete and concise proposal to modify the Commons:Image guidelines (for example increasing the image resolution as I have proposed in the past) then I'm all ears. In the meantime let's keep things as objective as possible and focus on the technical quality. --Trougnouf (talk) 13:06, 11 November 2021 (UTC)[reply]
  •  Agree I support the views of Alvesgaspar. I strongly disagree with the suggestion from Augustgeyler to raise resolution to 12mp. That is fine for those with high quality cameras and might be OK for 90% of the images that populate QI. It is not appropriate for wildlife where it is often impossible to approach close enough so that cropping is not needed. There is no easy solution, but I have a few suggestions. 1. increase the number of positive votes required to 2, with 2 vs 1 resulting in promotion. 2. Increase the importance of composition (two examples from current nominations are footballers and vehicles). 3. Improve guidlelines to recognise the 'technical quality' standards for an image of a sunny building should not be the same as an image of a forrest-dwelling insect. 4. Alter the software so that a comment on a 'green' nomination does not remove the green box. 5. Alter the guidelines to discourage multiple nominations of the same subject from the same POV. 6. Further emphasise to discourage downsizing. Charlesjsharp (talk) 13:33, 11 November 2021 (UTC)[reply]
  • Quality image in my eyes means, taking a picture and not or just carefully cropping it. In those days, when shooting on film was standard, cropping more than 50 % was seen as no good practice. Today we should keep that practice when selecting QI. If the photographer has to crop more than 50 %, he is just using the wrong lens. One could shot a small bird, crop the result at 70 %. The picture might be OK but not QI. Don't get me wrong, I would support a new minimum resolution which is half the size of an ordinary DSLR sold in the last 6 years. I think this would be something around 6 MP.--Augustgeyler (talk) 15:40, 11 November 2021 (UTC)[reply]
  •  CommentI think we must not exaggerate, not to discourage photographers and also because even those who do not have considerable equipment can still access IQ. Instead I would give a limit to those who present images that are too similar on the same day, so that the same photographer first examines well the photo he is submitting, and presents the best shot. Often one gets the impression, that almost identical shots are submitted to each other, to leave all the work to be done to the examiner. For the technical qualities I consider essential, perspective, focus in most parts of the photo, right framing and right cut. I thank everyone, because I have learned a lot by attending QI, and I would like others to be able to follow the same path as me, and for this reason I would not further tighten the rules. (⧼Anna Massini alias PROPOLI87⧽) (talk) 13:46, 11 November 2021 (UTC)PROPOLI87(⧼Anna Massini alias PROPOLI87⧽) (talk) 13:46, 11 November 2021 (UTC)[reply]
  • The purpose of QIC - This gets lost in many of these discussions, I think. The most visible purpose of FPC is to supply POTD, but it also supplies POTY and serves as "our very best". With QIC, the purpose is less clear. We talk about providing a level of quality for end users, but it's ultimately a community exercise. Our rules are set up not to identify all quality images (we don't allow nominations of work imported from Flickr, etc.) but these discussions always make it seem like we're doing a disservice to users by promoting lower quality images. With the tools we have now (or lack thereof), end users don't really benefit from QIC at all. The one tool we have, FastCCI, is broken most of the time, and has been for years. So end users don't know what QIs are and have no good way to find them. As that's unlikely to change anytime soon, we might as well put that aside and focus on making this the most productive community process possible. That means orienting everything to be about (a) motivating people to upload good work, and (b) helping people become better photographers. I certainly benefitted from feedback at QIC to learn some fundamentals. The challenge to all of these discussions is in that tension between setting useful standards and helping our community, and I'd argue that until we get more people to help on the technical side, making QIs useful at all to end users, we should focus on the community.
This should be a bigger conversation than just a back-and-forth about specific technical standards. There are those demotivated by seeing low-quality promotions, those who like more casual "gamification", and those who just want things to be fair. Since it's a community process first and foremost, none of these are wrong, but we should also be considering other kinds of possibilities. For example, a focus on more meaningful feedback on fewer nominations. Or to base our standards not on technical benchmarks but on something like "getting the most out of whatever equipment you have, including smartphone cameras". Right now, QIC is really only for the small number of people with a real camera. As this is a community process, I think it probably makes sense to consider how to motivate people who don't have a camera but could learn to take better photos with their smartphone. TL;DR - let's not lose sight that this is a primarily community-focused project. — Rhododendrites talk |  14:47, 11 November 2021 (UTC)[reply]
 Comment (Edit conflict) It might be wise to figure out the actual purpose of QI before trying to steer it in any direction. If it's just about wikimedians feeling good about themselves, there's not much reason to change anything. But if QI is supposed to be of any use for people who come to Commons to find pictures they can actually use for whatever project thy are working on? Well, then we should maybe try to adjust our perspective a bit. For example, from the perspective of a random reuser, there is no sane reason to exclude material from external sources from QI. Out there, nobody cares about a bit of CA or that little stitching error you can only see when you zoom in to 100%. Composition & lighting are what makes or breaks an image. --El Grafo (talk) 14:52, 11 November 2021 (UTC) (The main problem of File:Kreuzweg Berghausen Station 13.jpg is not that it is soft, under-exposed and has a big fat purple halo around the cross. The problem is, that it is a lazy snapshot taken with no respect for light and environment.)[reply]
  •  Agree Following Rhododendrites and El Grafo's cue of keeping in mind the purpose of QIC, here are my four cents (the first two assume that QIC continues to be a community exercise, while the last two go out on a limb and envision it as a tool that serves end-users):
  • QIC as a community exercise
  • If the tool's purpose is to motive users to upload better images and learn photography, stricter requirements and more comprehensive feedback seems far more useful than a mere "good quality" response
  • To ensure a more even playing field for evaluating the image's sharpness and (lack of) aberrations/noise, Colin's suggestion of using MediaWiki thumbnailer to open a version of the image scaled to a standard resolution seems like the way to go. Without it, we will continue muddling in the pitfalls of pixel-level comparisons
  • Lightning and composition are less measurable than sharpness and (lack of) aberrations/noise, but that doesn't make them subjective, only more difficult to define. Perhaps, rather than trying to boil them down to rules (e.g., like the well-intended but misleading "Rule of Thirds"), the guidelines should stress that each and every one of the photographer's decisions needs to have a purpose e.g.:
  • The crop should not cut off parts of the subject, unless there's a reason to do so (e.g., chopping off the top of a head to create a closer and more intimate portrait)
  • Lighting should tell a story e.g.: soft and warm for a soothing landscape, directional to bring out the subject's details, harsh for a busy street scene, dark and gloomy for a shadowy back alley
  • Depth of field should focus the attention of the viewer on the subject: wide for a landscape where everything is interesting, narrow to isolate a portrait from the background
  • Creating stricter guidelines will surely lead to some initial confusion, but we should not use that as an excuse do no nothing. Things can and should be improved iteratively
  • QIC as a tool for end-users
  • If the tool's purpose is to support end-users in finding quality images, then it should apply to all uploaded images, not only to the miniscule proportion that is nominated
  • No number of reviewers would be enough to evaluate all images that are uploaded daily to Wiki Commons, so it would need to be done by an algorithm. It would be great to hear from technically savvy users if the existing automatic culling algorithms could be adapted to Wiki Commons
  • The automatic culling algorithms I know of are great in evaluating technical criteria (e.g., sharpness, aberrations, noise) and straightforward images, but get easily confused by creative uses of lighting and composition. Human reviewers could focus on these fringe cases, which would serve two purposes: i) ensuring that the algorithm is not rejecting good images; and ii) showcasing to less experienced photographers the nuances of good lighting and composition
Julesvernex2 (talk) 16:56, 11 November 2021 (UTC)[reply]
 Comment In my opinion, the average quality of the photos that are now turning green is lower than it was before.--Famberhorst (talk) 17:25, 11 November 2021 (UTC)[reply]
  • I certainly agree with Rhododendrites in recognising the value of the community objectives he puts forward, but it seems to me that the primary objective of Commons should not be forgotten. It is a huge bank of royalty-free images, but most of them are of poor or mediocre quality. It is therefore very useful, even if the sorting tool is too often out of order, to provide the end-user with a benchmark to identify images of acceptable quality for editorial use (they can in any case be found via the "quality images of..." categories). Therefore, I am not in favour of raising the file size criterion, but would rather support Colin's proposal for a standard size of review (assuming that this is technically possible). If, whatever the file size, the image has sufficient technical qualities under this condition, then it does have use value for an end user. For the rest, it seems to me that the possibility of challenging and discussing a review, or simply not reviewing or evaluating an image, is sufficient to limit the number of obvious errors of appreciation, including those concerning composition or lighting (which are, at least implicitly, at stake in the absence of review of an image). I would also support the nomination of images imported from Flickr --Velvet (talk) 18:21, 11 November 2021 (UTC)[reply]
    In my opinion, an increase to 4 or 6 MP is a reasonable compromise. Our effective standards for higher-resolution images has long exceeded 2 MP, as can be seen by the unanimous rejection of my QI candidate from 2016 which looks perfectly fine at 2.7 MP. -- King of ♥ ♦ ♣ ♠ 18:31, 11 November 2021 (UTC)[reply]
  •  Comment Interesting conversation. I asked a fellow Finnish Wikipedia writer what does he expect to see in QI category of Commons. Foremost, the basic expectation seems to be (reasonably) clear and bright pictures of a given subject. Quality Images are helpful to find the "technically best" pictures of a subject. I pretty much agree – I think the human interaction is good in the review process, especially since the subjective nature of what makes the picture a good one. It's difficult for me to hit strict rules on the desk. My main driving motivation for photography is Wikipedia as well, and I try to take good photographs with Wikipedia articles in mind. I mostly think the guidelines for Quality Images are quite good, maybe slightly larger minimum resolution would be welcome (not too much though!). On a side note, If some people use Quality Images as the main motivator for own photography, and then quit photography because of opposing votes, it's a damn shame though... It's better to have atleast an average picture on Wikipedia than no picture at all, so I'd welcome anyone to contribute regardless of camera model and technical mastery. In addition to todays technical requirements for the Quality standards, I'd want to give emphasis to reviewing from the Wikipedia standpoint as well. If the picture is technically good in detail but has no purpose, is it still a Quality Image? --Ximonic (talk) 19:34, 11 November 2021 (UTC)[reply]
  • Unfortunately, the Wikipedia view limits the world of photography. In my opinion, every picture has a purpose, an intention, even if you may not recognize it at first glance. --XRay 💬 20:07, 11 November 2021 (UTC)[reply]
  • I see this viewpoint. But I would mention that also the Quality image rules already limit some of the world of photography. What the current standards emphasize is the documentary style of photography, rather than artistic. And I like that. It's a very different photography world you see on Instagram for example. It's not necessarily a bad thing on Wikimedia Commons' part that you will consistently find photographs with more accurate and believable color representation, rather than oversaturated, just for example. The style of photography here is certainly driven more towards naturalistic representation of things and stuff. I recognize this from a standpoint of a fine artist doing nature paintings as another hobby. So I wonder, what are we aiming for with the quality images and it's terms in the end? Are there some photography styles that are then excluded and some that are included? --Ximonic (talk) 20:27, 11 November 2021 (UTC)[reply]
  • The difference is that Wikimedia Commons is a media archive for every purpose (and not only Wikipedia). IMO QI and FP shouldn't be limited to the Wikipedia view. I don't have a idea how to solve these different views. Or maybe it's just a matter of acceptance. Please have a look to en:Abstract photography. These kind of images should be a candidate for QI - some of them. --XRay 💬 20:41, 11 November 2021 (UTC)[reply]
  • I appreciate the existence of such photographs and visuals. I have to admit, I would be very confused of how to review such image on QIC though. Seems to go deeply into the subjective field of how to perceive photography, and in this instance, I wouldn't think myself as qualified to review it. --Ximonic (talk) 21:06, 11 November 2021 (UTC)[reply]
  • IMO it isn't a problem. Most of our rules are made with "should" instead of "must". If an image is described in a way like abstract photograph we can decide to promote it. (Same with animal or sports, 6 MP may be the rule, these kind of photographs may be the exception.) --XRay 💬 08:08, 12 November 2021 (UTC)[reply]
  •  Comment First things first. I've been here for a while (in four days it will be exactly 10 years since I got the first QI) and I don't share the opinion that the quality threshold has been dropped. If it has been like that, it has happened whitout me being aware of that. Sometime ago we had a similar discussion and we checked images promoted and declined some years ago and my takeaway from that discussion was that some pictures that could pass today were declined then and the other way around. I do think also that the criteria of the reviewer can be very diverse. I've always supported to increase the minimum MPx requirement to 4, 6 or 8 MPx and I believe this measure would increase the quality overall and is easy to implement and verify. That is IMHO the point, if we define rules with the purpose to increase the quality threshold, those rules must be verifiable (a "good composition" isn't). If not, then those good meant but bad implented rules will guide again and again to endless discussions and frustration instead of using our limited ressources in prolific matters. I've been wondering also for a while whether the standard criterias of QICs could be automatized. I believe that a good algorithm should distiguish wether images are sharp or not, have CA or dust spots in the sky or are tilted. Getting help in such topics would ease our work a lot and we could review more images, even some which haven't been even nominated. --Poco a poco (talk) 22:06, 11 November 2021 (UTC)[reply]
  • An automated system for sharpness might work some categories of images, but might reject all of your underwater shots and many of my bugs. I have no idea how such a system would cope with images which are sharp (say for the eye or head) but have a shallow depth of field so are 80% blurred. Charlesjsharp (talk) 11:24, 12 November 2021 (UTC)[reply]
  • Yeah, unfortunately the decision of whether a out-of-focus background is appropriate to the subject is not something that can be decided by a simple algorithm. Maybe we can have a tool automatically compute the "effective resolution" of each newly uploaded image, and instead of using that to promote QIs, just make that information available to end users so they can filter by it. The effective resolution is defined as the sum of the resolutions of all portions of the image shrunk to the smallest resolution such that there is no significant detail loss. -- King of ♥ ♦ ♣ ♠ 14:59, 12 November 2021 (UTC)[reply]
  • I've read this whole discussion and have some thoughts. First, I do think it makes sense to increase the minimum acceptable size. I also agree that it's unfair to judge huge photos by the same standards as 2-MP photos. However, there is an issue with judging every photo at the same size: Panoramas are at least often meant to be looked at somewhat closely and not only as entire compositions, and it's reasonable for them to be larger photos than macro photos of tiny insects or invertebrates and judged as such. So I agree with the intention of the idea of judging every photo at the same size, but I think it's a little more complicated than that (we do that at COM:VIC, which is exclusively intended to identify images that are best to use as thumbnails online, not to necessarily be looked at at larger resolutions or printed). In terms of suggestions that a greater degree of information than "Good quality" is needed to be really helpful with reviews, I think that the volume of photos to judge here means that we are not going to get lots of detailed commentary from reviewers, and COM:Photography critiques is more the place for that. On composition, I think it's OK for reviewers to disagree on the relative importance of different criteria and to resolve such issues in CR, but I for one think that the idea that a good photograph can't be taken because there is less light during overcast days, or because there's a shadow somewhere, is not a reasonable standard for quality (and no, I'm not saying that such photos might not be of poor quality sometimes, depending on what the shadows do and so forth), and I also think that unusual or eccentric compositions should be given some license, and that there should be room for artistic compositions that are not sharp because they're shot in fog, mist or what have you to be considered of good quality, regardless of whether they may or may not be used on Wikipedia pages (and we should have enough humility to avoid prejudging what can or can't be used on an encyclopedia, anyway, as there have been photos that were opposed as non-encyclopedic that were afterwards inserted as thumbnails in appropriate Wikipedia articles). I also would think that substituting an algorithm for the views of individual human beings would be a pity, because even if we can never look at everything, the act of looking at and judging photos has a value in itself. -- Ikan Kekek (talk) 15:22, 12 November 2021 (UTC)[reply]
  • First of all I want to thank Alvesgaspar for inviting me into this discussion. I have tried to write something several times, but the discussion keeps growing and I am unsure of what I have to contribute here, as most of you have been here for so much longer. Also I often feel I am just "dabbling" and many people here have a lot more expertise than I have - but because I think this process also needs Wikimedians like me, who at times produce good quality content, who have an eye for good shots, who are happy to showcase the work of others and see their best work showcased in return, but who are not photography experts. For me this is the essence of being a Wikimedian: being allowed to dabble and try my best. And with this I feel I have reached my main point : let us keep the process easy enough so that people like me can enjoy it and dare to participate. I have no qualms, if anyone wants to raise some technical requirements or make certain points a bit clearer. But I would hate to have the images evaluated by algorithms, or by people trying to imitate algorithms. I would feel super estranged by that. Also let's make sure that you do not need the most expensive equipment to participate. A good shot from a simple camera or a good phone under good conditions can have a quality that should be recognized. Let's keep the QI process one that is inclusive .. or strive to make it even more inclusive, like Ikan suggested above. --Kritzolina (talk) 17:27, 12 November 2021 (UTC)[reply]
  •  Comment In my opinion, the level at QI has not dropped over time.When I joined QI, the evaluation of the images was strictly limited to the technical quality. The artistic evaluation is, as Poco says, very difficult and often leads to long discussions that often lead to no result. It is not a good style to accuse photographers who provide their work here for free of any bad intentions such as convenience or sloppiness. Not everyone can photograph equally well. Everyone has a different way of looking at things or a preference for certain motifs and all have their more or less sophisticated processing techniques. A friendly correction should always motivate positively und being personally offended on the other side is counterproductive. As far as automatic evaluation is concerned, I agree with Ikan's and Kritzolina's opinion. Manual work is first choice everything else would perhaps create more problems than solve. I also think that the image should always be in the foreground of the evaluation and not the process of creation or even the equipment with which it was created as Kritzolina also describes.The approval process works well in my opinion. Any complication adds more time to the judging process.The file size could perhaps be raised and on the other hand limited. I now avoid images where I can go to the shower during the loading time. Such images do not have much practical value for me since the end user has to resize the image again, unless he wants to create a billboard.--Ermell (talk) 21:44, 12 November 2021 (UTC)[reply]
  •  Comment The idea of judging the photos in a preselected size (eg. 4k resolution size, around 8 Mpix) sounds nice. Maybe it can also be easily implemented if it is offered with just a click along with other preview resolutions (there is already 320 × 240 pixels, 640 × 480 pixels, 1,024 × 768 pixels, 1,280 × 960 pixels and 2,560 × 1,920 pixels). As far as the lowering of the bar is concerned, I haven't noticed somenthing, there were always reviewers with higher standards than others. Trying to add more specific restrictions about lighting or composition wouldn't work. I have seen a user removing the QI badge from one of my photos because he thought the shadows were to harsh (it was a midday daylight image), so an overcast sky that produces soft shadows (if any) should be a reason alone to decline a photo, especially if with correct editing its flatness can be fixed. Other problems I have encountered is the presence of parked cars in front of a building, where I believe that if they don't hide the main subject and aren't cropped in a distracting way, they shouldn't be a reason to decline. --C messier (talk) 12:07, 13 November 2021 (UTC)[reply]
  • At first I thought the judgment happening in a certain resolution would be a good way to go. But after reading the comment by Ikan Kekek above, I understand that it may be trickier than that, as pictures such as the bigger panoramas may not be the most representative for judgment in the fixed size.  Agree about the parked cars. In a city the cars are most often part of the usual cityscape and to find a building by a busy street without any cars below can be quite an extreme expectation. It's like trying to photograph a city street without any cars on it, how about that. Yes, it would portray a clear view of the street, but wouldn't represent the usual reality of the place quite so well. Maybe some of the documentative value is lost. That's why it's weird to judge a picture by it's cars imo, if they don't harm the composition itself. As matter of fact, I often personally, very consciously, like to include some cars in city views as they can increase the documentary value of the image in my viewpoint. The age of the cars are often a good clue of what time in history the picture was taken. --Ximonic (talk) 20:08, 13 November 2021 (UTC)[reply]
  •  Comment Before focusing into more detailed subjects, let me say how pleased I am with the quality of the contributions to this discussion. I was particularly impressed with the comments of Rhododendrites, El Grafo and Julesvernex2 about the purpose of QIC and its two parallel components (as a community exercise and a tool for end-users), which may help in making a correct diagnosis and taking the necessary steps for improvement. I was also touched with the words of Kritzolina about inclusiveness. When I came to Commons for the first time, I knew almost nothing about digital photography. Engaging into QIC and FPC proved to be a very rich learning experience, as I was in contact with learned and talented photographs, who boldly criticized my first clumsy attempts and taught me how to improve. I never felt excluded as a newbie, even though many of my photos were rejected without mercy. The message I want to pass, Kritzolina, is that inclusiveness applies to people, not to photos. In fact, we don’t learn with appraisal, only with critique. In the limit, having most of our photos promoted with no comments, other than a “good quality” label, is not a rewarding experience. Let me tell you a secret. Taking good photographs is not that difficult, even for a beginner, provided we have good light, and are aware of some basic facts concerning exposure and how the subject should be composed, framed, and illuminated. We don’t even need a fancy camera, which is only required for more difficult subjects (e.g. macrophotography and sports) and conditions. All we need is patience, hard word, a good eye ... and love. -- Alvesgaspar (talk) 12:14, 13 November 2021 (UTC)[reply]
  •  Comment Good to know —kallerna, that you feel the same as I do regarding how the QIC bar has dropped in the last years. One of the main reasons appears to be the way creators now use the forum, more like an administrative service where they can systematically stamp their photos than a place where quality is assessed, recognized and appreciated. That is, quality has given place to quantity. One of the subjects I would like to discuss regarding the revision of the guidelines is the maximum daily number of candidates to QI and the number of positive votes needed for promotion. Another reason, as I've been repetedly emphasizing and Colin has just pointed out below, is the incapacity of many edtors to assess the more subjective qualities, taking refuge in quantitative criteria, some of them silly. Or even the incapacity of opposing a promotion. QIC is ill, IMO. -- Alvesgaspar (talk) 20:49, 18 November 2021 (UTC)[reply]

Resolution increase

Starting a separate thread on a potential resolution increase. I propose the following replacement:

Images should have at least 2 real megapixels of information (with the exception of animations, videos, and SVGs), for example, 1600 × 1250. For "easy to take" images, reviewers may choose to demand more if the image would benefit from it. This rule excludes images computer generated and constructed using a free licensed source code available in the image description.
Images should be sharp throughout when scaled to 6 megapixels, unless otherwise called for by the subject matter (e.g. backgrounds can be intentionally blurred when appropriate, and subjects which are difficult to photograph such as certain types of wildlife can have lower resolution). This rule excludes animations, videos, SVGs, and images computer generated and constructed using a free licensed source code available in the image description.

So basically, for images where everything is supposed to be sharp corner-to-corner, the nominal resolution doesn't matter, nor the pixel-level sharpness. The only thing that matters is that the image looks sharp at 6 MP. If you have a super-sharp 5 MP image that can handle a 10% linear upsample with no obvious signs of degradation, then fine it's a QI, why not? And if you have a 50 MP shot which has bad corners or an out-of-focus foreground/background but you can't see the issue at 6 MP, then that's also a QI. King of ♥ ♦ ♣ ♠ 20:45, 12 November 2021 (UTC)[reply]

  • Images should be sharp throughout when scaled to 6 megapixels - What is the easy, 2-clicks-or-less method to view an image at exactly 6 megapixels? if there is not one without the use of user scripts, I would oppose this for purely practical reasons. — Rhododendrites talk |  21:57, 12 November 2021 (UTC)[reply]
    I don't think Mpx is a good unit of measurement here, as it really only leads to halfway comparable results when looking at regular aspect ratios. For things like panoramas, this just does not make sense. A while ago, someone (Colin?) proposed to use a minimum count of pixels along the shortest side. That seems like a much more reasonable choice, and would be much easier to implement. Width and height in pixels can easily be specified in regular Wiki syntax – just need to tell the template that does the displaying whether the image is in portrait or landscape orientation. El Grafo (talk) 09:36, 15 November 2021 (UTC)[reply]
  • 1. "sharp throughout" is not an appropriate term where an image (of wildlife say) has shallow depth of field. 2. "such as small wildlife" should include wildlife that cannot, by necessity, fill the frame - i.e. flying animals. It is considered bad practice to approach some wildlife too close e.g. courting/mating/nesting animals. Charlesjsharp (talk) 16:34, 14 November 2021 (UTC)[reply]
    Exactly, which is why "unless otherwise called for by the subject matter" is there. I replaced "small wildlife". -- King of ♥ ♦ ♣ ♠ 00:26, 15 November 2021 (UTC)[reply]
  • Let's keep it simple please - remember non-English speakers need to understand the rules as well. The more exceptions and rules beyond the basice requirement we have, the more difficult it gets to translate and understand. --Kritzolina (talk) 07:16, 15 November 2021 (UTC)[reply]

 Oppose I do not think that this proposal makes it easier. --Hillopo2018 (talk) 08:53, 15 November 2021 (UTC)[reply]

  •  Comment I know it's been there before, but while we're at it: What is […] images computer generated and constructed using a free[sic] licensed source code […] supposed to mean? And why would it matter for the resolution whether the source code is freely licensed and available on the file description page. If that's a requirement for QI, then that should be moved to a separate section and not mixed in with considerations about pixel counts. --El Grafo (talk) 09:17, 15 November 2021 (UTC)[reply]

Exposure, lighting, composition, and framing: subjective factors indeed?

  • It has been suggested in this discussion by some editors that poor conditions at the time the photo was taken are mitigating factors for the lack of quality in the process of QI assessment. Those poor conditions may refer to lighting, as well as to all kinds of limitations preventing a clear and well composed image of the object to be taken. According to such interpretation – please correct me if I’m wrong – a photo should be considered as a QI if the best possible results were obtained under those particular circumstances. In the limit, and by absurd, we could have several QIs of the same subject with wildly variable quality.
In my opinion this interpretation subverts the very spirit of QIC in both the components proposed by Rhododendrites: as a community exercise and as an end-user tool. Concerning the educational component, by accepting such principle would imply passing a paternalistic message to the beginners, the one that good intentions may prevail over the results and that knowledge, patience and hard work are not really necessary. Having had a set of poor photos accepted by the QIC community (or assisting to the promotion of such photos, made by experienced photographers), the newbies will eventually give up trying to improve the quality of their own work and gladly engage into a useless collectionism of QIs. Concerning the utility of QIC as an end-user tool, I concur with the interpretation of julesvernex2 that exposure, lighting, composition, etc., are not entirely subjective parameters. Firstly, because they intimately relate to the purpose of the photo. A photo of a building made for illustrative purposes should be well lit, composed and framed, so that its features are clearly shown. Under this light, it doesn’t make sense to promote an underexposed photo just because conditions were not good enough at the time of the shot. And secondly, because some of those parameters (e.g. lighting and contrast) can be numerically expressed, and the result used in the assessment.
Finally, and contradicting the believe of an editor above, it is just not true that poor environmental conditions can be entirely compensated by a good enough gear and digital post-processing. As any serious photographer knows well, most of the success of a good photo is achieved before the button is pressed; and manipulating an image in order to correct a bad shot is always an undesired second choice. -- Alvesgaspar (talk) 19:13, 13 November 2021 (UTC)[reply]
  •  Comment The original photography conditions shouldn't be definite though imo. I would like to give a perspective from living up north in Finland where light is generally drastically different than in central or southern Europe, and we have very distinct seasons. So when someone who doesn't know this, might propose editorial changes to a picture which wouldn't no more represent the reality of this location. Such as making drastic white balance correction to a picture that should represent the blue hour or kaamos (the constantly dim winter period when sun doesn't rise above the horizon in the north at all.) Or maybe you want to photograph the northern lights which are indeed a very challenging subject to photograph in a very crisp quality and have pictures as bright as day. This is a situation where you put almost any kind of a camera and lens in their true limits. But back to my point: I think this is where the original intent of the photograph comes in question. If the subject is clearly kaamos and the picture has therefore a bluer white balance than usual pictures, it should be judged by it's own terms. Otherwise it may result not having any Quality Images that purpusely represent certain events and conditions, such as winter in Lapland. Of course, I encourage photographers to do the best they are able in any given conditions. --Ximonic (talk) 20:40, 13 November 2021 (UTC)[reply]
  •  Comment I don't think anyone is arguing that it's great to underexpose, but there's a difference between underexposure and truly representing the light during an overcast or rainy day, and while Commons is not a solely encyclopedic site, I would appeal to the encyclopedic thinking among all of you to make the point that encyclopedias don't pretend every day is sunny - nor, for that matter, that cityscapes in 2021 lack automobiles. Isn't pretty much everything viewable worth photographing in some fashion? -- Ikan Kekek (talk) 20:54, 13 November 2021 (UTC)[reply]
  • Many difficult photographic situations can be improved in post-processing, but just as not everyone has a high-resolution recent-model camera, not all contributors have Lightroom or Photoshop. Some contributors, however, do not put enoughy effort into post-processing. One solution might be an orange button that says post-processing required with the voter indicating issues with shadows, crop, exposure, noise, dust spots, horizon or whatever. Charlesjsharp (talk) 16:40, 14 November 2021 (UTC)[reply]

The tyranny of the “vertically correct”

  •  Comment The literal observance of what the Image Guidelines suggest about making all vertical lines parallel is becoming quite annoying at QIC, deviating the attention of all (especially those interested in learning) from much more important issues. I did not intend to start this particular discussion yet but was driven to it because two of my pictures (this one and this one) are now suffering from what I consider the tyranny of the “vertically correct”. I’m usually very careful with the composition and framing of my pictures and am perfectly aware of the fact that vertical lines are only shown as vertical when the line of sight is horizontal and the distance to the object is significantly larger that its height. If you point your camera upward and are close to the subject, convergent verticals are unavoidable. Whenever I choose not to make that kind of perspective adjustment, it usually means that I prefer to depict the subject as close as possible to what my eyes perceived at the moment of the shot. That is certainly the case of those two pictures of mine.
Some photographers have rules of thumb regarding when one should, or should not, apply perspective corrections; for example, by considering upper limits to the angle between the two external converging lines. I prefer to experiment the possible solutions having in mind how the picture was taken and how I prefer to depict the subject.
Having said this, I believe that the Image Guidelines on this subject should be corrected as to avoid passing the message that vertical lines should be shown vertical, as a general rule. Discarding the cases where images are accidently tilted or affected by unwanted lens distortions, that should be left to the creator’s decision. It would be more useful to explain that making converging lines to become parallel is more effective (and useful) when the angles are very small; and that the correcting process may introduce distortions that are detected by our eyes and can make the pictures look unnatural, sometimes ridiculously so. -- Alvesgaspar (talk) 15:17, 15 November 2021 (UTC)[reply]
  • I can at least agree with you regarding your examples. In both cases, the insistence on vertical lines is critical, as the angle of view does not allow it and other historical buildings have no verticals at all. (Just one example: With nominating of such pictures I always had and still have problems, because verticals are required. However, the two towers actually lean outwards - in real.) --XRay 💬 15:37, 15 November 2021 (UTC)[reply]
  • BTW: The guidelines say "Images of architecture should usually be rectilinear." There is no "must". Your examples are exceptions and verticals shouldn't ("should" ;-) ) have verticals, the first one should have horizontal lines. --XRay 💬 15:45, 15 November 2021 (UTC)[reply]
  • This is also a good conversation to have. To put my mix in the soup, this one has also been very puzzling to me. I generally do vertical corrections, but in this case the result was absolutely hilarious, and decided to stay far away from those fixes. The belief that verticals should always be vertical in QI, made me decide to not try putting it as a candidate (and because of some noise, which I decided not to fully reduce because it would lose some of the textures.) I understand that the non-vertical verticals may irritate some, but there are many perspective correction cases where the "reality" might become distorted straight into other dimensions. --Ximonic (talk) 19:38, 15 November 2021 (UTC)[reply]
  • I also tried, with the same hilarious results. The problem is that these distortions were caused by a wide-angle lens, and I doubt that any linear transformation will solve the problem. You did well ... -- Alvesgaspar (talk) 21:06, 15 November 2021 (UTC)[reply]
  • That's another great example of what's wrong with QI (and to some degree FP): blindly insisting on "rules" in situations where they don't apply. "Perspective correction" is a misnomer. It is impossible to change the perspective of a photograph after it has been taken. That's why File:Colombo Lisboa November 2021-19.jpg would look utterly ridiculous with straightened verticals: Straight verticals would suggest that the camera is located above the ground at about half the height of the building, looking straight at the building (i.e. line of sight is parallel to the ground). But the fact that we can see the underside of the roof but not the top of the bridge are strong indicators for the true position of the camera: pointing up. In this case, the verticals need to converge towards the en:Vanishing point for the image to make any sense.
The "perspective correction" tools can be useful when use of a wide angle lens exaggerates the perspective. And if the facade of a building is reasonably flat, you might even be able to simulate a higher point of view by making the verticals more vertical. But if you have something like File:Old vs new Tikkurila, Vantaa, Finland, 2021 May.jpg, where due to perspective and ultra-wide angle you can see the underside of the balconies, no amount editing can fix that.
Unless you're shooting from far away with a long lens, 100% perfectly straight verticals are hardly ever desirable (the definition of "far" varies depending on how tall your building is). I've found that even in those cases, a tiny little bit of conversion (sometimes so small you won't be able to spot it without a ruler) is what is needed for an image of a building to look "just right". --El Grafo (talk) 08:17, 16 November 2021 (UTC)[reply]
  •  Comment Too much fo me. I hope there will be a summery of the intended changes before the final voting. I think an update of the giudlines is usefull, but I have no ressourses to join the discussion entensive now. --Milseburg (talk) 18:04, 17 November 2021 (UTC)[reply]
  • I've felt for a long time that Commons:Image guidelines should be completely revised. As Alvesgaspar notes, there are some voting habits that have developed which mark Commons QI/FP as ridiculous compared to other image assessment forums. I think partly that has been due to our clumsy website that only makes it easy to look at an image scaled to fit the browser window or 100% full size and nothing else. So reviewers have got used to examining images in a way that is effectively looking at a 3 or 4 metre wide poster from a distance of 50cm. The other reason is that many reviewers are unable to assess subjective qualities for various reasons (inexperience perhaps but also lack of artistic sensibility or lack of confidence) and are conflict avoiders, so are reluctant to oppose if it means arguing over subjective values. The consequence is they oppose due to objective reasons even when they are silly (like pixel peeping for CA, noise, sharpness or demanding architectural perspective).
Wrt size, this is a perennial request. And the same few photographers oppose it, even when their peers are now routinely uploading higher resolution images. We have a minimum size that belongs in the days of CRT monitors, not 4K televisions. That is long overdue an upgrade, but I do think we need better support from the UI. See User:Colin/ShortestSide for some data and ideas.
User:Ikan Kekek further up commented that some photographs reward a closer look vs as entire compositions. I reflected on this a bit in my old User:Colin/PixelPeeping page, where I split images into Bold and Detailed subjects. I think that could be used to help influence when folk should be making complaints about small details and when they should concentrate more on the entire image.
I think some of our Image guidelines could be replaced by reference to or links to Commons:Photography terms. The latter comments on Architectural perspective and Perspective correction, for example. Both making the same observation that one ideally needs to be as far away from the building as it is tall. -- Colin (talk) 15:53, 18 November 2021 (UTC)[reply]
  • @Alvesgaspar: I see that you opposed File:Bad Dürkheim Kirchgasse 14 003 2021 10 10.jpg, against the opinion of every other !voter. What we can't have is inconsistency, i.e. some people will promote the corrected version and decline the uncorrected version, and others will promote the uncorrected version and decline the corrected version, because then there is no way to win. I think we should, within reason, allow for photographer's discretion in these cases and promote as long as the result is sensible, even if it is different from what you would have personally done. -- King of ♥ ♦ ♣ ♠ 23:35, 18 November 2021 (UTC)[reply]
  • Well, since I was the first to vote we should rather conclude that all the other voters went against my opinion! Whose rational I have explained (contrarily to all but one of the supporters), and which I keep, based on my interpretation of what a QI should be. Is it implied in your comments that one should comply with the opinion of the majority, so that a unanimous decision may emerge? I hope not... -- Alvesgaspar (talk) 00:09, 19 November 2021 (UTC)[reply]
  • I am for relaxed standards when it comes to perspective correction, allowing the photographer to do whatever they want as long as it is reasonable. I thought you were also for that, but now it looks like you want to maintain the same strict standards but in the opposite direction, which is worse since it is contrary to what the rest of the community believes. This is a problem when the majority of QICs are reviewed by one person. -- King of ♥ ♦ ♣ ♠ 15:08, 19 November 2021 (UTC)[reply]
  • I could use your terms to define my own position, as I consider my requirements quite reasonable. Please take into consideration the fact that, until now, the subject of "verticality" was taken quite strictly and that people do not change opinion easily. However, we now have the opinions of some knowledgeable editors in the discussion above, who agree that verticals should not be corrected in situations where an unnatural outcome results. We even have some practical rules of thumb, such as keeping a distance to the subject at least equal to its height. I very much expect that the less experienced reviewers/photographers take these opinions into consideration before automatically support or oppose a nomination. -- Alvesgaspar (talk) 15:53, 19 November 2021 (UTC)[reply]
  • I think we all agree on the broad idea that you've brought up ("don't always correct verticals"), but there is vast disagreement on where to draw the line. If an image requires a +5 correction in Photoshop, then IMO it is unacceptable to leave it uncorrected in most cases; it is large enough to be noticeable, but small enough that correcting it would almost never lead to an unnatural result. On the other hand, I would personally rarely do more than a +40 correction, but if other people do it I would probably be OK with it. -- King of ♥ ♦ ♣ ♠ 16:20, 19 November 2021 (UTC)[reply]
  • Sorry King of Hearts, I don't understand what a +40 correction means, as the only application I have is the poor man's Corel Photo Paint :) But I'm glad to realize that we agree on the general principle that there must be limits to the geometrical corrections. In same cases, it happens that neither the original nor the corrected version are good enough due to geometrical distortion. -- Alvesgaspar (talk) 01:00, 27 November 2021 (UTC)[reply]

Revising the submission and promotion processes

QIC has become a boring place, something like a public service where editors queue up for having their documents approved and stamped. Because there are so many pictures to be evaluated, it is impossible to give to each of them the attention they deserve. Thus, the assessment process was reduced to the very possible minimum, where a single vote is enough for promoting a picture. When we combine this fact with the unwillingness of some reviewers to use the oppose vote or to address such fundamental quality criteria as composition, framing and lighting, we end up in a situation where both the educational and the instrumental components of QIC have dropped to an unacceptable level. Because old habits do not disappear easily, it would be naïve to expect that people will be more selective in nominating their pictures and more competent in reviewing them, just because someone is suggesting so. If we really want QIC to improve, we should start by implementing very simple administrative measures aimed at reducing the number of images waiting to be assessed while improving the assessment process. Here is what I propose:

- To reduce the daily number of images a single editor can submit to 2 or 3.
- To discourage (or prevent) similar images to be submitted in a row.
- To increase the number of support votes needed for promotion to 2 or 3.

Thoughts? -- Alvesgaspar (talk) 23:01, 18 November 2021 (UTC)[reply]

  • One of the best ways to improve quality of the nominations would be IMO reducing the daily number of images. It's hard to find rules to prevent similar nominations, should be common sense, but all of us do not think same way. I would support reducing the amount of daily images to 2. It's quality images, we should promote quality over quantity. —kallerna (talk) 07:09, 19 November 2021 (UTC)[reply]
  • I would personally be fine with a daily nomination number of 2 or 3. By time, it would probably make the individual quality images more valuable and would probably make reviewing a little less laborous, for now. Obviously a large number of very good pictures wouldn't get nominated because of the limit, but is it bad? --Ximonic (talk) 09:15, 20 November 2021 (UTC)[reply]
  • What might seem a boring place to you is a useful place for others. I like the idea of reviewers giving pictures more attention, but if the only way you think about attention are oppose votes, I don't think I like the alternatve, interesting place better. Also I don't like the idea of restricting people from bringing their pictures for review. This is the opposite of inclusive. What about a second different process, where people whose images don't get promoted can turn to to ask how they can improve? Where they might get help in processing if they want that? A mentoring process of some kind? --Kritzolina (talk) 08:58, 19 November 2021 (UTC)[reply]
And how would one find it, when one has ones image not promoted? ^^ Also the title is not really inviting - and yes, I might be a typical woman here, but what I am looking for is not critique, it is help to improve.. --Kritzolina (talk) 10:06, 19 November 2021 (UTC)[reply]
@Kritzolina Good point about the title, I can see how that could look off-putting to some. Any suggestions? Maybe something like "Photo feedback"? And where else do you think it should be advertized? It's already right in the intro at the top of COM:QIC, but I suppose few people actually read that … El Grafo (talk) 13:47, 19 November 2021 (UTC)[reply]
Hmmmmmmmmmmmmmm, I looked twice and did not find this, only on the third pass did I read the top line - making this a bit more visible and linking it more than once (compare to how often the image guidelines are linked) could be helpful. And yes, something more in the lines of "Feedback on how to improve photos" might be more attractive. Thanks for thinking about this! Kritzolina (talk) 15:49, 19 November 2021 (UTC)[reply]
  • I have no problem admitting that I submit images for review to get them approved and stamped, as it seems to me that this makes them easier to find for a potential user (which, after all, is the purpose of Commons). However, I am not opposed to a modification of the labelling process that would provide better guarantees of quality for labelled images. But I must admit that I don't really see what improvements are being sought. I would therefore like to suggest to Alvesgaspar and kallerna (or any other like-minded volunteers) that they look at the labellations of the last two or three days and identify the images that they think should not have been labelled. This would probably enable us to identify more concretely the nature of the defects that they feel should be avoided and therefore to imagine ways of taking them into account in the evaluation guidelines. --Velvet (talk) 09:59, 19 November 2021 (UTC)[reply]
  • Hi Velvet, I went to the last archive and (very quickly) browsed through the promoted images. Here is the list of those which, for me, do not deserve the QI seal:
File:Dülmen, Börnste, Teiche in der Heubachniederung, Pflanze -- 2021 -- 6815.jpg – unfocused
File:Dülmen, Börnste, Teiche in der Heubachniederung, Baum -- 2021 -- 6807.jpg – idem and unbalanced composition
File:Dülmen, Börnste, Middlers Heide -- 2021 -- 6873.jpg – idem, not a good lighting
File:Radovljica Linhartov Trg Pfarrkirche hl Petrus Hauptportal und Rosettenfenster 18082017 0428.jpg – Extreme perspective correction with unnatural result
File:Woman in the shadow of a conical hat made of repurposed bags of laundry detergent in Don Det Laos.jpg – subject unclear (what’s the point?), distracting foreground
File:Grund- und Mittelschule Bayerisches Vogtland 20210808 HOF02479.jpg – poor framing, extreme perspective correction
File:Tiburón azul (Prionace glauca), canal Fayal-Pico, islas Azores, Portugal, 2020-07-27, DD 31.jpg – unsharp, noisy, undetailed
File:Tiburón azul (Prionace glauca), canal Fayal-Pico, islas Azores, Portugal, 2020-07-27, DD 30.jpg – unsharp, noisy, undetailed
File:Tiburón azul (Prionace glauca), canal Fayal-Pico, islas Azores, Portugal, 2020-07-27, DD 16.jpg – unsharp, noisy, undetailed
File:Protestant church in Daun (2).jpg – poor framing, partially unsharp
File:JS-Bach-Platz 22 in Ansbach (4).jpg – partially unsharp owing to perspective correction (a pity because JSBach is my hero)
File:Dokkum, molen De Hoop. 20-09-2021. (actm.) 01.jpg – too soft and noisy, colours look unnatural to me, poor foreground
File:2018 Petergof Cascade Ceres 01.jpg – poor composition and framing, subject too soft -- Alvesgaspar (talk) 11:58, 19 November 2021 (UTC)[reply]
  • Thank you Alvesgaspar for doing this exercise which I find very instructive. Indeed, most of these images are not very beautiful. But, what can be reproached to them is based on criteria already provided for in the Commons:Image guidelines (sharpness, blur, noise, lighting, distortion or even composition). What is at issue is not the guidelines as they stand, but the way they are implemented by the evaluators. Moreover, the discussions above are not about the inadequacy of the criteria, but rather the excessive rigidity with which they are applied (pixel peeping, vertical tyranny...) even though the guidelines do not imply it. I also note that most of the files are 10 MB or larger and that they have been submitted to the label by experienced contributors, many of whose images have been labelled. It is therefore only a small percentage of the images they have submitted to the label. The only conclusion that can be drawn from this is that the labelling of these images should have been discussed. How could they have been? Perhaps by introducing a "weak support" category which would imply at least another evaluation and/or, by introducing a routinised and easily actionable post-labelling appeal procedure. --Velvet (talk) 08:57, 20 November 2021 (UTC)[reply]


IMO a limit of 2 is a very low limit. There is a risk that some good photographers could stop nominating pictures. --XRay 💬 10:33, 19 November 2021 (UTC)[reply]
If someone is productive enough to take more than 700 worthy per year, fine. But then again, then the photographer in question would be able to nominate 700 of his/hers best work per year, and maybe (just maybe) we could see some improvement in QIC standards - which was the point of the conversation. I do not see any risk. —kallerna (talk) 11:50, 19 November 2021 (UTC)[reply]
In this you are assuming every person has time and the internet bandwidth to edit commons more or less on a daily basis. This is definitely not true. I know of people who buy extra internet packages for a month to upload their best pictures - not letting them nominate those quickly would be a severe disservice to those dedicated users. --Kritzolina (talk) 13:27, 19 November 2021 (UTC)[reply]
Limiting the amount of QI candidates to 5 per day was a fatal error, let alone 2. I will stop following or participating in any threads here. Alvesgaspar keeps opening new threads although there is ongoing discussions in older ones or no result have been achieved there (it would then make more sense to me to focus on a few topics and revive those than open more and more). I'm starting to think that all this threads are going nowhere, and therefore are a waste of ressources. We need less and not more regulation. I prefer having 1,000,000 QIs in the project meeting the criteria between 80-120% (as a result of what 5 reviewers would say, with different opionions and approaches) than using the same ressources to have 10,000 QIs with 5 reviewers confirming that they are QI and meet criteria to 100%. Poco a poco (talk) 23:48, 19 November 2021 (UTC)[reply]
I think your shark nominations are a very good example why we should lower the limit. You have uploaded some 30-40 images, of which some are brilliant. But the project doesnt't need those unsuccessful shots, let alone we shouldn't promote the bad ones. Why would you need 30 QIs of the same subject, why you get so upset? Maybe five good quality images from different perspectives would be enough? —kallerna (talk) 08:15, 20 November 2021 (UTC)[reply]
—kallerna Yes, I've uploaded a lo of images of sharks in the Azores, 36 to be exact, of which 7 became so far FP, which is proof that five would have been too little. Those 36 images are the result of let's see, 5 days traveling, preparation, execution and postprocessing. Those pictures are the result of 1200 images, of which I finally kept 756 and uploaded those 36. I've discarded 97% of the images and uploaded 3% and you keep telling me there are bad, noisy, unsuccessful, unsharp, of course I'm upset. Have you ever invested so much time in one image at all? I believe all those images are over the bar and this is an exceptional case of a subject for which I've uploaded a bunch of pictures, I don't think you will remember another case like that among my updates, so, please, don't talk about "a very good example" because it is a very good exception of my contributions here. Poco a poco (talk) 12:17, 20 November 2021 (UTC)[reply]
  • Poco a poco, as an accomplished photographer who has contributed with so many excellent images to Commons, most especially to the FP galleries, it’s hard to believe that you really need to prove anything here or see your effort rewarded the way you suggest. In fact, making the permissiveness of QIC proportional to the difficulty of the shots would seriously subvert its very purpose and spirit, as I’ve been alerting to. -- Alvesgaspar (talk) 12:52, 20 November 2021 (UTC)[reply]
  • I don't need to prove anything, my assessment above is the answer to the suggestion that I don't select my best images before proposing them to QI / uploading them on Commons. If you don't want to understand that taking pictures with a 1200 mm lens, underwater or under whatever specially difficult conditions is comparable to a studio photo, then it is your problem. You just cannot use to all of them the same QI criteria to evaluate them, and this even not considering the resolution provided. I demand here fairness and some consideration to those providing material under difficult conditions. Affirming that this image is "Unsharp & noisy" or this one is "noisy and undetailed" is exactly what I talk about.
  • What I critisized above is rather the way you are doing things here, I see long discussions about a bunch of topics and see no efforts to channel them to come to a kind of conclusion.
  • And last but not least, if the community opts for applying more rules and more bureaucracy increasing the time we need to invest here for a nom or review, then I'll keep off QIC. --Poco a poco (talk) 15:26, 20 November 2021 (UTC)[reply]
  • Proposal - The more I think about the assessment process the more I become convinced that a single positive opinion is not enough for correctly identifying a QI. The reason is, of course, the inevitable biases all of us have. I would go for a minimum of 2, with a twist inspired in the brother FPC forum: an image can only be promoted with a minimum of 2 support votes, plus a minimum difference of 2 to the number of oppose votes. That is, 2-0, 3-1, 4-2, etc. implies promotion, while 2-1, 3-2, 4-3, etc. does not.
Ahh, one more thing: I propose reducing the maximum daily number of nominations per user to 3.
I would also discourage chatting in the small nomination templates and make more flexible taking them to CR. That is, a nomination could be moved there even without any explicit vote, so that any suggestions or discussion could be held there.
Do you think it would be appropriate to start a poll on this proposal now, or should I wait for further comments? -- Alvesgaspar (talk) 13:00, 20 November 2021 (UTC)[reply]
  • Commons includes a large number of poor quality images, a number of good quality images and some exceptional images. In my view, all good quality images should be tagged. The current system is incomplete and homemade, but it is the only realistic one. As far as the images proposed for tagging are concerned, the system is rather efficient: most of the images proposed are unquestionably of good quality and are quickly accepted, a few are unquestionably of poor quality and are quickly rejected, and a certain number, say 10% to 15%, are of questionable quality. It is these 10% to 15%, and only them, that we need to manage as best we can. At present, a number of them are ignored, others are discussed. But some are accepted or rejected without discussion. It is this small fraction of images that are accepted without discussion that we are dealing with here. In my opinion, the search for a better treatment of marginal situations should not lead to complicate the functioning of the whole system. This is why :
- I am not in favour of limiting the number of images proposed daily by a single contributor (yesterday, 16 contributors proposed more than 3 images. Most of them are very experienced and their images are mostly of good quality)
- I do not believe that a double validation procedure should be applied to all images when it is only of interest for a few
- on the other hand, it could be useful to be able to bring an image to the discussion without taking a prior position as clear-cut as for or against (I notice that the number of images brought to the discussion has increased a lot in the last few days, probably in connection with this thread).--Velvet (talk) 12:04, 21 November 2021 (UTC)[reply]
  •  Comment I give up the idea of proposing the number of daily images and the number of positive votes to be changed, as I see no significant support to it. Please go ahead if you feel it is worth. But I'm not abandoning the discussion about the evaluation criteria. More to come in the next days, I hope. -- Alvesgaspar (talk) 21:28, 28 November 2021 (UTC)[reply]
@Alvesgaspar:  I appreciate your effort for get more quality in this project, but sincerely having more restrictive rules may be discouraging for newbies, and the purpose of Commons to get free images may be damaged, QIC VIC and FPC are promoted in teaching courses about WMF projects and participants seems interested in participate here, so if the rules are each time more difficult will be counterproductive, 5 images per day approved by only a reviewer is enough. Regards!! Ezarateesteban 21:40, 5 December 2021 (UTC)[reply]

Purpose of QI

I was invited to comment by Alvesgaspar as the creator of this project some 15 years ago, I'm not actively following the process on a day by day basis. Back then FP was dominated by the likes of NASA, US government, and similar sources scraped from the Web it was easier to scrape an FP than it was to take one. Though Commons had a substancial contributor base, when anyone went to look for our best work it wasnt ours and we couldnt point anyone to place to find it. I spent a couple of months scouring the categories trying to put together 12 user created photographs(basically a calander) of a similar subjects, I couldnt even find one set based solely on colour or even high wikipedia edit interest areas like trains or automobiles. COM:QI was created to address this need and encourage contributors hence the work must be uploaed directly to Commons by the photographer requirement. From that we built a set of criteria aim to recognise the effort individuals put into making Commons more than just a google image search. We chose to display these photographs either being for the subject matter it represented, remember that calander challenge or for the technical aspects making it part of a learning process for everyone to follow. One of best examples of that is @Poco a poco: who would go on to present at Wikimania in London about how QI helped him be a better photographer, I must admit I had one of those proud father moments listening to that talk QI had achieved what I set out for it to do. Key to getting to this moment is the fact that images were declined as much as they were promoted, there is nothing wrong with saying sorry its not quite there you need to do xxxx. That came with balance of recognising that some subject you can go back to over and over again improving every time while others are fleeting moments that cant easily be repeated hence balance in favour of subject over technical.

There was another vision I had with QI its where you are finding yourselves now, which was why QI was a one way process once promoted it stays promoted. QI is an archive of photographic capabilities in 2006 10mpx cameras were the high end, phone cameras were barely distinguishing contrast and colours. We are 15 years from that technology and the technology has taken many leaps in capabilities as has open source software to edit images. QI is about lifting the quality of images being uploaded to do that the criteria must also lift average images are not good enough for QI, FP is for the 1% QI is for 10-15%. QI first step was to limit nominations that was meant to reduce the queues and get the uploader to think about which image is their best, reducing that quantity to 2-3 could help that aspect.

The next step is acknowledging technology has advanced my average phone(Samsung A51) gives me shots that with effort on lighting would pass QI perhaps even nudge FP on current criteria, its video quality was good enough to record my presentations for Wikimania 2021. QI's purpose is to strive for better, to encourage each other to get the best they can out what is readily available the average has changed so should QI.

If I was creating QI now the minimum would be around 12mpx, but QI is at 2mpx I would suggest laying out a timetable to step up slowly let everyone adjust one leap like that is too much and will scare people away. I a small step to 3mpx now, then in July 2022 step again to 4mpx, for 1 Jan 2023 step to 6mpx. That way the whole community is able to take the steps with you. For me I'm proud of what QI has achieved, what you all have done every milestone 1,000, 10,000, 100,000 and now fast approaching 300,000 is amazing. Gnangarra 05:40, 6 December 2021 (UTC)[reply]

Thanks for the very interesting history and thoughts. I looked at some old QI archives, and I can confirm that there was close to a 1:1 promotion vs. decline ratio. -- Ikan Kekek (talk) 05:00, 7 December 2021 (UTC)[reply]
  • Thanks for your comments, Gnangarra. I went through the first year (from July 2006 to June 2007) and found an overall failure rate of 46%, ranging between 32% and 54%. Here are the numbers (please read: month / #noms / #not_promoted / %)
July 2006 /100/33/33
August 2006 /34/11/32
September 2006 54/24/44
October 2006 111/56/50
November 72/34/47
December 35/15/43
January 2007 168/61/36
February 159/69/43
March 2007 109/52/48
April 142/73/51
May 2007 235/104/44
June 2007 359/194/54
Totals 1578/726/46
It is also instructive to go through the reasons for declining the nominations, often invoking composition, framing and lighting issues. If someone is courageous enough (or has the appropriate tools) to make the statistics for all nominations, that would be very useful to the on-going discussion-- Alvesgaspar (talk) 12:30, 7 December 2021 (UTC)[reply]
@Alvesgaspar I think its worth discounting the first two months as the aim there was to get images promoted and encourage the projects development. there was an automated tools first to remove complete nominations, and then later to make it possible to nominate easily so it would be worth those points in time if someone is doing statistics. I also know some local WLE/WLM competitions encourage submissions going through here as well not sure what impact they are having either. Gnangarra 14:30, 11 December 2021 (UTC)[reply]
Wow. Thank you very much for that very impressive insight into QI's history. I really agree with the idea of raising the bar to 12mpx by using a slowly adjusting timetable. August Geyler (talk) 13:01, 7 December 2021 (UTC)[reply]
Retrieved from "https://commons.wikimedia.org/w/index.php?title=Commons_talk:Quality_images_candidates/Archive_26&oldid=620829707"