Open main menu

Commons talk:Quality images candidates

Again: downsized candidatesEdit

It seems to me that I am one of the few who look if the image size matches the possibilities of the camera. I always come across resistance with the reference to the 2 MP Linmit. Even less than 30% of the possible resolution for simple landscape motifs are waved through here. If only the 2MP-limmit should play a role and not the possibilities of the camera used, we should abolish the downsampling-rule. I do not mind. But either this rule applies or not. --Milseburg (talk) 17:19, 1 May 2019 (UTC)

I'm looking for the resolution too. ;-) 2 MP is good for cropped images like sport or wildlife. In general photograps should not be downsampled. --XRay talk 17:44, 1 May 2019 (UTC)
Downsampled or not doesn't matter to me at QIC, my criterion for judging sharpness is that it be entirely perfect when scaled to 2 MP. -- King of ♠ 03:49, 2 May 2019 (UTC)
This is an interesting approach, but completely different from how I understand the rules and less demanding. Here more acordance should prevail in my eyes. I'm afraid here is arbitrariness on this point.--Milseburg (talk) 12:38, 2 May 2019 (UTC)
Not in every case, but similar to that "sharp enough @ 2Mpix" I'm often using "good enough to be printed in letter size" as criterion. Printig with 150dpi then results in - tadaa - slightly above 2 MPix. If you want to discourage downscaled uploads I'd suggest a completely other approach: Don't pixelpeep high resolution uploads. --Smial (talk) 13:54, 2 May 2019 (UTC)
Smial, I do not see that the current wording of the downsampling-rule allows your interpretation. If it is optional to apply this rule, we can abolish or at least should reformulate it. I´m open to this but against arbitrariness, if some think that an existing rule does not apply to them and make there own one. --Milseburg (talk) 12:44, 3 May 2019 (UTC)
  • To me, the rule is simple: even if the maximum resolution of the camera is 1.99mpx, such images are ineligible since they contain less than 2mpx. That's the minimum. The most important part of the rule is that we shouldn't allow downsampling, since this would allow for the typical cheating tactic: "Oh my image isn't sharp, but I'll downsize it so it looks sharp enough, then it will be approved." Additionally, it would change the speed at which the standards move. An image taken in 2006 doesn't have to be as good as an image taken in 2016, because cameras have gotten better since then. This means that we have to evaluate images according to the standards at the time they were taken. This will never happen if we allow downsizing, because a 30mpx camera shooting in 2019 will always shoot at 2006 standards if we downsize its images to 15mpx.--Peulle (talk) 12:23, 4 May 2019 (UTC)
    • Peulle, there is a natural tendency to feel that someone who downsizes their 24MP image to 6MP and gets an easy pass is cheating compared to someone who uploads at 24MP and gets pixel-peeped over sharpness, CA, noise and other issues. But this isn't a contest to collect QIs. The real aim should be that (a) Commons QA selects images that are high quality for a wide range of purposes and (b) Commons QA encourages photographers to upload large high resolution images. Not every photograph is taken with a high-quality prime lens, stopped down two stops, on a large sensor, at base ISO, on a tripod. Non-optimal conditions or equipment can mean the maximum resolution raw file image is poor and perhaps there really is only 6MP worth of photo in it, even if shot at 24 or 36MP. In that case, I really don't mind some modest downsizing (and as noted above cropping will drop the MP real fast). What's more harmful is photographers with a perfectly decent 24MP image who only upload 2, 3, 4 MP images to Commons because that's "all that Wikipedia really needs" and keeps the larger size for themselves. We discussed before, raising the minimum, but couldn't agree on what or how to measure. I don't think anyone is happy with the 2MP threshold held over from a decade ago, and I don't agree with KoH insistence that they will assess at that very basic level. But those folks evaluating images at 24, 36 or 50MP are doing great harm -- not only causing folk to downsize but also rejecting useful pictures. I suggest review by downsizing the minimum length to 2000 or thereabouts. Anything above that is a bonus. Reward those who upload full-size with praise rather that wasting effort trying to discover cheats. Resolution is only one measure of a photograph, and one we seem to spend too much time worrying about at QI. -- Colin (talk) 14:17, 4 May 2019 (UTC)
      • "I don't think anyone is happy with the 2MP threshold held over from a decade ago"? When last debated, I was against a blanket rule that might penalize heavily-cropped wildlife images. Charles (talk) 14:42, 26 June 2019 (UTC)
      • Basically what I'm saying is, if someone using a particular lens on a D40 is capable of passing QI, they should never be punished for using a D3200 with the same lens. I see many 24 MP DX candidates and as long as they look OK at 6 MP, I pass them. -- King of ♠ 21:13, 10 May 2019 (UTC)

Images should not be downsampled (sized down) in order to appear of better quality. This rule is simple and does not give you any individual leeway for any other interpretation. So make specific suggestions for phrasing to modify the rule, stand up for abolishong it, or be mindful of it, especially by reviewing a candidate. --Milseburg (talk) 08:29, 25 May 2019 (UTC)

  • Remember that this is from Commons:Image guidelines shared by both FPC and QIC, and one would expect FPC to be more strict than QIC. In practice it has never been a strict requirement; for example most of Diliff's panos are downsampled (otherwise they'd be something on the order of 30,000 x 20,000 pixels), yet garner near-unanimous support at FPC. The spirit of the guideline is to prevent loss of information, and it could very well be that when using a weak lens (e.g. a film-era ultrazoom), there is not much resolution to begin with, and it will not resolve significantly more detail when used on a 36 MP sensor than on a 12 MP sensor. In that case, I wouldn't mind if they downsampled to 12 MP to satisfy the pixel-peepers. -- King of ♠ 02:43, 26 May 2019 (UTC)
    • Yes, but in fact the phrasing of the guidelines make no difference between QIC an FPC concerning downsampling except images of living people. Other exceptions aren´t mentioned, where downsampling is allowed. Current reason for my new advance was File:2019_-_Nationalpark_Jasmund_-_03.jpg among the current candidates. A great picture with FP potential. But it has been scaled down for more sharpness. This is strictly against the current version of the rules. Don´t you or anybody see any need for action for a new version of the rules? --Milseburg (talk) 10:05, 26 May 2019 (UTC)
    • Milseburg that's a 25MP image taken on a 21MP camera. So likely stitched, cropped and perhaps downsized a bit. Ask yourself: if this photographer had instead taken a single shot on their camera, and submitted it here, we'd have only 21MP and probably softer when pixel peeping. Ok, so perhaps if they had not downsized it would be even bigger. And if they only had a 12MP camera, the image would still look just as good on your monitor, but less detail when pixel peeping. So if the 12MP image could be a QI, why are you complaining about a 25MP image. There's only so much the "rules" can do to encourage people not to downsize, and most people are not that bothered unless folk are downsizing to 4 or 6MP silliness. The "rules" can't be written to handle all cases, and if they are too prescriptive and detailed then some folk will take that as a guide to downsize all their photos or we get arguments about the thresholds rather than just trying to use common sense. Common sense tells you this is a high resolution image that is very sharp. It's a QI. If you would like the creator to submit a larger size, why don't you just ask nicely rather than saying it is needed for the rules. -- Colin (talk) 10:48, 26 May 2019 (UTC)
      • I'm afraid I did not understand your point of view. What speaks against an adaptation of the guidelines? The example is a landscape shot and a standard case. The current rule should support promotion, but it does not. --Milseburg (talk) 11:45, 26 May 2019 (UTC)
        • Again, COM:IG is not a rule. It says right there on the page: "Happy nominating, happy judging, and remember…rules can be broken." Different people can have different opinions, and while this has unfortunately led to some lack of consistency in judging QICs, it has worked very well at FPC, where the criteria are whatever voters believe them to be. -- King of ♠ 21:26, 26 May 2019 (UTC)
          • Obviously you understand "guidline" here as a rule, that prescribes the handling of a thing, but does not contain the compulsion to adhere to it. If no one here join into the discussion also wants more reliability and less arbitrary, then that is probably generally wanted and there is no serious need to discuss about non-obligations and everybody continou doing what he wants to do. --Milseburg (talk) 12:26, 27 May 2019 (UTC)
This rule has always been nonsense when it has been interpreted stubbornly. There are already several exceptions for motifs that are technically difficult to photograph, such as sports or stage photography under adverse lighting conditions. Other exceptions apply to stitching and correction of distortions of the lens or perspective used. I'd be very happy if we could phrase something there with "not excessive" or "reasonable" or "meaningful". --Smial (talk) 19:20, 5 June 2019 (UTC)
  • Abolish rule - The guideline against downsizing is absurd. Macro photographs are almost always downsampled or severely cropped since you can only get so much sharpness on tiny subjects due to diffraction softening. This is a limitation of light itself, not of cameras. Scanning electron microscopes were invented specifically to get around this limitation. If you want an ant to be sharply in focus at full resolution, it simply isn't physically possible for it to be 8 megapixel (i.e. not downsampled). Since the rule effectively disqualifies an entire genre of photography, it should be abolished. Kaldari (talk) 15:35, 23 July 2019 (UTC)
  • I've only just spotted your comments above Kaldari. I never downsample my macro photographs and the best ones are not 'severly cropped' either. You do not necessarily get quality problems with a specilaist macro lens. I do agree with you that for very small (or difficult to appraoch) animals like ants, cropping would be essential with my 100mm macro lens. Charles (talk) 11:42, 2 September 2019 (UTC)
As I wrote in the beginning, I do not mind if we do that. But what should we discuss then?--Milseburg (talk) 08:14, 3 August 2019 (UTC)

Categorised in Quality images and Unassessed QI candidatesEdit

Correct me if I'm wrong.

  • If an image is nominated as QI candidate and nobody approves or declines it, it is sorted in category Unassessed QI candidates by a bot when the image is removed from the candidates list
  • You are allowed to nominate the image again for QI
  • When the second time the image is approved, it is sorted in category Quality images

Shouldn't category Unassessed QI candidates be removed when the image passes the second time? --D-Kuru (talk) 13:47, 25 August 2019 (UTC)

Normally it should; but the QICbot doesn't, nobody cares, and the category is irrelevant anyway. --A.Savin 19:29, 25 August 2019 (UTC)
It's not totally irrelevant but luckily fixable with PetScan and QuickCategories. --MB-one (talk) 20:03, 25 August 2019 (UTC)
Does anyone know why and for whom this category is useful? --Smial (talk) 20:31, 25 August 2019 (UTC)
I use the category to find images, that haven't been assessed during the last nomination period. --MB-one (talk) 21:15, 26 August 2019 (UTC)
Und da findest du etwas in den derzeit 14688 Einträgen, die alphabetisch sortiert sind? --Smial (talk) 12:22, 3 September 2019 (UTC)

Problems nominating QIsEdit

I have Qinominator enabled. When I click on the nominate button, this happens:

Great banded grayling (Brintesia circe), Bükk National Park, HungaryNOTE: This image is a focus stacked image consisting of multiple images that were merged using software. As a result, this image underwent digital manipulation which may have included blending, blurring, cloning, and colour and perspective adjustments. As a result of these adjustments, the image content may be slightly different than reality at the points where the images were combined. This manipulation is often required due to lens, perspective, and parallax distortions.I, the copyright holder of this work, hereby publish it under the following license:

Then when I go to QIC and click on the green tab, nothing happens. Anyone got an answer please? Charles (talk) 10:36, 3 September 2019 (UTC)

The "oldfashioned" nominate button works normal. --Smial (talk) 12:26, 3 September 2019 (UTC)
Which button do you mean, please Smial? At the top on my image pages there's a button   Nominate this image for QI. This doesn't work. Charles (talk) 10:09, 4 September 2019 (UTC)
Oh, sorry, I misunderstood. I just used the same button and everything worked as usual. --Smial (talk) 15:00, 4 September 2019 (UTC) https://commons.wikimedia.org/w/index.php?title=Commons:Quality_images_candidates/candidate_list&diff=prev&oldid=364589579
Yes, I saw that, thanks. Weird. I wonder if anyone else knows what's happening. Charles (talk) 22:04, 4 September 2019 (UTC)

Image disappearedEdit

Where has File:Narphu Village, Mustang.jpg gone? It's still listed and promoted at https://commons.wikimedia.org/wiki/Commons:Quality_images_candidates/Archives_June_24_2019 -- Smial (talk) 16:07, 10 September 2019 (UTC)

Fixed. --A.Savin 16:32, 10 September 2019 (UTC)
ARGL... & thx. --Smial (talk) 18:42, 10 September 2019 (UTC)

QIC nominator problemEdit

I don't see the green bar to autonominate images, only the red Review bar. I've de- and re- installed the gadget in preferences but no luck. Any ideas? Rodhullandemu (talk) 21:21, 16 September 2019 (UTC)

IP address as nominatorEdit

Hello, I'm surprised to see from time to time nominations at QIC by IP addresses. Is that ok? To me it looks like an obvious way to bypass the max amount of noms per day. I feel that having an account as prerequisite is not asking to much, other thoughts? --Poco2 19:38, 17 September 2019 (UTC)

I'm with you. --Magnus (talk) 08:02, 19 September 2019 (UTC)
I'd then decline those candidates and ask the user to login in or create an account, if nobody disagrees Poco2 19:18, 19 September 2019 (UTC)

What to do about images that shouldn't have been promoted?Edit

It occurs to me I don't know of any process to deal with images that shouldn't have been promoted. For example, this morning I looked at some QIs of owls to see if it made sense to create a new genus-based subcategory. I clicked on one and realized it did not contain any attempt at identification, File:Owl,Birds Of Bangladesh.jpg (it's also very noisy, but that gets more into subjective matters that I wouldn't have otherwise brought up). As far as I know, images of animals need to be identified to at least the genus. In this case, it may be possible to simply find an identification for it and leave it as-is, but what to do in such cases otherwise? — Rhododendrites talk |  14:28, 19 September 2019 (UTC)

Demoting QIs is a reoccurring question on this page. There are plenty of QIs that have slipped through the QIC system without being detected in time as not really up to the quality and documentation you'd expect from such an image. Right now there is no tool/form/procedure to remove the QI assessment once it is given. Ideas about having a sort of delisting system for QIs too, have been tossed around from time to time, but so far nothing has happened. --Cart (talk) 14:45, 19 September 2019 (UTC)
Return to the project page "Quality images candidates".