Commons talk:Quality images candidates/Archive 10

From Wikimedia Commons, the free media repository
Jump to navigation Jump to search
Archive 5 ← Archive 8 Archive 9 Archive 10 Archive 11 Archive 12 → Archive 15

Early Decision on CR

Could we probably agree to stick at least remotely to our rules on CR, which state that the decision on a CR is to be done a minimum of 48 hours after the last entry in the discussion?

I just saw some decisions done considerably earlier than that:

I don’t think that’s good practice. A discussion should always allow others to express themselves, even if this last statement has been made after several days of silence.

Maybe I didn’t notice an alteration of the rules. I just think it’s fair to wait for those two days before closing a discussion. And if the user who is about to close a discussion feels the urge to make another point before, he has to wait another two days, of course.

--Kreuzschnabel 13:17, 8 February 2015 (UTC)[reply]

… all right, that seems to be normal now, I seem to have missed an alteration on the rules. So, after one supporting and one declining vote on CR, we can add our own vote and close the review immediately, making our own vote turn the scale. Interesting.

--Kreuzschnabel 18:47, 8 February 2015 (UTC)[reply]

I take the following view:
  • Discussions should generally last about a week
  • Discussions should not normally be closed less than two days after the last comment
  • SNOW-closures are fine
  • Closures where there is no one arguing for one side are also fine (eg Promotion -> Discussion -> Improvement -> Opposer recants opposition)
But a relaxed view is good. -mattbuck (Talk) 14:50, 9 February 2015 (UTC)[reply]

It’s going on:

User:Hubertl closed the CR on File: Kirschblüte mit Biene (2007-04-06).JPG 39 hours after the last statement, which was his own one, turning the scale (before it was 3 support vs. 3 oppose). This looks as if he didn’t want to leave others too much of a chance to put in another vote. The last statement came from PIERRE ANDRE LECLERCQ – my fault, sorry. But this one being a supporting vote, and Hubertl being supporting too, this does not alter the facts.

User:Hubertl also closed the CR on File:Chao_Say_Tevoda,_Angkor,_Camboya,_2013-08-16,_DD_03.JPG 38 hours after User:Dnalor_01’s statement, giving no chance for User:Averater’s opposing statement which came in 20 minutes after Hubertl’s closure but still much earlier than 48 hours after the last vote, so it was in time and had to be counted actually!

User:C_messier did a nice trick on File:Deux cigognes nid.JPG, closing its CR 3 days before the last vote, taking into account 3 of the 4 votes given after decision.

This won’t work, guys! This is not a battle of pushing opinions through or have one’s own nominations promoted by any rate. This is meant to be consensual assessment of images by their quality, judging from our image guidelines, and we should respect each other’s opinions as well as our own ones.

I am nearly out of QIC because it’s increasingly losing its idea. Up to now, when picking QIs out of a category, I could be sure to get good images. From now on, I can be sure to get images of arbitrary quality but taken by an author the respective reviewer was fond of, or came from the same country, or something like that. Worthless. --Kreuzschnabel 09:33, 12 February 2015 (UTC)[reply]

I checked the CR closing procedure first time. Nearly every closing was done less than 48 hours from the last entry. But there are new votes done after closing. May be the reason is the question “More votes?" within the summary.--XRay talk 09:54, 12 February 2015 (UTC)[reply]

I did no trick in File:Deux cigognes nid.JPG, I added [1] the QICtotal template, not the the QICresult, to ask for more votes in 8 February. The QICresult was added by HumbertI [2] three days later.

Proposal: We should should change the rules in the following way: After at least 8 days from nomination the QICtotal template has to be set as first step and at least 48 hours from this or the last entry the QICresult template should be set as second and last step.--XRay talk 14:53, 12 February 2015 (UTC)[reply]

  •  Support Good proposal. I fully agree. --Code (talk) 15:42, 12 February 2015 (UTC)[reply]
  •  Support Generally agreed. Regarding the cases where there are no (more) diverging opinions (sometimes after rework, all opposes are <s>-ed), the discussion should be closed 48 hours after the last entry regardless of the 8-days-rule (because then it’s a normal promotion and no demand for further discussion). --Kreuzschnabel 18:50, 15 February 2015 (UTC)[reply]

Remarkable

Christian Ferrer wrote above: „Fully agree, there are systematic massive votes based on who previously voted or based on who are the nominators.“ Remarkable! I agree too that such votes should not be. -- Lothar Spurzem (talk) 10:09, 17 February 2015 (UTC)[reply]

We seem to disagree on what constitues "quality"

I've noticed over the past few weeks a disturbing trend here, how QIC is splitting into two groups. One group seem to have high standards for what constitutes a quality image, the other seems to be less concerned about technical aspects. People have brought up various files over the past few weeks here. Here's a few examples from the past week which concern me:

In addition to these there have been other files which get promotions despite clear issues such as overexposure, CA, noise, etc. Others have been declined in what seems to be more or less "I don't like it". Now I don't mean any offence to the authors of these images by this topic, nor to the people who disagree with me, I just think there needs to be a discussion.

It worries me that people will support an image despite others pointing out serious (perhaps fixable) flaws. That means we're not all singing from the same hymn sheet. While there are bound to be differences in how we all review - I know I tend to be more concerned with overexposure and blur, others notice CA, still others dust spots or perspectives - that's natural. But we should agree that major flaws should not be allowed to pass unfixed.

I'm not sure there's anything which can be done about initial reviews which miss faults - everyone is guilty of that sometimes. QIC is a process whereby we start out assuming an image is good quality and then we look to disprove that assumption. But we need to improve the discussion process. For instance people supporting or opposing images without comment, despite contrary votes above. The discussion-initiator is expected to state their reasons, if you're voting later you should be able to justify why you are right and those with differing opinions are wrong. For instance  Support - I think the lighting is acceptable here or  Oppose - the subject is just too dark. In particular a supporting vote should address any issues raised by opposers, and an opposing vote should say exactly what it is that user finds disqualifying about the image.

In the longer term I think we need to reconsider exactly what it is that makes an image "quality", as the dissonance between those who require high standards in all aspects (and I know I'm in that group and that I rightly have a reputation for being picky) and those who seem to be willing to let things slide is increasing. Specific questions we need to answer include "Does a higher resolution excuse unsharpness at 100% zoom?", "Is perspective distortion ever acceptable?", "Do arty photos have a place at QI?".

I look forward to your comments, and before you shoot the messenger I know my own images are not perfect and some are deservedly declined. Some, like most people I imagine, I feel are unfairly declined. Let's try and keep our discussion on-topic rather than picking out specific people. -mattbuck (Talk) 09:56, 12 February 2015 (UTC)[reply]

Fully d’accord. What worries me are reasons like „I like it“. To like a picture is nothing but a matter of personal taste. There are many images I like but that does not mean they fall into our guidelines. There are some images on CR right now which I really like but still decline due to their quality drawbacks. After all, this place is called "quality images" instead of "nice images" or "likeable images".
Even harder to understand was the reason „I just have to support, you know why“ on the tilted lamp image. Asking what that means, I was rudely told not to intrude into things I know nothing about. (Which, of course, was my reason for asking in the first place.) No explanation given, I had to make up my own conclusions on this occurrence of just-having-to-support. What’s going on here?
By the way, there’s a similar phenomenon detectable next door at FPC. Of course this image looks nice, but it’s still unsharp (and wouldn’t need 3.500 euro equipment to be taken correctly focused as was held against my vote).
--Kreuzschnabel 11:30, 12 February 2015 (UTC)[reply]
  • Fully agree, there are systematic massive votes based on who previously voted or based on who are the nominators. Almost aggression when opposed. -- ChristianFerrer 13:09, 12 February 2015 (UTC)[reply]
  • I'm afraid I have to agree. --C messier (talk) 15:58, 12 February 2015 (UTC)[reply]
    I was wondering about that, though I'm not sure it's organised, but certain people do seem to crop up supporting the same images. That said Christian and I quite often support or oppose the same images, and that's not due to coordination. I guess the real question is what can we do about it? -mattbuck (Talk) 11:46, 14 February 2015 (UTC)[reply]
    After all it is democracy and apparently it is the majority that decides what can be called QI here, it's a pity for the label because it is not hard to find in the archives, QIs promoted with many defects (CAs, dustspot, OE...). The only thing to do is to try to double check the votes of the easy/faster promoters, and to continually look what happens in CR for to prevent the images with defects are promoted and also to fight against the wicked/revenge votes. If democracy works in one direction it also works in the other...-- ChristianFerrer 20:03, 14 February 2015 (UTC)[reply]
    Exactly. --Code (talk) 20:31, 14 February 2015 (UTC)[reply]
    Unfortunatley democraticity is not an infallibe way to objective truth and sometimes leads to embarassing and ridiculous outcomes. A very good example was the Indiana Pi Bill (here). Less interesting ones are the many promoted QI with obvious quality flaws as well as those declined on the basis of the incompetence of some laborious reviewers. Alvesgaspar (talk) 22:19, 14 February 2015 (UTC)[reply]
    Well, we will have to live with that. There is no such thing as "objective truth". That's one of the great advantages of democracy. --Code (talk) 05:24, 15 February 2015 (UTC)[reply]
  • I agree too. Quality is the goal. I'd learned a lot, especially if one of my images was declined. All the photography communities have a kind of "like" but not a meaningful criticism. I can't improve the quality of my images with "like" and "dislike". (Wikimedia Commons isn't a kind of social media.) So the only way is too fulfill the rules of QI, even if there is sometimes a blind eye. (Hopefully a good translation of "ein Auge zudrücken".)--XRay talk 16:36, 12 February 2015 (UTC)[reply]
  • QI is a fast process its meant to be a simple process, looking through galleries and saying you can find images that dont meet the requirements or standards now being imposed is an expected outcome as QI progresses through time and technology improves both in the camera and our viewing capacities. When QI started the majority of people where looking at images on a 17" crt monitor now we have lcd,led and plasma screens in sizes from 7inch to 70inch, though I would hope people werent reviewing QI images on the smaller screens. With QI people not only learn how to improve their photographs but they also learn how to review the works of others with a one person review system its going to higher percentage of failures in the process, unless its such that a significant majority are passing/failing because of the result of multiple reviewers failings then lets keep it simple. The last thing anyone wants is QI becoming like QA on en where you start talking months to get a review finalised Gnangarra 04:59, 15 February 2015 (UTC)[reply]
  • When I add something to a review I have seen the image at 100% in Chrome or up to 200% in XnView. Combined with a laptop vintage 2011 and eyes vintage 1958 you can trust that I miss lots of issues—but not the validation of SVGs with a quick glance in the source, mostly because I need the tool info for {{igen}}.:-) –Be..anyone (talk) 10:52, 15 February 2015 (UTC)[reply]
  • May I add an aspect to the debate: According to the Grant Proposal of WMAT one objective is Increasing the number of decorations (valued, quality, featured) of images supported by Wikimedia Österreich by 500; this applies regardless of the creation and upload date of the images (and I assume, other chapters may have similar objectives). Defining this as a metric to be rewarded by future grants, always seemed a quite bad idea to me. It creates a kind of coupling between the successful nomination here and financial sustainability of a chapter. It mixes quality with quantity aspects. It could be a motivation to try to promote harder. Given that objective, trying to successfully promote is fully legitimate. --Herzi Pinki (talk) 21:20, 15 February 2015 (UTC)[reply]
  • Wow, everything is explained now! I understand the attempts passages strengths and aggressive responses to scare off potential opponents. For those who love this label we have to pay attention to the quality of images promoted and to the CR section. -- ChristianFerrer 23:10, 15 February 2015 (UTC)[reply]
    • @Christian Ferrer, to state it clearly: the WMF wants metrics that can be evaluated, they give money for metrics met (not only, but this is my strong impression, also after some discussions) and the world begins to spin. You should not blame individual uploaders. Nevertheless it could impose a threat on the process. --Herzi Pinki (talk) 23:58, 15 February 2015 (UTC)[reply]
I agree that this is a problematic conflict of interest, especially given the fact that it is more or less undisclosed. I don't want to accuse anybody of wrongdoing, but by definition such targets can easily lead to devaluation of any neutral process, since they incentivize lower standards. In any case such thresholds are quite ridiculous since they are most easily reached by taking many photos of similar subjects (just changing angles etc.). There are a few nice examples of this to be found in some cats (1, 2, 3, 4). I am not saying that such alternatives of essentially the same image have no value for documenting purposes and I am not sure it makes QI weaker (fair enough for me if all images are good enough by the guidelines (by the way, there is a healthy amount of CA on some of these images, but they were promoted by third parties)) but it shows how absurd fixed QI targets are. --DXR (talk) 08:07, 16 February 2015 (UTC)[reply]
  •  Comment -- I share @Mattbuck: 's concern over the two contradictory groups of reviewers: those applying technical criteria and those apparently trying to facilitate the promotion process through a more or less "like-it" approach. Although both approaches are theoretically respectable, we should recall that the outcome of the discussion we had recently, where a maximum of 5 nominations was agreed, clearly pointed to the first one. It also concerns me that some reviewers are avoiding any kind of explanation when supporting a picture in CR. The title "Consensual Review" is eloquent enough to suggest that this is the place to discuss the merits and demerits of images, not to decide democratically on their promotion. Alvesgaspar (talk) 12:01, 17 February 2015 (UTC)[reply]
  •  Comment I've read through this whole thread. @Mattbuck: has properly summarized the core of the problem. There are questions that have not been ever been voted on, if they've even been discussed at all.
    1. "Does a higher resolution excuse unsharpness at 100% zoom?"
    2. "Is perspective distortion ever acceptable?"
    3. "Do arty photos have a place at QI?"
There are other similar questions. Those that want higher standards feel that consensus has shifted and higher standards are normal and accepted. Yet these higher standards were not voted upon to determine consensus. It's clear from the discussions that the guidelines are being ignored both by letting images with glaring issues through (i.e. downsampling) and by rejecting those by holding them to higher standards than the guidelines dictate (i.e. CAs). It's so bad that User:Cccefalon became enraged when I attempted to enforce the guidelines as written. If you want higher standards, have discussions and vote on those standards rather than try to force them on everyone. Until that is done, expect a lot of frustration. -- Ram-Man 19:41, 17 February 2015 (UTC)[reply]
  •  Comment @Ram-Man: your points are worth discussing but I don't think they are the main issue here. I understand that each one of us has slightly different ways of interpreting the guidelines and selecting a QI. What is difficult for me to accept is that QIC is used to pursue an agenda or fight some regional cruzade. Something disturbing is happening in QIC right now, please take a look into CR. -- Alvesgaspar (talk) 00:03, 18 February 2015 (UTC)[reply]
    • I can say for sure, but I think you are just being trolled. And it's hardly "slightly different ways of interpreting the guidelines" when there are clear instances of disregarding what the guidelines actually say (as opposed to shades of emphasis). -- Ram-Man 02:17, 18 February 2015 (UTC)[reply]

Please stop to initiate your own rules, Ram-Man

Dear Ram-Man, if you don't like the rules in QIC, then initiate a discussion to change them. But don't implement new rules by facts, like moving files untimely to CR [3]. --CEphoto, Uwe Aranas (talk) 14:50, 17 February 2015 (UTC)[reply]

Really? -- Ram-Man 15:48, 17 February 2015 (UTC)[reply]
As far as I understood the procedure for the last years, it is the task for the bot to move the images for CR. Beside that, I apologize for reacting impulsive, but your insinuation, that CA can be excused in an otherwise good image, enraged me. --CEphoto, Uwe Aranas (talk) 15:59, 17 February 2015 (UTC)[reply]
  1. Keep in mind both AGF and IAR. Even had there not been a rule supporting my action, procedures exist to serve a greater purpose, not to exist for their own merit. We use a bot because is convenient, but the notion that bots and process are more important than editors and common sense? I will never support that. I did nothing that would violate the purpose of QI whether it be explicitly stated or not.
  2. The issue of CAs is complicated and review is perhaps not a place for a very long explanation. There are a couple things at work here. (1) There are lots of complaints about CAs, but very few reviewers are willing to correct the issues. I've read reviews like "CAs are easily fixed", but no one bothers to do it. We've already disincentivized this since you lose your vote if you do fix them. For me, I don't have any software that can fix these and this is probably true for many others. Since I don't feel that CAs are that major (see point #2), I don't think there is reason to oppose for these reasons. (2) Images like this have lots of CAs, but does that make the image bad or not useful? CAs are mostly caused by less than perfect optics. And I refuse to require high end SLRs with high end optics as the only threshold for QI. For pixel peepers, CAs are obvious and distracting, but they are not for most people. If the CAs do not actively detract from the image, then it can be a QI, especially if there are no other serious defects. In this photo, the CAs are visible, but not distracting at normal magnifications. I suspect that most people wouldn't even see them if they were not looking for them. (3)The rules may list sharpness and CAs as both criteria for quality, but a blurry, unsharp photo is far worse than a tack sharp picture with a few CAs. Not all criteria are equally important when evaluating for a QI. -- Ram-Man 16:27, 17 February 2015 (UTC)[reply]
  • Sorry, but the CA are definitively visible even at 800px in the image that is the original subject of this debate. CAs are most efficiently removed by the owner, using a fitting profile, possible with most image editing tools (such as the free GIMP) or alternatively by almost any modern camera in JPEG. Based on JPEGs, the removal is much worse since there is more guesswork involved concerning the lens and e.g. LR does not have many JPEG profiles. Not removing CA is a clear reason for declining an image in my book, also according to the image guidelines. Don't get me wrong, but basically this argument looks a lot like: "I can't remove CA, so it shouldn't be considered a major issue". --DXR (talk) 16:38, 17 February 2015 (UTC)[reply]
    • The image guidelines list the CA evaluation guidelines as "reasonable". It is not a strict rule. There is no requirement that an image contains no CAs or that they be invisible at small magnifications, only that they be at a reasonable level. What constitutes reasonable is up for interpretation. Could I fix the issues? Maybe. But I'm not going to spend time fixing CAs that ultimately don't matter. The argument is "CAs are not a major issue, so I shouldn't have to fix them". -- Ram-Man 16:42, 17 February 2015 (UTC)[reply]
  • You can do what you want. But if I am not completely mistaken, the general consensus of those that actually bother to check an image seriously here is that fixable CAs ought to be removed if an image is promoted. Nobody can double-check all reviewers, but this approach is going to mean that many of your images are going to have a tough time in CR. The problem is that you have basically decided on your own what matters or not. Anybody can say: Who cares about noise, it's only visible at 1:1. Or: Who cares about straight lines? If your argumentation is strong enough to change the consensus on reviewing, so be it, but I guess it is unlikely. --DXR (talk) 16:52, 17 February 2015 (UTC)[reply]
  • I don't know many photographers. People who are not themselves photographers never notice the things like noise at 1:1 or CAs on objects in a scene that are on the fringes. I've had my work published in many different media: color print magazines, books, websites, etc. and these are not issues that have ever come up. Virtually all my published work is 6MP or smaller. The reason my recent set of photos are of lower standards is because I am using a different camera with lens prone to CAs and corner softness. This is not all that different from those users who don't have or want to use an SLR. What's the point of a 2MP minimum if you require equipment that far exceeds this threshold? If you want the standard set so that you require SLRs with quality optics, then boost the megapixel standard to 8MP or higher. Or add a "Gold Seal" for those QIs that reach technical perfection, the "best of the best". I maintain that QI is intended to highlight the good from the bad. By focusing on things that don't really matter to our end users, we do not accomplish our goals. Everyone decides on their own what's important and what is not. I'm not unique in this. We should require the best that we can get (e.g. non-downsampled images) without being unreasonable (e.g. noise free at 1:1). -- Ram-Man 17:05, 17 February 2015 (UTC)[reply]
  • Well, it might sound funny, but we don't disagree that much as it might seem. I agree that we should make an effort to distinguish good and bad, but I'd rather have a simpler system below QI that doesn't require all the tamtam. I see your point: Online publication is not very demanding on the technicals and 90% of people are happy with an iPhone shot. Yet people like me go to great lengths to make high-res panoramas. Perhaps I'm naive, but I'd like my images to be printable as a poster and still look good and that is why I cringe seeing problems unresolved that could be resolved. --DXR (talk) 17:15, 17 February 2015 (UTC)[reply]
  • And when I'm shooting macro shots, I take my Nikon D7000, macro lens, and a tripod and take razor sharp nearly perfect images. But I can't always do that. Sometimes I'm traveling with the family in a low light museum and I take my light weight Olympus E-M5 with its exceptional 5-axis stabilization and can take blur-free shots hand held at 1/6 s. The cost is higher CAs, but the images themselves are otherwise very good. QI should be flexible enough to allow for multiple types of photography/photographers. CAs are different in that they have little to do with the photographer and more to do with the equipment. I think we should allow for equipment variation and focus on things that photographers have more direct control over, such as exposure, noise, sharpness, etc. -- Ram-Man 17:24, 17 February 2015 (UTC)[reply]
  • Well. So many lenses have a lot of CA, even very expensive ones. I know that many people are not fans of spending money on PC programs, but frankly I think it is hard to spend 140$ as efficiently on any photo gear as for Lightroom, which will take care of any issue of that kind instantaneously and easily lasts several years. That aside, do you shoot raw? If you don't, I'm almost certain that the camera can take care of such issues on its own. --DXR (talk) 17:32, 17 February 2015 (UTC)[reply]
  • I don't shoot raw because I don't have the time to add another step into workflow. I have an image backlog going back many years. Alas, the EM-5 does not have in-camera CA reduction. They added that in the EM-1. I'm forced with two choices: Use different equipment or use different software. When the image guidelines state that CAs should be "reasonable", I don't think either of those two options are a "reasonable" way to require dealing with the issue. -- Ram-Man 17:39, 17 February 2015 (UTC)[reply]
An example of unreasonable CAs
  • The guidelines for CA state that it must be distracting and be unreasonable. This standard is not clearly applied to CA evaluations. Non-distracting CAs are often flagged and it's relatively unusual to see CAs that are truly unreasonable with the quality of modern equipment. Why are CAs overemphasized? Is it because quality of submitted images is so high that we've resorted to being nit-picky? There is nothing in the guidelines that ease of fixing CAs is any justification at all for requiring it. -- Ram-Man 18:26, 17 February 2015 (UTC)[reply]
    • My two centavos: AGF and IAR are english wikipedia concepts. Here we are on Commons. Maybe AGF can apply, but IAR for sure not. Second: in february 2015, CA are always distracting and unreasonable, because they are often avoidable, and always correctible. A picture with CA cannot be a QI. Thank you.--Jebulon (talk) 18:51, 17 February 2015 (UTC)[reply]
      • Always?! This is clearly and obviously not the intention of the guidelines. This goes way beyond higher standards. If there are legitimate complaints that some reviewers are too soft, it is positions like this that go way to far in the other extreme. -- Ram-Man 19:21, 17 February 2015 (UTC)[reply]
        • Your argument has some merit. But imho it is unrealistic to believe that the guidelines are the only thing that is used in practice. There is precedent that is largely accepted, yet not formally written down. I strongly believe that there would be consensus to affirm a statement like "CA that can be removed using a modern image retouching program should be absent in an QI", after all nobody has made much fuss about the CA issue before you, who finds himself in a somewhat unlucky position on the technical front. Most people have accepted this and have found some way to remove CA and to arguably improve the quality of their images. --DXR (talk) 19:54, 17 February 2015 (UTC)[reply]
          • If not CAs, then something else. Unwritten rules are not helpful to inexperienced participants. It creates a hostile environment. The burden is not on me to back down, nor should we write off those people who just accepted rejections to avoid a fight or simply stopped participating in QI. If there is consensus a simple vote and explicit rule change is all we need. -- Ram-Man 20:16, 17 February 2015 (UTC)[reply]

Happy Chinese New Year 2015

"Xin Nian Kuai Le" or "Guo Nian Hao"
... and might members of QIC behave like lambs in the New Chinese Year.
Cheers from Singapore, --CEphoto, Uwe Aranas (talk) 16:30, 18 February 2015 (UTC)[reply]

Same to you and everybody else! Have fun! Also nice to see an exif date in the future --DXR (talk) 16:32, 18 February 2015 (UTC)[reply]

Rule clarification concerning CA, etc.

As you can see above, the fact that many reviewers require CA correction for QI promotion has been challenged. While I think that the common interpretation of "distracting CA" is generally quite strict here and backed by a large majority of reviewers, it has been noted that the image guidelines are not sufficiently clear on the actual meaning of this term. The proposed alternative interpretation (which is quite generous with respect to CA, effectively arguing that no modern DSLR lens creates CA that is so severe that it should be a obstacle to promotion) would in my opinion lead to a clear decrease in the required quality of images, harming the goals of QI. By definition, QI is concerned with technical quality and removing one major aspect of technical quality would in my opinion question the nature of QI as a whole. I support the stricter interpretation of the rule because I think:

  1. QI is a project that attests technical quality.
  2. The presence of CA significantly decreases technical quality.
  3. An image without CA is technically superior to one with CA, ceteris paribus.
  4. The creator of the original file is best placed to correct such issues.
  5. It is not an unreasonable burden on the creator to require CA correction for promotion.

Therefore, I suggest to add the following line to the QI rules (or alternatively the image guidelines):

Quality Images should be free of noticeable and correctable lens defects (including chromatic aberrations, purple fringing and geometric distortion). Geometric distortion is acceptable for fish-eye lenses.

Feel free to comment or suggest alternative formulations. Once we are reasonable happy with the phrase, we should vote on it. --DXR (talk) 08:45, 18 February 2015 (UTC)[reply]

  • In my opinion geometric distortion should be out of the bag. Quite often it is not a good idea to fully correct the non-vertical lines of buildings even when the pictures are not taken with fish-eye lenses. Furthermore many people (including myself) don't have access to the expensive software needed to fully correct chromatic aberration and specific lens distortions. Alvesgaspar (talk) 09:45, 18 February 2015 (UTC)[reply]
    • As a clarification: By geometric distortion I mean barrel distortion and pincushion. This can be corrected with GIMP, perhaps not perfectly but at least adequately. Since this paragraph would only be about lens issues, I wouldn't want to address perspective distortion issues at all here. --DXR (talk) 10:02, 18 February 2015 (UTC)[reply]
    • I don't think being completely "free of noticeable ... defects" is necessary for a high quality image. It is customary to ask nicely if the creator can fix a minor correctable defect, but really if it is only noticeable at 100% on a 24/36MP image then we are fretting about unnecessary things and will only encourage radical downsizing. This year all the major camera manufacturers will launch 50MP cameras, meaning that no 100% image will be perfect unless one uses a Zeiss Otus :-). And I agree the "geometric distortion" isn't specific enough. Some ultra wide angle lenses produce moustache shaped distortion that cannot be easily corrected but may not be very noticeable unless one is picky. Another comment: QA should (imo) not be wholly about technical quality. It may not require (or indeed be saved by) the "wow" requirement of FP, but it should still tick some artistic and creative boxes, making good choice of lighting, subject arrangement, framing, etc. And both quality forums seem over concerned with the small things. Wrt Alvesgaspar's comment on expensive software, well I don't think Lightroom is expensive compared to camera equipment but I don't also want to see so high a bar set on these minor things that only wealthy European photographers can participate. -- Colin (talk) 10:45, 18 February 2015 (UTC)[reply]
      • A few more words: Firstly, I would like to stay on the topic of "correctable lens defects" here. I agree with you on compo etc., but that is not the point of this debate. Generally speaking, writing rules for this forum is near impossible because there is the innate conflict between precision and leniency. I don't want to scrutinize on Otus-level. But almost every lens from a 18-55 for 100€ to a 2000€ lens has CA and often distortion that could be removed. Most cameras after 2008ish do it automatically in JPEG, most decent programs can also do it. How are we going to set an appropriate level of scrutiny? Scale everything to 6MP? Again, I'd be very happy to hear alternative formulations that you prefer. If nothing emerges out of this, I will continue to oppose images with significant CA, especially if it's blatantly obvious even in 1080p. --DXR (talk) 11:07, 18 February 2015 (UTC)[reply]
      • (Edit conflict) Colin, you're raising a point that has been bugging me for quite a while now. Composition and lighting are essential in photography. These are things you notice about an image when first looking at it from afar, well before you start noticing details like focus, depth of field or CAs. But even though they are in our Commons:Image guidelines (which are the basis of both QI and FP), it seems that they are rarely considered in QI reviews unless theres something extremely disturbing about them. We are nitpicking about smallest amounts of CA visible only at high magnifications, while regularly promoting images with random compositions at snapshot level. I'm not gonna point fingers here, but I can find some examples in my own uploads if requested. QI don't have to have "WOW", but they shouldn't be en:Meh either. --El Grafo (talk) 11:20, 18 February 2015 (UTC)[reply]
        • Yes, I'm not keen to have specific rigid rules about e.g. CA that just encourages technical-only reviews. Really, if the only thing stopping a great image being recognised at QI is a little bit of purple fringing in the tree leaves in the edge of the frame, then we have our priorities wrong. And, btw, minor CA can be successfully removed by Lightroom but significant CA just requires Lightroom to cover it up in grey, which might even look worse or contaminate other genuinely purple features. The level of permissible minor defects is always going to be a judgement-call, and requiring that none be noticeable is too far. I agree with El Grafo, I'm much rather see stronger reviews of genuine photographic and artistic concerns, than failing images for pixel peeping. -- Colin (talk) 11:58, 18 February 2015 (UTC)[reply]
        • A few months ago I opposed a large number of images on compositional grounds only and got a huge backlash. So I backed off a little, but the priorities are reversed. -- Ram-Man 13:27, 18 February 2015 (UTC)[reply]
          • This is a problem with opposing/supporting on more subjective things: people will argue. Doesn't mean they aren't important. Lots of photographer's and reviewers here aren't confident enough in their artistic judgement to make such reviews, but know how to spot some purple fringing or noise. Thus, especially on QI where reviews are brief and often unchallenged, they tend towards easy options. If QI exists for the benefit of Commons' photographers then it needs to help them make better photographs by full critical review and to learn from each other what makes a great image. Nobody became a better photographer by learning where the "Remove CA" check box was in Lightroom. -- Colin (talk) 16:05, 18 February 2015 (UTC)[reply]
            • There's more to it. If I decline an image because of CA, it's basically the fault of the lens and the uploader just didn't notice. If it's out of focus, the autofocus missed - shit happens. Perspective distortion? Probably correctable, no biggie. But if I decline because of composition, I'm basically saying "YOU screwed it up, and there's no way to fix it." So declining because of another reason doesn't only require less confidence in my own judgement, it also feels less cruel. The result being, that I don't review some images and later shake my head when someone else reviews them with a default "Good quality". I'm trying to overcome this, but it's not easy. I think a better interface with more space for more detailed reviews like on VIC could help the project much more than any new rule. --El Grafo (talk) 17:37, 18 February 2015 (UTC)[reply]
  • A photographer directly controls the most important things: composition, lighting, exposure, ISO, focus/sharpness, depth of field, and resolution. Software manipulations (sharpness, noise reduction, cloning, curves/levels) are for fixing those mistakes. The photographer should position themselves and choose a focal length to limit perspective issues, but equipment issues (CAs, lens distortion, noise at 100%), while often correctable in software, are not about taking good photos and cannot be fixed in the camera by the photographer. Nor should we require high end camera equipment for a QI. Does this mean we should ignore equipment issues? No, but they should never be a review focus and they should never be a required fix unless they rise to the level of being distracting. Leave guidelines as they are and just follow them. -- Ram-Man 13:24, 18 February 2015 (UTC)[reply]
Obviously there is no middle ground between our opinions. That's why I actually wanted to hear the opinion of others. If something is visible without magnifying the image, it is distracting to me. That is often the case with CA or bad distortion but not with sharpness or noise. Distracting or reasonable or other words of that kind are wobbly, and probably by design unclear. Distracting is a word that means many things to many people. If we accept distracting as a measurement, images will end up in CR and a majority vote will decide on them. I can't say that I know what is distracting to you, but the same holds vice versa. I simply would like to know whether there is a majority for being strict as suggested by me, lenient as suggested by you or whether everything will remain in that unclear state of reasonableness. I am worried that your stance will lead people who have previously corrected CA to think: "well he doesn't care about CA, so why should I?". This will not give us better composition or better light or anything, just technically worse images. --DXR (talk) 15:56, 18 February 2015 (UTC)[reply]
You can still ask then creator to remove the CA please. Most of us are keen to fix small issues where possible. If someone routinely supplies image at QA that lack basic post-processing such as a degree of NR, CA-removal, horizon straightening, then the solution is the community to have a word with them rather than to try to prevent such laziness through rule-making. -- Colin (talk) 16:09, 18 February 2015 (UTC)[reply]
Okay, so what is your opinion about the current underlying case, where the user is aware of the issue but says that it doesn't matter enough to make the effort to fix it (I guess that is a fair representation). Would you support such an image, would you abstain? I mean, I can't rule out that I'm too harsh, I just always felt that CA removal was generally desired and eventually required by most. --DXR (talk) 16:31, 18 February 2015 (UTC)[reply]
Well, since you asked: I think personally I'd prefer the current state of wobbliness. If something is disturbing, someone declines it and nobody disagrees, it's probably truly disturbing. Otherwise it's obviously not an obvious case and goes to CR for a majority vote. We should have more of those anyway, imho. --El Grafo (talk) 17:14, 18 February 2015 (UTC)[reply]
  • I think the wording is reasonable, though issues with scale do arise. I for one review images at 100% zoom, and do not go closer unless I want to confirm something I noticed at 100%. (I do also zoom out and look for compositional / lighting / etc issues as well, so please don't misinterpret what I'm saying). I hold the view that unless you are on 100% zoom you are not reviewing the image in question. If I can see CA at 100% I will not promote the image, but if I don't see any CA at 100% then I'm not going to look closer to try and find some. CA and barrel distortion are eminently correctable (often by the camera itself), and having either should generally disqualify an image from QI. -mattbuck (Talk) 15:54, 19 February 2015 (UTC)[reply]
mattbuck, if you are reviewing a 36MP image at 100% on a 100DPI monitor then the image is over 6 feet wide and 4 feet high. Can you imagine studying a huge gallery print that size from 2 feet distance? You physically can't see the whole image at that perspective, even if you had a monitor the size of your wall. Nobody would look at a picture like that. They'd perhaps stand 10-12 feet away from such a monstrous print. So tell me if you can see CA at 100% when you view the image from the far side of your room (assuming it is big enough to stand that far back). And I'm gobsmacked that anyone might want to try to find CA at 200%. This is really lame technical reviewing and makes Commons look foolish. Our re-users will not care one bit about CA or noise that is only visible at 100%. Most image users are happy to buy microstock at 6MP for print or < 1 MP for web use. So do you think our users would rather we judged QI on whether the image was high quality image or whether the individual pixels looked fine? I suggest that full-sized images be typically reviewed at 6MP for such issues. -- Colin (talk) 16:28, 19 February 2015 (UTC)[reply]
I honestly don't care how massive the photo is in terms of pixels - if I review your 10,000,000,000 megapixel image at 6MP I'm not really reviewing your 10,000,000,000 megapixel image am I? Instead I'm reviewing the quality of my browser's thumbnailing software. Does this mean that if you are producing higher resolution images I'm expecting you to take a better photo? I suppose so, in percentage terms. But I think that's the wrong way to look at it. If you downsample enough, every image will be pinprick sharp, every defect invisible, the question is what size of image are you left with? To me, QI means that "this image, at full size on a decent monitor, looks excellent" - whether the image is 2MP or 200MP. It doesn't matter whether the image is used in a magazine or a 300px thumbnail on wikipedia or printed out on the side of a skyscraper - what size images may be suitable for different media is a matter for reusers. That said, if I were a skyscraper coverer, I'd like to know that if I find a 20MP image, the QI seal would mean that image was pixel-perfect (more or less). -mattbuck (Talk) 17:59, 19 February 2015 (UTC)[reply]
Well you should care how many megapixels there are. Because someone uploading a downsized 3MP image is going to pass all your quality thresholds no problems at all and the same person uploading their 36MP original would be torn apart. Please explain how that matters to our re-users? One needs at least 5MP to print A4 to high quality. I suggest that 5/6 MP be a suitable resolution for review. Certain images (birds, for example) may have mitigating factors that allow a small 2MP image to be acceptable. But worrying about the pixel-level quality above 6MP is a bonus, not essential at all. "pixel perfect" is something I'd look for when buying an expensive computer monitor, but "pixel perfect" is not something any serious photographer aspires to. What a mad game we have here. -- Colin (talk) 21:41, 19 February 2015 (UTC)[reply]

A list of things which QIs should possess

This is a list of things which, to me, define QIs.

  • The following are required for any QI:
    • Image size at least 2 megapixels
    • Work of a wikimedian
    • Good composition
      • Clear subject
      • No distracting elements
    • No more than minor unsharpness throughout the focus area
    • No dust spots
    • No pixel errors (such as "hot" or "dead" pixels)
    • No cropping beyond the image boundary
    • No disturbing chromatic aberrations at 100% zoom.
    • No posterisation
    • No disturbing dither patterns
  • The following are almost always required (but can in certain circumstances be let slide)
    • No barrel distortion (except in the case where it is a part of the composition, for instance using a fisheye lens)
    • No tilt (unless it is a compositional choice)
    • No clipping (under or overexposure, even if remapped to grey)
    • Subject well-lit (except where this is a clear compositional choice)
    • Realistic white balance
    • Subject sharp and crisp at 100% zoom (except where this is a clear compositional choice)
  • The following are generally required, but can semi-regularly be let slide:
    • No lens flare
    • No perspective distortions

-mattbuck (Talk) 17:59, 19 February 2015 (UTC)[reply]

I note only two issues that really matter: clear subject + good composition. The rest may or may not be important and most are not. There is far far more to what makes a great image and a high quality image than these things. -- Colin (talk) 21:41, 19 February 2015 (UTC)[reply]


this is a great image that is very usable
this is a poor image that would only be used in desperation

On the thumbnails right, one passed, one declined. We are declining great highly usable images on minor technical issues or pixel-peeping, and passing dull dull dull images because, well, because the camera did it's job???? Are we photographers or photocopiers? -- Colin (talk) 21:52, 19 February 2015 (UTC)[reply]

Of course you are right in principle. When I started this debate, my aim never was to defend boring images or to claim that a technically correct image of nothing is better than a great image captured with a phone. I'd like to think that most images I personally nominate here are at least successfully illustrating their subject (asking every image here to be amazing is probably missing the point) or even interesting and technically correct and I'd hope most people think before nominating, not only looking at a checklist but asking "Is this image good for something?". What you are asking for is probably never going to be achieved because it will be controversial (and controversy always appears to end up badly around here) and too many here are not capable to defend their ideas well enough linguistically and an ironic comment is going to set off another fire etc. I probably expressed it badly, but I didn't stress the ceteris paribus for nothing. A nice image is even better and more useful (at least if printed) if it is at least technically decent. If people go to amazing places and share their impressions with us and the world, I feel we should strive to have the best output out of the camera in their hands and not correcting flaws that can be corrected fairly easily without resorting to buying a new camera or lens is part of that for me. --DXR (talk) 22:27, 19 February 2015 (UTC)[reply]
  • File:Khustain Nuruu National Park (2).JPG @Colin:  Question Is this image a QI in your opinion? -- Christian Ferrer 21:56, 19 February 2015 (UTC)[reply]
    • I think it is borderline. The light on the hills and the subject are really good. Plus it is an image in Mongolia which isn't so well photographed as train stations in the UK :-). The technical quality is poor however, even downsized to 5MP. It is an image that could be useful at small size and given the scarcity of the subject that might be enough mitigating factor. Perhaps QI rules should take that into account that images we have no shortage of might need to pass a higher bar than those that are hard to find. -- Colin (talk) 22:07, 19 February 2015 (UTC)[reply]
      • @Colin: Is this version a QI now? @PIERRE ANDRE LECLERCQ: if you like you can revert on it as this version is on the file history. -- Christian Ferrer 22:32, 19 February 2015 (UTC)[reply]
        • Don't know. You seem to have downsized it and boosted the colours. I'm less interested in fiddling with a specific picture than why you or anyone might or might not think something is QI. What purpose does QI serve? I thought it helped our re-users find good images they can use. Images that are crap are not saved by being "pixel perfect crap". How about File:St Michael's Mount II5302 x 2982.jpg which passed QI but failed FP. Now some people might say both forums have the same quality standard, and indeed that FP has the option to allow wow to mitigate poor technical quality. I think it should have passed both forums. Clearly one can find and point out technical weaknesses if one wants to (and is armed with a check list like Matt supplied above) but that shouldn't be our purpose. To pixel peep nominations to death rather than stand back and admire or dismiss as an image overall? -- Colin (talk) 22:44, 19 February 2015 (UTC)[reply]
          I would have opposed that at QI due to the lack of fine detail in the island. However I agree it is a beautiful image, and so probably should have got FP.
          That said, I disagree, I believe that finding and pointing out technical weaknesses is exactly our purpose. We don't assess on value or on wow, we assess on technical quality. I go into each review with the assumption that the image is technically perfect, and then try and disprove that. If the flaws are minor that's fine, support. If they're not minor then the image shouldn't be QI, oppose. I can admire a photo for being beautiful, for being amazing and for being valuable, all while thinking that it is technically very flawed. -mattbuck (Talk) 23:11, 19 February 2015 (UTC)[reply]
@Colin: - I agree with you, I think the upper image is beautiful and the lower image boring. But QI is not and has never been about what the image is about - we care solely about technical quality. For "wow" we have COM:FP, for usefulness we have COM:VI, but we are at QI, and what we care about is technical merit. Poco a poco gave a presentation on QI at WLM this year, and he said that he looked back at the start of QI, and it was explicitly started as a place for images which were good quality but which wouldn't pass FP because they lacked "wow". QI was for the boring photos.
I'm perfectly honest - I do not care what I am reviewing a photo of (though I'll skip photos of diseases and suchlike) - all I care about is whether the image is excellent technically. If yes promotion, if no decline. Rinse and repeat. I like photos like the upper one more, but that doesn't excuse the fact it's quite clearly tilted. Tilt is one of the most basic things a photo needs to get right, and unless it's a compositional choice (in this case it certainly isn't) it's a very easy decline. -mattbuck (Talk) 22:56, 19 February 2015 (UTC)[reply]
  • Maybe it would be a good thing to create a new assessment and divide QI into two label : technical grade and visual quality in this way :
  • this have a visual quality
    this have a visual quality
  • this have a technical grade
    this have a technical grade
  • this have a technical grade and visual quality
    this have a technical grade and visual quality
  • It has neither visual quality nor technical grade
    It has neither visual quality nor technical grade

Of course we can discuss and not be agreed on the choice of my examples but it is just for the principle. -- Christian Ferrer 08:11, 20 February 2015 (UTC)[reply]

    • I think Christian's idea is a great one. The common image guidelines that FP and QI share mention both technical and visual aspects of what makes a good image (though they aren't very well written). I strongly disagree with Matt that QI should solely focus on technical aspects of an image. My camera can take a technically fine image nearly every time, why would I want to participate here as some lowly Quality-Assurance technician double-checking that the Nikon/Canon/Sony engineers have done their job? As QI may not have the wow that a FP should require but it still needs to be a good image. Are we assessing photography here or just camera sensors and focus optics? If QI really is only about the pixels then I wouldn't bother with it any longer.. -- Colin (talk) 08:49, 20 February 2015 (UTC)[reply]
      • Hmmm. First impression: this really sounds like an awesome idea. Would probably need a huge amount of work to make this work (technically), but on the other hand the QI interface etc. urgently needs a serious overhaul anyway. I think that's worth a thorough discussion. --El Grafo (talk) 10:17, 20 February 2015 (UTC)[reply]
  •  Comment -- I don't think it would be a good idea to split QI into two parts. All we need here is a bit of common sense and some discussion, so that we can bring to a common ground two extreme approaches: the "like-it" type one, now very much alive among an agreesive group or reviewers; and the old "pixel-peeking" one, which will not let a picture pass unless is 100% free of all dust spots. In the middle stays essential components of a good photograph - very often ignored - such as a good exposure, framing and composition. For me (and I have repeated this many times here) a QI is nothing more than a picture that we can take as an example of a good quality photograph for its subject. Alvesgaspar (talk) 11:06, 20 February 2015 (UTC)[reply]
    • Splitting QI like that seems a bad idea unless you explicitly mean split it into two separate projects.
      • @Mattbuck: yes one project with all the criteria you mentioned, the current one maybe rename in images of technical grade and another one visual quality image much more based on visual criteria. So yes, two projects. Therefore not devaluing this one but a rich addition to Commons. -- Christian Ferrer 16:18, 20 February 2015 (UTC)[reply]
        • Ok, sorry, I misread Christian's proposal. I took from it the idea of having two kinds of quality to consider and then to balance these in the final judgement. I wouldn't be keen to participate in assessing images for visual quality if the technical merits were lacking just the same as I see no point in promoting a technically adequate photograph that has no visual appeal at all. How does this serve (a) our users and (b) our photographers. Our users need visually appealing photographs of a wider variety of subjects than make people go wow, and that are technically good enough to use in a variety of purposes, and our photographers want to get review so they can improve all aspects of their photography. Now perhaps there are some here that are just happy to point a decent camera at a subject and claim to be a photographer, but that isn't my definition of the term. Neither is my definition of "image assessment" an activity concerned only with the pixels. Learning to review good images is part of learning to take them. I suspect some people here have mastered pressing the buttons and don't want to progress so don't want their images reviewed other than what Canon/Nikon/Pentax/Sony engineers have done for them. -- Colin (talk) 18:14, 20 February 2015 (UTC)[reply]
    • I don't think there can be any rapproachment between the two groups. The "I (don't) like it" crowd explicitly reject all that QI is about, and should frankly to COM:FP instead, where liking something is relevant, leaving those of us who actually care about technical quality here. That said, I also think Alvesgaspar is right that the "middle" elements are part of QI. -mattbuck (Talk) 11:41, 20 February 2015 (UTC)[reply]
      • Alvesgaspar, mattbuck, I don't think this was about "like-it" vs. "pixel peeping". I understood it as an attempt to strengthen those "middle element" essential components of a good photograph. --El Grafo (talk) 12:09, 20 February 2015 (UTC)[reply]
        The "like it" vs "pixel-peeling" (as you call it) is literally all this entire argument is about. -mattbuck (Talk) 12:20, 20 February 2015 (UTC)[reply]
        • Well I don't see any representatives of the "like it" camp in this discussion. Perhaps there are some and they never bother to even load the thumbs to see a larger image. That's equally as bad and unhelpful as those who seem to care nothing for the visual qualities of an image (as long as one can see the subject) and fret over CA, noise and other minor things. Both camps, if they exist, have to meet in the middle, why does anyone think that is impossible? And if it needs guidelines/checklist of things to consider for both visual quality and technical quality then that sounds helpful. mattbuck your view that the "like it" crowd should go to FP seems motivated simply by getting them out of your hair rather than concern for Commons -- a "like it" support is very much unwelcome at FP and I've only seen that among novice voters who need steered towards more serious reviewing. I actually care about technical quality -- and in my own images I expect very high standards. But I also hope they are interesting, useful and visually engaging photos too. A QI that only cares about technical quality will just be a dumping ground for the memory card contents of users who own a DSLR and have figured out what buttons to press. And some days it seems that way when I look at the candidate list. The monument image I thumbnailed above is a bad photograph and should not be QI. Who does it serve to promote that -- we have thousands of badly taken but technically ok images that nobody ever wants to see. If QI means anything it has to mean an image someone might actually appreciate, whether the subject is mundane or magnificent. The concert photograph is a visually good photograph with only minor technical concerns that are offset. This is the balance we as reviewers should use. A wonderful 60MP image of the Taj Mahal should not fail QI because someone spotted a little purple fringing at the corner of the frame. That's why I'm opposed to absolute requirements that QI be blemish free. They should be of a high technical standard, but perfection is not a good goal. By all means ask the photographer to try to fix the blemishes, and I'd hope they would. -- Colin (talk) 13:15, 20 February 2015 (UTC)[reply]
        • Thanks Colin, I agree to all of that. --El Grafo (talk) 14:12, 20 February 2015 (UTC)[reply]
Retrieved from "https://commons.wikimedia.org/w/index.php?title=Commons_talk:Quality_images_candidates/Archive_10&oldid=154421603"