Commons talk:Featured picture candidates

From Wikimedia Commons, the free media repository
Jump to: navigation, search

Cutoff for panoramas[edit]

When categorizing FPs, we treat panoramas as special due to technical limitations of displaying each image; we dump all wide images into Commons:Featured pictures/Places/Panoramas while using regular categories for images with normal aspect ratios. My question is, how wide is wide enough to be considered a panorama? I've seen 1:2 images on either side of the divide (and even greater ratios - how is File:Île de la Cité shortly before sunrise, West View 140320 1.jpg not a panorama?), so we should really come up with a rule for what is a panorama for the purposes of categorization. -- King of ♠ 22:40, 7 June 2015 (UTC)

In my opinion, resolution is no matter about this, I think aspect ratio is wider than 2:1 and angle is wider than 180°. --Laitche (talk) 10:36, 8 June 2015 (UTC)
I thought a long time for this. Laitche give a reasonable definition, because if we start to include all assembled images we must include all Dillif's interiors church. Personaly I would not use this category, or at least I will place all the images in a second category too. Because a cityscape is a cityscape, and your recently promoted image, King of Hearts, has also clearly, more in my opinion, his place here, in the gallery for cityscape. It is more logical for who search a FP of cityscape to search in this gallery rather than in panoramas gallery. IMO all images in the panorama gallery should be also placed in a second relevant gallery. Panorama is maybe useful but not enough for to define an image. -- Christian Ferrer 17:40, 8 June 2015 (UTC)
If I would venture a guess, perhaps panoramas only exist as a category for legacy reasons. Back when ordinary categories were used, panoramas would be extremely tiny because you would be trying to fit a round peg into a square hole (where "round" here is actually "extremely wide"). Now that we use packed categories, this may no longer be necessary, since panoramas could stretch to fit the screen. I have a radical proposal: Abolish the panorama category entirely and move everything to the standard categories. -- King of ♠ 00:19, 9 June 2015 (UTC)
I mostly agree with Laitche, but I don't think it should have to be 180 degrees. For one thing, we don't always know what the angle of view is so it's an unrealistic requirement, and secondly, I think many genuine panoramas are not as wide as that. I think it's more about the aspect ratio than about the angle of view. A skyline photo from the distance such as this one would not be 180 degrees wide but what else would you call it other than a panorama? Diliff (talk) 10:47, 9 June 2015 (UTC)
I felt near things as Diliff but if a photo which aspect ratio is 2:1 and angle is 150°, is that panorama? so I chose 180°. --Laitche (talk) 11:09, 9 June 2015 (UTC)
I would say that yes, 150° is a panorama. Perhaps 90° is even a panorama if the aspect ratio is more than 2:1. The problem is that there is no actual clear definition of panorama. Even dictionaries describe a panorama as "a wide view", but wide views can be accomplished either by being a long way back, or by being close with a wide angle of view. The perspective is very different, but both are 'wide views'. So I think the only way to define it is with an aspect ratio. Diliff (talk) 12:12, 9 June 2015 (UTC)
  • +1 I rather agree, of more the BOT fail to sort the images in the unsorted section so I would stop to use this category to close the nominations, I already ask for help, but the BOT operator is too busy. -- Christian Ferrer 05:17, 9 June 2015 (UTC)
  • If the whole point of the panoramas section was because they don't fit nicely in with the packed gallery of other images, then any threshold should be based solely on the shape of the image. The angle of view is utterly irrelevant. While 2:1 is a common definition, it probably isn't elongated enough to be causing a problem in galleries. Perhaps 3:1? However, I think we should abolish the separate category. If panorama-format images don't fit into the gallery, then list them as a separate group at the end of the section (e.g. Country) they are in. -- Colin (talk) 12:04, 9 June 2015 (UTC)

Proposal[edit]

I think it's time for an official proposal. The Commons:Featured pictures/Places/Panoramas category was useful because images with too wide of an aspect ratio would not look good when forced into a standard-sized gallery. This problem has been solved ever since the introduction of packed galleries: now images can expand as much or as little as they need to to fit the page. In light of this, Places/Panoramas is an archaic category that should be removed, with the current featured pictures inside it moved to the appropriate categories. -- King of ♠ 04:52, 10 June 2015 (UTC)

  • Symbol support vote.svg Support -- King of ♠ 04:52, 10 June 2015 (UTC)
  • Symbol support vote.svg Support A very good idea your majesty. :) --Hedwig in Washington (mail?) 06:55, 10 June 2015 (UTC)
  • Pictogram voting comment.svg Comment I'm not convinced packed galleries solves the problem. It can lead to awkward sizing of adjacent images. It may still be necessary to hand-format panoramas but I think this should be done in the subject-relevant category rather than categorising based on the shape of the frame. -- Colin (talk) 07:08, 10 June 2015 (UTC)
  • Symbol support vote.svg Support The most important is that the images are in subject-relevant category. After that, "packed", "packed-hover", or "packed-hover" heights=200px galleries, have all advantages and disadvantage, especially for this kind of image, however disadvantage are not so serious, such images not so many, and it will still be possible to arrange and format sections if necessary. It is also the great advantage of galleries with respect to categories, the possibility of formatting and presentation. -- Christian Ferrer 11:22, 10 June 2015 (UTC)
  • Symbol support vote.svg Support Sounds reasonable. — Julian H. 14:33, 10 June 2015 (UTC)
  • Symbol support vote.svg Support Poco2 18:29, 10 June 2015 (UTC)
  • Symbol support vote.svg Support Daniel Case (talk) 02:47, 11 June 2015 (UTC)
  • Symbol oppose vote.svg Oppose I agree with Colin's first two phrases. I've seen in recent photo contests that even in packed galleries standard and panoramic images do not mix well. And panoramic images are more than just the aspect ratio, argument supported by the panoramic categories on standard commons. -- KlausFoehl (talk) 12:22, 11 June 2015 (UTC)
    Indeed, vertical images next to panoramas do not look good. But that can be addressed by careful arrangement of the images in each gallery. --King of ♠ 23:16, 11 June 2015 (UTC)
  • Symbol support vote.svg Support From Panoramas to Cityscapes, Natural Places or other categories. 😄 ArionEstar 😜 (talk) 00:35, 12 June 2015 (UTC)
  • Symbol oppose vote.svg Oppose a lot more work goes into both the taking and processing of panoramic images, they require/receive far more scrutiny in FPC than a single shot image so I think they deserve to be displayed in a separate gallery. Gnangarra 06:25, 12 June 2015 (UTC)
    Gnangarra you are confusing panorama with multi-stitched images. Few of Diliff's current photos are panoramas yet most of them stand up to serious scrutiny and have a wealth of detail. On the other hand, plenty landscape panoramas of 24/36MP are no more detailed than a standard portrait from a modern camera. The work that went into taking/processing an image is irrelevant to the FP category -- mostly this is unknown and under-appreciated. It seems to me the only reason we have a panorama category is because such images weren't presented nicely when mixed along with standard landscape aspect ratios. That's a problem of presentation, not categorisation. -- Colin (talk) 07:34, 12 June 2015 (UTC)
    Indeed. In fact, under this new proposal panoramas will inevitably be given more visual weight because all images in a row must have the same height. Not saying that's necessarily a good thing - in fact, it is a problem if it makes portrait orientation photos too small, but that can be addressed with a rule of thumb like "don't place a vertical image next to a panorama." -- King of ♠ 04:49, 13 June 2015 (UTC)
    @Gnangarra: I agree with Colin you are confusing panorama-like format whitch can be cropped single shots (exemple) with multi-stitched images whitch can have all formats (square-like format exemple). -- Christian Ferrer 08:55, 13 June 2015 (UTC)
    I not confusing the difference between a cropped image, and a panoramic, I m saying that panoramic images shouldbe in category of their because they are substancially different in the work involved, anyone can take an image remove a third to make it look wide where as actually taking multiple images and stitching them together requires more work an dlevel of skill if FP considers the two the same thing then the problem lies in FP not the category. IMHO works of this unique skill do deserve to be displayed separately just like animations or drawings Gnangarra 13:38, 14 June 2015 (UTC)
    I agree with you, they do take a different skill set (some would say a greater skill set) and should have their own category. On another subject though, that article has some really terrible quality contemporary panoramas. It's a bit embarrassing really. I think we can do a much better job of showing what panoramic photography is capable of. Diliff (talk) 13:45, 14 June 2015 (UTC)
    @Gnangarra: you are confusing the format with the one popular technique for generating high-resolution panoramic images. The word panoramic merely means a wide view (literally all-around-view). Image stitching is a popular method for generating panoramas but is also widely used to generate an image of higher resolution that would otherwise be possible with current affordable camera technology. For example File:Elizabeth Tower 2014-09-21 205MP.jpg is a 205MP photograph of Big Ben with a very small angle-of-view that certainly isn't panoramic yet required the same techniques and effort as many FP panoramas. Focus stacking is another skill that can create fantastic images yet we do not separate out those images. Generating stitched photographs is no longer a "unique skill" but no more rare or specialised than macro photography or bird photography, and we should equally celebrate those who have talents in those areas. Diliff's cathedral photographs take the skill to a higher level still, combining exposure-bracketing with multi-image stitching to generate high-resolution HDR. Featured Pictures celebrate our finest work and guide those using our site towards high-quality images. If you want to praise the photographer, do so at the FPC or on his or her talk page. We have other images that required a large investment in time, or experience or great skill to create so I disapprove of singling out just one particular technique for praise. Let's leave the FP categories for still photographs to focus on the subject rather than the photographer or their techniques. By all means create some other page on Commons where you celebrate stitched images. -- Colin (talk) 10:35, 15 June 2015 (UTC)
    I don't think he implied he wanted to praise any particular photographer (it was Christian Ferrer who mentioned me, not him), he just said that panoramic photography is a style that requires a different skill set. And given that it also has a different aspect ratio, it certainly makes sense to put them in a gallery/category that allows them to be viewed in a manner suitable for that aspect ratio. Just as insect macro photography and telephoto bird photography are both 'pictures of animals', they are clearly a different category of photography. Panoramics can in theory be created without stitching, but it's rare that you could generate that kind of aspect ratio and still have the resolution and detail necessary for a FP. So in practice, they are all mostly stitched images. But panoramics are a subset of stitched images, because as you say, some stitched images are not of panoramic aspect ratio. I wouldn't argue that all stitched images should have their own category, just those that have a 2:1 aspect ratio or greater. Some stitching is merely a means to an end (high resolution) rather than a desire to get a wide angle of view. Diliff (talk) 10:48, 15 June 2015 (UTC)
    Diliff, read what Gnangarra said. If you replace the words "panoramic" with "stitched" then what he says is correct. But if you don't then it is just plain wrong. Then read what King of Hearts said in his proposal. KoH is talking about images with a wide aspect ratio that are awkward to display in a gallery. This is nothing to do with techniques or scrutiny or effort involved in creation. So can we get back to the proposal rather than this complete distraction about image techniques. We don't group images at FP by photographic technique. We don't have a macro category, a high-speed flash category, a stitched image category, a long-exposure category. If someone wants to group our FPs by photographic technique, then by all means create some other pages for that. But I'd like the category of Bridges, say, to include bridges no matter what aspect ratio and no matter how many exposures or frames were used to create it. That's really all that matters. -- Colin (talk) 11:11, 15 June 2015 (UTC)
    Points taken. But in practice, FP level panoramic photography is stitched, because it if you have to use a wide angle lens and crop it to get an equivalent wide angle of view, you will lose so much resolution that it's unlikely to pass. Yes, there may well be exceptions to that rule, but you can usually assume that panoramas will be stitched. And I agree, we don't group images by technique, but we do group by subject, many of which can only realistically be captured with a specific technique. Birds can't easily be captured with a wide angle lens, insects can't easily be captured with a telephoto lens (mainly due to minimum focus distance), and panoramas can't easily be captured without stitching, so many of these categories are just as easily identified by technique as by subject. If the problem is simply that we need to display the images better in galleries then that should be pretty easy to solve. But here's an interesting question.. why do we only give our FPs one category/gallery? Why couldn't a panorama of a bridge be a bridge FP and a panorama FP? Diliff (talk) 11:48, 15 June 2015 (UTC)
    In any case, pictures can of course be in two galleries. -- Christian Ferrer 05:16, 16 June 2015 (UTC)
    They can still be in sparate galleries with different attributes accoding to the proposal (read Christian Ferrer's support comment). For each country in the subject categories you should have both a packed gallery for the photos with normal aspect ratios followed by another gallery, formatted such that it better displays the wide panoramics. -- Slaunger (talk) 05:14, 13 June 2015 (UTC)
  • Pictogram voting info.svg Info As I had warned several time, I went back to the old form of gallery because as I said several time the BOT don't work with the new form, now you have the old gallery form before I reworked it, of course I added all the recent promoted images since I touched it. This is now a totaly unuseful gallery as if you search an image, you must read 250 or 300 description. It is now exactly like a category with the disadvantages of to not have the fastCCI tool and take longer than a category to display on the screen. For who want to use this gallery for his candidate, well as you want but if nothing is done regarding the formating, virtually nobody will find your picture in this gallery as there is too much image without classification. It is totally obsolete in this form IMO. -- Christian Ferrer 11:43, 16 June 2015 (UTC)
  • Pictogram voting comment.svg Comment I think there is enough support to move ahead with this proposal, while taking into account the concerns expressed. I'm on vacation next week but when I get back (hopefully with lots of good pictures Face-smile.svg) I can work on recategorizing our panorama FPs. -- King of ♠ 02:08, 26 June 2015 (UTC)
  • Pictogram voting comment.svg Comment I already strated to copy (not to move) the panoramas to relevant galleries. Now I think Slaunger was right for to create separate galleries, but not for all panoramas IMO. Explication : the most common display resolution is 1366*768, so no images size should not exceed 1100px wide IMO. Exemple here, or further down the same page, where I created a section with "heights=140px". The purpose is to not have an horizontal scrollbar for 1366*768 screens. "heights=150px" is for this specific image and if some images need "heights=120px" or only "heights=180px", no problems, we will create specific sections for this ones too. -- Christian Ferrer 10:38, 28 June 2015 (UTC)

Discussion on formatting[edit]

  • I think whatever the outcome, I would copy the images from the panorama gallery to other relevant galleries, and I think I will use "packed-hover" instead of "packed-hover" heights=200px. Exemple by adding the German FP from natural and panorama:

Of course all suggestions are good to take. -- Christian Ferrer 05:29, 13 June 2015 (UTC)

@Slaunger:I prefer the one, for severals reasons. We can sort the images in the order they are promoted. My main point, it give a more homogeneous appearance, maybe more compact, by having all the same height, this is visualy more pleasant IMO. Admixing formats show the big diversity of our finest images. I'm now not very favorable to block the image size to 200px because some screen, including laptops and tablets may have small screen. -- Christian Ferrer 09:56, 13 June 2015 (UTC)

I would rather separate in two galleries, one with height 200 px for less than 2:1 aspect ratio images, and one with height of 140 px for the larger than 2:1 aspect ratio images, like this

-- Slaunger (talk) 06:05, 13 June 2015 (UTC)

Pixel peeping[edit]

Our Featured Picture reviewing guidelines are pretty terrible, should anyone care to read them. And the so-called complete guidelines are worse. A mix of rules, advice and educational material. They are in need of a complete rewrite and considerably shortening. However, the biggest problem is the contradiction between the stern rule against downsizing (which is often ignored) and the faults to be found only when pixel-peeping.

The worst offender seems to be "Images should not have distracting amount of noise when viewed in full size." This might have seemed fine in the day when 5MP DSLRs cost $1000 but now we routinely see images uploaded with 24 and 36MP dimensions (and this year will now doubt see some 42 and 50MP out-of-camera images). Considering that a 5MP image prints just fine at A4 page size, the additional resolution of these is therefore a bonus that allows for cropping, or detailed investigations with zoom-viewer software, etc, etc. Few of our images will be printed as posters that actually require 36MP resolution.

I think that issues such as noise/CA/sharpness should be considered at a more reasonable resolution than expecting a 24/36+MP out-of-camera image to have perfection at 100%. Many images do look great at this magnification when taken in ideal conditions. But most experienced photographers know that the 100% view exposes flaws in an otherwise just great image and they do not actually matter. Not every reviewer at FP has this understanding, many are still learning how to review, and when exposed to Flickr downsized photos, some expect everything to look perfect at 100%. Our guidelines should therefore help steer reviewers away from taking a flawed approach to review, and to learn to look at the image rather than the pixels.

For judging minor flaws in a non-stitched images, it seems a 6MP resolution is appropriate (this is 50% reduction of a 24MP image). For stitched images, a 50% downsizing is perhaps a better choice than a fixed MP.

For simplicity we can combine these into one anti-pixel-peeping rule:

When reviewing an image for noise, CA, sharpness and other small defects, do so at no higher than 6MP or 50% reduction (whichever is larger).

This does not stop reviewers from giving advice on how to take sharper pictures, remove noise, CA and moire, etc.

Comments (not votes please)? -- Colin (talk) 08:06, 12 June 2015 (UTC)

I agree with your reasoning and also with the chosen values. I think for stitched images, it would make sense to give a pixel size for the shortest edge of the frame, for example of about 2k pixels - but 50% might also work (the two are actually equivalent if 24MP frames are stitched horizontally). In very wide and narrow panoramas, this would encourage a size at which the image always has reasonable vertical resolution.
Additionally, however, I'm also not sure if the "no downsampling" rule is useful in all cases. In action shots or other scenarios with limited sharpness expectations and a lot of noise potential, raw processing software is often more successful at downsampling with simultaneous noise reduction and sharpening than a browser would be. And there is such a thing as useless resolution, if the image simply doesn't have the resolution that the sensor captures. This is becoming more frequent with 50 MP cameras on the market today and more to come. I think making this less of an absolute rule and more dependent on the situation would represent the actual voting practice that we see today. — Julian H. 09:14, 12 June 2015 (UTC)
I agree that sometimes a modest amount of downsampling can be appropriate for images that are high noise for unavoidable reasons, or images that have been stretched by perspective correction or stitching software. It's a controversial issue, though, with polarised views on this (never downsize vs take what you are given and be happy about it). So I'd like to leave any amendment to that guideline for another day.
Commons seems to like rules so it appears some legislation against pixel peeping is now required. My hope is that if we see fewer such pixel-peeping opposes, that reviews will shift from this kind of "easy picking" technical fault finding approach to one where people actually review the picture and not the pixels. Then perhaps there will be less pressure on nominators to downsize so heavily. An HDTV 1920 × 1080 is 2MP and a 4KTV 3840 × 2160 is 8MP. So there's continual movement in our media for higher resolution images and we need to encourage people to upload them.-- Colin (talk) 09:44, 12 June 2015 (UTC)
Unfortunately, I dont believe that common sense and personal judgement can be replaced (or rather regulated) by a single rule.
I always take into account the size when I review. But FP is not only about wow.
Some images have strong wow and I find it easier to overlook technical defects. Some pictures are taken under difficult conditions that cannot be avoided (it is simply impossible to use a tripod and ISO 100 for indoor sports).
Other images have less wow but, for example, high EV and then the size, level of detail and sharpness (even in larger sizes than 6 mp) can then be part of the assessment criteria. One aspect of the FP is that it should be one of our finest images. I find it in most cases hard to see (unless there is a high wow) that a 6 MP landscape, panorama or church interiors is among our finest work.
I think that the proposed rule would limit my opportunities to continue with thoughtful reviews. That said, I strongly agree with that we should urge reviewers to consider the size.--ArildV (talk) 10:26, 12 June 2015 (UTC)
ArildV I fully agree with your considered approach to reviewing outlined above. And I do wish that accepted norms (e.g. against pixel peeping) or common sense didn't have to be legislated for (ultimately an impossible task). But we have at times a combination of beginner ignorance and stubbornness that can lead to a response "There is no rule says I can't ....". So here we are.
The 6mp/50% guidance is only for reviewing minor defects. I am not proposing we review in image in-total at 6MP. Certainly we need to balance image size, technical issues, circumstances and wow. And I agree that for many situations a 6MP is totally underwhelming and not our finest work. But for others (a 6MP butterfly or bird or concert) it would be absolutely fine. I'd certainly like our guidelines to describe this more considered approach to image size than our minimum 2MP threshold does at present. However, the issue I want to address with this minor change is just that of how to review pixel-level defects. Currently, those who do not apply such an approach will support a 6MP downsized image yet would reject the same image if uploaded at 36MP. That clearly is harmful to the project.
Also, the guidelines are just that: guidelines. And like the "rule of thirds", they are helpful to beginners and new reviewers, but experienced people know when they can be broken or bent. So I'd hope that no guideline would prevent a well-argued support/oppose.
So can you consider that the proposed text does not limit your opportunity to take size into account when reviewing: my intention is in fact the opposite -- to provide a tool where people can take size into account when judging noise and such pixel defects. -- Colin (talk) 11:25, 12 June 2015 (UTC)
For example, we can appreciate that Diliff can create a 60MP cathedral image that looks sharp at 100% and that counts towards the wow and technical excellence on show. Another person's 36MP out-of-camera image will struggle to look as sharp and detailed, and that has less wow on the technical level. However, the 36MP should not be opposed just because at 100% it doesn't look as crisp and noise-free as Diliff's downsized stitched images. It may still be an absolutely fine image and among our finest work. Looking at it at 9MP (50%) is a better way to judge its noise/CA/sharpness. -- Colin (talk) 11:34, 12 June 2015 (UTC)
I think it is a good point Colin is raising about judging for CA, noise, and sharpness at a baseline and practically relevant resolution. For images less than the chosen baseline resolution, they shall be scrutinized in full resolution for CA, noise and sharpness. If larger than the baseline resolution, a template, a tool or the mediawiki software shall make it possible to provide a version of the full resolution properly resized/downsampled to the baseline resolution. Regarding downsampling, I think it is correct that moderate downsampling may be relevant at times, but downsampling below the baseline resolution should be strongly discouraged as then you need to upsample again to display on, e.g., an 8 Mpixel monitor, and that is less optimal than first downsampling, even if there is visible noise or it is not super sharp at the pixel level.
For the baseline resolution, I think it is reasonable to consider state-of-the art display on a monitor and reprints crisp on an A4-sized prinout. Colin mentioned above that 5 Mpixel is enough for A4 and a 4K TV is 3840x2160 pixels or 8 Mpixels. The latter is the toughest baseline, so I would propose to use 8 Mpixels or 2160 pixels on shortest side as the baseline resolution for review of technical defects. -- Slaunger (talk) 20:21, 12 June 2015 (UTC)
But Slaunger, it has also been pointed out that some FP candidates rely on full resolution (50-100 megapixels, for example) for wow, and at that baseline resolution, you wouldn't appreciate the image at its best and may be inclined to oppose instead of support. I don't think this rule is a suitable replacement for common sense judging. It's also quite patronising for those who can be objective in their judgements and understand perfectly well the relationship between resolution and quality and what should be expected of cameras in different photographic styles. This is just another facet of that age-old problem that Wikipedia (and Commons) have: Balancing high technical quality with inclusivity. Sometimes more of one is at the expense of the other. Also, every additional rule and procedure that we impose on FP brings it one step closer to a robotic production line "if x, then y" process with less and less room for subjective opinion. I know this rule aims to rationalise and conform judging, but I fear people may still nitpick images at full resolution even if they're not supposed to (I can't see how we can enforce it really), and people may end up simply not mentioning their reasoning to avoid being questioned about it. In any case, we can't force people to explain their votes or even strike them out in extreme situations, as has been demonstrated in recent discussions. I feel that forcing people to conform using rules is a futile effort. It might be frustrating and it might not feel like an efficient way of managing the issue, but continually discussing and educating people who seem to be judging images unfairly is probably the only way forward IMO. Diliff (talk) 21:51, 12 June 2015 (UTC)
Diliff Note I am not stating you should not review and appreciate a nomination in full resolution, but that for judging CA, sharpness and noise a baseline resolution is sufficient. But I also understand what you say about patronizing and making rules, that it tends not to work. Perhaps a nudging approach is better, where a recommended baseline resolution pic is made easily available for evaluating CA, noise and resolution, but it is not required to evaluate according to that, as it really depends on the kind of nomination you have? -- Slaunger (talk) 22:02, 12 June 2015 (UTC)
I agree that nudging is a better approach, and I'm not saying the baseline shouldn't be written down somewhere for those whose only argument consists of "But where is the rule that says that?", but yes I think it's best that we don't require it too rigidly because many of us can and do judge fairly, I think. We also have to take into consideration that while pixel peeping is a problem, we all put different emphasis on technical quality. We can't insist that anyone judges an image to our own satisfaction. We should have guidelines so that people with unrealistic expectations don't oppose every nomination that isn't 100 megapixels or razor sharp and free of any noise, artifact or whatever the case may be. But I don't see how we can enforce strict, specific rules. As I said, we can't force anyone to justify their votes, so I don't see how we can really impose rules on how to judge an image either. Diliff (talk) 22:26, 12 June 2015 (UTC)
  • Yeah we should have better guidelines. For one, I think the phrasing "amount of noise when viewed in full size" should probably be changed to something like "when viewed at a reasonable resolution", allowing the reader to draw their own conclusions for what is appropriate for the image. It should probably be expressly mentioned on the Commons:Image guidelines, that a sufficiently high resolution can sometimes overcome issues of noise and sharpness, and that higher resolution digital sensors may have more noise but are usually preferable to lower resolution ones. For example, a 3x3 group of slightly noisy pixels in a 150 megapixel image can convey much more information than a single noise-free pixel in a 17 megapixel image of the same scene that is the mean of those 3x3 pixels. As for nudging... it is good if done right, but people are notoriously stubborn on the internet and the more you try to convince them of things, the more they get defensive. Dllu (talk) 23:20, 12 June 2015 (UTC)

┌─────────────────────────────────┘
Agree with the nudging, and a link to a downsampled version for comparison is not a bad idea. Also, I think we should call out people who oppose on technical grounds when they complain about sharpness unfairly, and emphasize that what we care about is "true resolution" (i.e. how many actual 100% sharp pixels worth of information does this contain?) rather than resolution or sharpness alone. This should ideally be done by someone other than the nominator, as they may be too modest to challenge the opposers. -- King of ♠ 04:38, 13 June 2015 (UTC)

The guideline proposed is just a guideline, and only for judging pixel level flaws, not for judging pixel level wow (which Diliff's images would still gain at full 100% view). So I don't think it is correct to think this will be "patronising" to those who those who are able to avoid pixel peeping already and use their own methods. For example, we have the 2MP guideline and nobody complains this is patronising when in fact many people regularly oppose images for being too small when they are well above that (though that doesn't stop some nominators insisting their 2MP landscape is acceptable!). Certainly I think we all agree the "amount of noise when viewed in full size" phrase should be dropped as no experienced reviewer thinks that's a healthy approach. How about:

When reviewing a large image for noise, CA, sharpness and other small defects, remember that that flaws visible at 100% magnification are not likely to be important to the overall technical quality of the image when viewed or printed at typical sizes. One approach is to check for such flaws at no higher than 8MP or 50% reduction (whichever is larger). Images that are already downsized or cropped to smaller resolutions deserve closer scrutiny. Some photographic scenes are inherently higher noise or lower sharpness than others (for example, low light concert or long-distance wildlife photography compared with a studio photograph).

I know we can't "enforce" rules, but at least this sort of advice gives us somewhere to point at when we spot pixel peeping. For new reviewers, who make the mistake of complaining about a little noise in a 24MP image, we can ask them to reconsider the level of noise when downsized to 8MP say.

Slaunger, I don't think it is generally possible to use MediaWiki to downsize all images to any dimensions we want. I investigated this a while back and found that the thumbnailing software has a limit on its internal memory that stopped it being useful for generating large thumbnails. I'll look again. It might be possible to use this for downsizing 24/36MP images to 8MP, but not for creating the 50% downsize of a huge stitched panorama. It's a shame. One suggestion might be that uploaders consider uploading a "for review" resolution version and then reverting it. Then they can link by URL to that version.

And King of Hearts notes "true resolution". I agree that sometimes a stitched image or very noisy image may be offered at a larger resolution than it really has captured, and may actually benefit from a bit of downsizing. There's not a lot of point in forcing people to download a blurry noisy 100MP image when a sharp 60MP image would be better appreciated. -- Colin (talk) 09:31, 13 June 2015 (UTC)

  • Too bad that common sense can't be enforced by simply setting up new rules. But I guess you're right, Colin et al --Martin Falbisoner (talk) 11:50, 14 June 2015 (UTC)

Happy to see the rules/guidelines changed to better reflect what is important rather than imposing any new rules. Frankly, I think the original problem of the bloated reviewing guidelines and complete guidelines should be addressed and this educational and other changes made at that time. Saffron Blaze (talk) 15:48, 14 June 2015 (UTC)

  • I tried to play the game and I had rewiew this image at 50% of the full resolution but I was forced to download it and downscale it myself. However I'm not against this idea. It's a pity that the "Other resolutions" section do not work with percentage more than predefined sizes, maybe more predefined sizes could help. Colin, your idea of a "for review" resolution may be an issue for the images coming from Flickr or/and if the nominator is not the uploader, and the uloader not more active on Commons. The idea to rework the image guideline is also not bad. -- Christian Ferrer 11:31, 15 June 2015 (UTC)
    • In addition to the rescaling problems (which may not be easily solvable as someone mentioned), as I said above, it isn't fair to judge images at below their maximum resolution because some images need full resolution for their ability to wow. Some images may look 'the best' at lower resolution, some images need maximum resolution to be at their best. If you don't judge them at their best, how can you judge the image fairly? There is no 'one size fits all' approach to this, we just need to make an effort to be intelligent reviewers instead of trying to mandate specific rules or processes on how to judge. Diliff (talk) 11:45, 15 June 2015 (UTC)
      • Diliff, how many times to I have to repeat that we are not asking anyone to "judge images at below their maximum resolution". This is solely about judging whether pixel-level flaws such as noise, CA or sharpness are at a level for which an oppose is justified. There's absolutely no reason why someone can't still find wow in an image that has amazing resolution, or is wonderfully sharp and distortion/noise-free at 100% view, such as yours. We have guidelines for a reason, Diliff, and you might feel intelligent and experienced enough to not require them, but that doesn't stop them being useful for others. It seems many people here like having guidelines. Currently our guidelines state that images should be noise-free at 100%, which is something nobody in this discussion agrees with. -- Colin (talk) 11:56, 15 June 2015 (UTC)
      • The proposal is "When reviewing a large image for noise, CA, sharpness and other small defects," and "One approach is to check for such flaws at no higher than 8MP or 50% reduction (whichever is larger)". The proposal is absolutely not saying "When reviewing large images for wow, first downsize to 8MP so that all of Diliff's skill and effort is for naught." -- Colin (talk) 12:03, 15 June 2015 (UTC)
        • Colin, you might not be asking anyone to judge images below their maximum resolution but it didn't seem clear that Christian understood was doing in his 'game' above (he didn't say what he was reviewing in detail but he gave no mention of specifics so I took it to mean a general review), and I'm not convinced that others have necessarily understood what you're proposing either, but I suppose I could be wrong. Also, note that I'm not replying to you, I'm replying to Christian. The fact that you understand the reasons for it doesn't mean that he does. Diliff (talk) 13:16, 15 June 2015 (UTC)
          • David, but you weren't correcting Christian on any misunderstanding of the proposal (he's French, and I'd hope any French translation would make it as clear as the English text is should you take care read it carefully). And if you did think his comment was unclear then why don't you ask him rather than (yet again!!) take the least helpful interpretation of what someone said. All you did was repeat your complaint about reviewing downsized images. A complaint which is nothing to do with the proposal. And I can understand why people might now be misunderstanding the proposal when people like yourself keep complaining about how unfair it is to review (generally) below maximum resolution. Can we please not have another "you said" / "I said" / "he thought" argument and concentrate on the proposal. Which I suggest you read again. Now, if you think the proposal is unclear, can you suggest how to make it clear, rather than rant about something that has never actually been proposed. -- Colin (talk) 14:36, 15 June 2015 (UTC)
            • I never said I was correcting the proposal, I was discussing what he had said. My response was not to the proposal, it was to him. I'm sorry if you think it's a "he said" argument. I'd rather you stay out of it to be honest - this had nothing to do with you and you're making it much more wordy than it would have otherwise been if you'd not got involved. And please don't compress a thread that you weren't involved in. I'm not saying I think the proposal itself was particularly unclear, I'm saying that I thought Christian may have misunderstood it. I could be wrong about that, but if so, I'll take confirmation from him, not you. You can be a bit of a bully sometimes. Just let the discussion flow instead of trying to control and suppress it. This thread had nothing to do with you. Diliff (talk) 14:52, 15 June 2015 (UTC)
              • If you want to have a private discussion with Christian, go to his talk page. This is my proposal, so comments like "as I said above, it isn't fair to judge images at below their maximum resolution" which nobody, not even Christian, has suggested anyone does, are just plain unhelpful. For crying out loud, Diliff, if you think someone may have misunderstood something, why don't you ask them? -- Colin (talk) 15:20, 15 June 2015 (UTC)
                • This may be your proposal, but it is our discussion and you can't dictate to me what I can and can't say in it. I don't want to have a private discussion with Christian, I'm happy to discuss things with just about anyone with you at this point. To use your turn of phrase Colin, "give it a rest". You've blown this way out of proportion. If I'd known you were going to react like this, I may well have phrased it differently, but I honestly don't think I've said anything so outrageous to justify your outbursts. Diliff (talk) 15:35, 15 June 2015 (UTC)
  • @Diliff: I want to precise, that I don't want to rewiew all candidates with a downsampling and it is beyond question that I bothers me to downssample an image witch is very good at full resolution. The subject is just that it is a pity to miss very good pictures for defects that can easily disappear with a reasonable downsampling. Reject images with minor defaults as these pictures have a high resolution is maybe not a good idea. Indeed an image with a very big resolution can have defects and its defects disappear or become acceptable with a downsampling, of course it requires that the downscaled resolution stay reasonably, to go from 70mpx to 3mpx is not reasonably. I took for exemple this image, currently candidate with more than 73mpx, (a big relution!), however at full resolution the sky is not acceptable for a FP IMO. But I listened Colin and I downscaled the image at 50%, that's give an image with 36mpx, witch is widely a featurable resolution. However even with the dowsampling I always see the defects in the sky so I opposed. But if they had disappeared it would have been a shame to miss an image with potential. Yours images are a good exemple too, as for most of them they are downsampled and the result is just fantastic, so when the resolution allow it why not a downsampling for to see the real potential of the image? -- Christian Ferrer 18:04, 15 June 2015 (UTC)
    • Thanks for confirming, Christian. I only asked because I wanted to make sure that people were aware of what exactly we would downsample to review, and what we should not downsample to review. As you can see, Colin has now made it pretty clear that it is not about only reviewing the image at one resolution, it is only about a baseline resolution where we should look for unacceptable problems (noise, sharpness etc). I still see the process as awkward because you would have to view the image twice, once at the 'baseline' and once at full resolution. But as Colin says, we can make it available to reviewers and encourage it as a way of being fair about image faults. I can't see myself using it because as I've said, I feel that I'm realistic about what we should expect from different photographic styles. Others may not have that experience and knowledge so it would be good, I suppose, to use the baseline as an educational tool. Diliff (talk) 18:17, 15 June 2015 (UTC)
    • Exactly we agree Diliff, it is an educational tool that wanted to offer Colin, thank you to him. And its view about potential quality of an image, that the lighting and composition are more importants than 100% perfect pixels is also very interresting and goes in the same direction. In summary it's a shame to leave aside images with a big and good visual impact only because they have minor defects. -- Christian Ferrer 18:38, 15 June 2015 (UTC)

Back to the actual proposal[edit]

Ok, let's get back to the actual proposal, which is all about "noise, CA, sharpness and other small defects" rather than what resolution one should do overall judging of the image for composition, lighting, subject, and wow, etc. I think we agree that a short-to-medium term goal is to shorten and simplify our guidelines. Doing so will take time, and the Common Guidelines are shared with QI who, from what I can see, are only concerned with pixel peeping. So it may be FP needs its own set of guidelines that reflect the focus of this forum. So in the meantime, I'd like to add the following text to the FP guideline text:

Don't pixel-peep. When reviewing a large image for noise, CA, sharpness and other small defects, remember that flaws only visible at 100% magnification are not likely to be important to the overall technical quality of the image when viewed or printed at typical sizes. One approach is to check for such flaws at no higher than 8MP or 50% reduction (whichever is larger). Images that are already downsized or cropped to smaller resolutions deserve closer scrutiny. Some photographic scenes and situations are inherently higher noise or lower sharpness than others.

To those who don't like rules, remember this is just a guideline not a policy and I've written it in the form of advice rather than regulations. @ArildV, Slaunger, King of Hearts, Martin Falbisoner, Christian Ferrer:. I know some would prefer a wholesale rewrite instead, but I think that could take quite some time to write and get any agreement on. In the meantime, is there anything in the above proposed text that is objectionable? This is a wiki after all, so we can edit and edit again if required: nothing is set in stone. -- Colin (talk) 17:42, 15 June 2015 (UTC)

  • Nothing objectionable for me. It's sufficiently vague and suggestive to avoid being a one-size-fits-all mandate approach, which was one of the things I was concerned with. Diliff (talk) 18:17, 15 June 2015 (UTC)
I like the text, but I am concerned about the lead Don't pixel peep as sometimes it is a real pleasure to pixel peep to dig into the details. How about Image prevails over pixel instead? -- Slaunger (talk) 18:43, 15 June 2015 (UTC)
  • Fine with me. Maybe Don't always pixel peep is better :) You can also add this kind of sentence: "But remember, even if the technical quality of an image is very important, a picture is primarily a visual object. And the visual impact of an image, as well as the emotion it gives you, are sometimes fundamental. The visually very successful pictures should not be dismissed too quickly only because they have minor defects. It's what we call the Wow factor". -- Christian Ferrer 18:53, 15 June 2015 (UTC)
At least in English speaking forums, "pixel-peeping" is an entirely derogatory term for someone who fusses and complains about issues that only manifest themselves at 100% pixel size. So I'd be reluctant to give up on that well-known term and replace with something less catchy. Christian, on your latter point, the Complete Guidelines already say "Given sufficient “wow factor” and mitigating circumstances, a featured picture is permitted to fall short on technical quality." -- Colin (talk) 18:56, 15 June 2015 (UTC)
I know but a bigger development of text in this direction would maybe not be a bad thing, to insist more on the visual impact end goes in your direction. -- Christian Ferrer 19:10, 15 June 2015 (UTC)
Yes, but I'm anxious about expanding into more areas than necessary for this immediate issue. There are other (non-pixel-peeping) technical quality issues such as exposure, lighting, crop, etc, that can be compensated for with "wow". Most of our current guidelines are in fact beginner lessons to teach the nominator to take better photos, rather than consensus guideline on how to review an image for Featured Picture. I'd just like to get the pixel-peeping issue out of the way before then looking at a radical reform. -- Colin (talk) 19:59, 15 June 2015 (UTC)
  • Symbol support vote.svg Support -- King of ♠ 23:51, 15 June 2015 (UTC)
  • Symbol support vote.svg Support sounds reasonably vague and explicitly instructive at the same time. Absolutely fine with me. --Martin Falbisoner (talk) 06:09, 16 June 2015 (UTC)

I would suggest a different text:

Consider the size of the image when reviewing images quality. We encourage users to upload and nominate high-resolution images. Remember that small technical flaws or defects are more visible in a high-resolution image. Downsizing may hide such flaws, but also remove valuable details from the image. When reviewing a large image for noise, CA, sharpness and other small defects, remember that flaws only visible at 100% magnification are not likely to be important to the overall technical quality of the image when viewed or printed at typical sizes. One approach when reviewing a large images is to check for such flaws at a smaller size. Some photographic scenes and situations are inherently higher noise or lower sharpness than others.
(I am not a native English speaker and the text could certainly be improved)

I don't think we need to use a derogatory term like "pixel-peeping". I think that the guidelines should focus on explaining (in an educational way) why you should consider the size and the problem of comparing high-resolution images with low resolution. --ArildV (talk) 10:42, 16 June 2015 (UTC)

I'm not clear what you think is to be gained by dropping a widely known term. It is a bit like us deciding not to use the term "chromatic aberration" and inventing our own "coloured edge-lines" term that is vague. Just because it is derogatory? Well, that's the point. There is no upside to pixel peeping, like their no upside to being bigoted, say. It harms the project. The whole point is to define what over-concern with pixel-level flaws is. Sorry, but I think your proposed alternative is wordier, and vague to the point of offering no practical advice to new reviewers. It also comments on the pro/con of downsizing which I think is a separate issue and complex enough to have its own guidance (aimed more at nominators than reviewers). I think it is useful to give some guidance as to how much smaller (if at all) it is reasonable to examine such flaws. That way, reviewers can say "Wonderful composition and no significant technical flaws when viewed at 8MP" (with a link if they like).
One problem with Commons is that it rarely operates as a wiki. We get bogged down in discussion and it is too easy to offer as many permutations of text as there are comments. ArildV, do you actually have a significant problem with the proposed text: can you live with it? If so, then we can put it in and move on. It isn't permanent. A wiki works by people perfecting through editing rather than through endless talk page discussion. We seem to agree on the fundamentals so I'm keen to move onto the more important task of planning a wholesale replacement of the guidance. And I hope that can be created like a Wiki should be rather than writing 10,000 words of talk page for every sentence of guideline. -- Colin (talk) 11:50, 16 June 2015 (UTC)
If you have the choice between an educational description and the use of a derogatory term. Why not choice an educational description? The fact that you can use a term does not necessarily mean that it is the best or most efficient choice. The comparison with chromatic aberration is of course not correct, chromatic aberration is not a derogatory term.
My goal was and is to do something more than just give us somewhere to point at when we spot pixel peeping. I think that is not only more helpful but also more efficient to have a better explanation. If you throw a derogatory term to someone, chances are that the user just goes into defence mode. And I do think the pro/con of downsizing is relevant (why downsizing is not the solution).
I will not spend much time to comment on your second paragraph. I think the rhetoric is kind of cheap and I can ask exact same questions to you (Colin, do you actually have a significant problem with my proposed text: can you live with it?) and say the same thing to you (A wiki works by people perfecting through editing rather than through endless talk page discussion) and point out that I have only written two short posts above.
You have presented a proposal, I have presented a proposal. None of us need to discuss it more, we can trust the other users' judgment. Our respective views are known.
--ArildV (talk) 12:44, 16 June 2015 (UTC)
Ok, we'll see if anyone else has a view. But I must stress this is a dysfunctional approach to editing on a wiki. The word "wiki" means "quick". We have written over 6,000 words on a proposal that originally intended to add one sentence of just 25 words, none of which anyone has any strong fundamental objection to (just different preferences of approach). We now have second proposal of 90 words and a third of 119 (the latter of which doesn't even contain the specific guidance I was proposing in the first place). One wonders why our guidelines/complete-guidelines fossilised many years ago? I have experience writing guidelines for Wikipedia and it really is best just to get something half-decent onto the page where people can see it in-place and see how they can live with it. Then edit it over time should better text be offered. My fear is we next see another proposal of 200 words additionally containing someone else's pet peeve and this process never ends. -- Colin (talk) 14:39, 16 June 2015 (UTC)
Just to answer your question (do you have a significant problem with the proposed text): yes. My main purpose in this proposal was to establish a rule-of-thumb for judging these minor defects. Like any rule-of-thumb, experienced reviewers will operate without it (or use it or similar technique instinctively without being told). Merely suggesting "a smaller size" is not specific enough to help a newcomer (who may wonder if we mean 2MP, which is the minimum threshold, or screen-resolution or even the thumbnail offered in the FPC page). -- Colin (talk) 19:25, 16 June 2015 (UTC)
  • Suggestions are just suggestions, wait a few day to see if there is no relevant oppositions, add your sentence to the guideline, and feel free to take or not account the suggestions. It will always be time later to make changes if the words are not suitable. -- Christian Ferrer 17:06, 16 June 2015 (UTC)

Image viewing tool for FP[edit]

Here's a conversation I had with Dschwen a year ago. I'm re-posting here to see if anything new can come of it. -- Colin (talk) 11:52, 15 June 2015 (UTC)


I'm afraid the MediaWiki thumbnailer is still broken for very large images. But the 8MP image should be practical with the current software. However, I don't know enough about MediaWiki API or Javascript to code this as a template or similar so that you only need to specify the megapixels or percentage reduction, rather than having to get your calculator out and enter the pixel height by hand. -- Colin (talk) 11:52, 15 June 2015 (UTC)

Remember mediawiki is no longer a community project, priority things are in the background. Things change when WMF want to change, according to their interests (as the media viewer). Btw, are you already opened a phabricator bug? --The Photographer (talk) 12:03, 15 June 2015 (UTC)
I just tried registering on phabricator but get an error about cookies. I'll try again later. I think that technically, it isn't a bug in the software. There is a configuration setting that limits the amount of memory/buffer the thumbnailer can use. If the value is increased, there may be issues with cached thumbnails (I know, a 24MP image isn't really a "thumbnail" but that's what the software's main purpose is -- the little images on Wikipedia). -- Colin (talk) 12:15, 15 June 2015 (UTC)
Yes I underestand the problem --The Photographer (talk) 16:33, 15 June 2015 (UTC)