Dispatches

Dispatches: Reviewing featured picture candidates

Pioneering aviator Hélène Dutrieu in her airplane. Nomination page: Wikipedia:Featured picture candidates/Hélène Dutrieu.jpg

Featured pictures (FP) are "images that the Wikipedia community finds beautiful, stunning, impressive, and/or informative". There are currently about 1,500 Featured pictures; one featured picture is highlighted as the picture of the day on Wikipedia's Main Page.

Featured picture candidates (FPC) selects images that illustrate a subject in a compelling and informative way. FPs must be of the highest technical quality possible. Taking the adage that "a picture is worth a thousand words", featured images should illustrate a Wikipedia article in such a way as to add significantly to that article, according to the Featured picture criteria. They should either illustrate the content particularly well or possess an eye-catching quality that prompts viewers to read the accompanying article. This dispatch will explain the FP criteria so you will be able to review candidate pictures and participate in the review discussions.

Featured pictures must meet eight criteria:

1) High technical standard – Featured pictures meet a professional standard of quality, though exceptions are made for historic images. A featured photograph clearly shows the subject, without any blurring or compression artifacts. The composition does not distract the viewer from the subject. Similar standards are applied to non-photographic works; for instance a painting or engraving must be competently reproduced, and a diagram must be accurate and illustrate the subject well.
2) High resolution – While, again, exceptions are made for historically important images, a picture's resolution should be high enough to allow print reproductions. The minimum is 1,000 pixels on the smallest side.
3) Among Wikipedia's best work – They illustrate their articles unusually well, and noticeably surpass other pictures in quality and encyclopedic value.
4) Freely licensed – All Featured pictures must either be in the public domain or freely licensed. No images with a "fair use" justification are accepted. Information supporting the license must therefore be provided on the image description page. (See related stories: Reviewing free images and Reviewing non-free images.)
5) Adds value to an article – Featured pictures increase readers' understanding of one or more articles. In this process, encyclopedic value is given priority over artistic value.
6) Accurate – Featured pictures do not promote original research. They are supported by facts in their articles, references cited on the image description pages, or the reliability of their sources.
7) Good caption – Featured pictures have an informative and succinct caption, which identifies its subject and context.
8) Avoids inappropriate digital manipulation – Limited digital manipulation that aims to correct flaws is acceptable, but any alteration which causes the main subject to be misrepresented is not.

Promotion process

The Featured picture candidates (FPC) page lists the current nominations. Each nomination contains a small version of the nominated picture, the nominator's reason it should be a FP, a list of articles in which the image appears, and an acknowledgement of the original creator of the image. Reviewers discuss whether the image meets the criteria. Reviewers are expected to evaluate the highest resolution version of the image before commenting on a nomination. All objections should be accompanied by a specific rationale that, if addressed, would make the opposer support the image.

After seven days, if an image is supported by four or more reviewers and the consensus is in its favor, it can be promoted. The opinions of anonymous and new users are generally not disregarded, however they are given little weight in the final outcome. Opinions added early in the process may be discounted if they do not address concerns that arise later in the debate. Therefore, reviewers should monitor the progress of nominations on which they have voiced an opinion.

Reviewing FPCs

The Starry Night, by Vincent Van Gogh, one of our featured paintings

Reviewing images against the Featured picture criteria can seem overwhelming, but most of it is common sense, and there are more experienced people there; an individual reviewer isn't expected to spot every problem. Pay attention to others' comments on the Featured picture candidates pages. Reading other reviewers' responses to images, and looking at the problems they point out, is a good way to improve your ability to review candidates.

This is only a rough guide to reviewing FPCs, omitting some of the finer points, more subjective matters like composition, and photography jargon.

Resolution

The size of the image is one of the more hard-and-fast rules on featured pictures. On the image description page, you will see a line under the image that says something similar to "Full resolution (1,944 × 1,296 pixels, file size: 1.49 MB, MIME type: image/jpeg)", which describes the size and file type of the image. Still images should have a minimum of 1,000 pixels along their shortest edge, though larger images are preferred. Exceptions are made rarely for this. Panoramas need substantially more than 1,000 pixels in the longer dimension so sufficient details can be seen.

However, vector images in SVG format can be arbitrarily resized without losing quality, therefore the resolution criterion is inapplicable. The 1,000 pixel limit also does not apply to animations and videos, but they should be as large as practically possible. That said, each image should be judged on its own merits. Some subjects, such as paintings, demand a higher resolution than the minimum requirements. As a reviewer, you must ask: is the image resolution sufficient to show the important details?

Examples (view these at full resolution):

At just 808 pixels wide, this image of William Hogarth's Gin Lane (1751) allows almost none of the text on the signs to be read, and most of the lines that make up the engraving are too fine to be seen at this resolution.
This featured picture version, a copy from around 1806-09 by Samuel Davenport, is 4,000 pixels wide. All the text can be read, and every individual line in the engraving is visible.

Accuracy

Diagram of the V-2 rocket: FP nomination

Criterion 6—accuracy—is particularly important for nominations of diagrams. The content should be verifiable and not promulgate fringe theories or original research. The sources should be listed on the image page, in the caption, or in the part of an article where the image is explained. For example, the diagram of the V-2 rocket to the right references v2rocket.com on the image description page.

Encyclopedic value

A boy playing with a burnt lamp in the city of Birao, Central African Republic: FP nomination page

All FPs must be included in at least one article. You can see which articles link to the image in the "File links" section of the image description page. The image should contribute strongly to the article, that is, you should be able to easily justify its presence in the article. An image shown only in a gallery is not considered to have encyclopedic value (EV). An image gains more EV by contributing strongly to a single article rather than contributing weakly to many. For this reason, adding an image to numerous articles to demonstrate EV is counterproductive and may annoy both FP reviewers and article editors.

For an example of the role of this criterion in featured picture nominations, consider the image to the left. In this FPC discussion, reviewers argued that the image did not specifically illustrate a person from the Central African Republic, the article in which the image was placed. One reviewer suggested, "This boy could be anywhere and from anywhere, therefore this photo is not very encyclopedic for any place. If we could figure out what ethnic group he belongs to, we could add him to the Demographics of the Central African Republic or to one of the sub-pages. If we had an article on the destruction of Birao, we could even add it there." In the end, the image was not promoted.

Licensing and image descriptions

For help understanding the proper licensing of free images and spotting copyright violations, see this dispatch. In short, if a nominated image violates a copyright or does not provide clear licensing and copyright information, you should vote delete on the nomination, clearly stating the reason.

The image description page should also provide enough information to put the image in context. It should include what and who the image shows (in as much detail as is appropriate), when it was created, who made the image (if known), and other relevant information. A scan of a photo or drawing from a book should state which book it came from, and ideally full bibliographic information for it.

Images should also have a caption, which appears on their featured picture candidate page; however, this is just a brief description of the subject to explain to reviewers what the image is, and to make it clear that it has value to the encyclopedia.

Image quality

This FPC of a poplar hawk-moth was not promoted because part of the caterpillar was cut off.

A full description of some of the technical aspects of image quality will have to wait until a future dispatch, but an explanation of the basics should be enough to get you started reviewing.

In short, a featured picture should not have obvious problems, with some exceptions. Very old photos might be a bit more blurry because the photography equipment of the time was less capable. Likewise, a 15th century painting might have gathered some damage in the six centuries since it was painted, but we can overlook that if the painting itself is interesting or important.

This would be a great illustration for battlefield medicine, except that it is over-exposed, making the surgery itself simply a mass of white. (nomination).

Here are some general rules:

These are considered to be among Wikipedia's best photos; various aspects increase their encyclopedic value, the first one being time lapse photography which is well suited for illustrating processes far more effectively than a single static image:

Pictures also benefit from a conveniently located person or some other object to provide a sense of scale, particularly if the subject is known for being big or small.

Rhagio scolopaceus (see Rhagionidae). The water droplets pin this insect's size down to an order of magnitude (10 mm) and add wow. Nomination page: Wikipedia:Featured picture candidates/Image:Schnepfenfliege Rhagio scolopaceus2.jpg
Palais Garnier. In this case, people do the trick. Nomination page: Wikipedia:Featured picture candidates/Palais Garnier

Pictures which illustrate a particular adaptation of a species, functionality of a machine or people doing their jobs also gain encyclopedic value.

Normally the subject blending into the background is a bad thing, but this is a chameleon (Chamaeleo jacksonii). Nomination page: Wikipedia:Featured picture candidates/Jackson's Chameleon 2 edit1.jpg.
An arborist felling a eucalyptus tree. Nomination page: Wikipedia:Featured picture candidates/Arborist felling eucalyptus.

It also helps if the image is hard or impossible to replace or depicts some valuable historic moment. This includes unusual subjects and images that counter systematic bias, whether caused by copyright problems (as discussed above) or a lack of good Wikipedian photographers in the area. The importance and rareness of an image may mitigate otherwise poor quality.




Also this week:
  • Flagged Revisions
  • Mobile devices
  • News and notes
  • In the news
  • Dispatches
  • Features and admins
  • Technology report
  • Arbitration report

  • (← Previous Dispatches) Signpost archives (Next Dispatches→)

    + Add a comment

    Discuss this story

    These comments are automatically transcluded from this article's talk page. To follow comments, add the page to your watchlist. If your comment has not appeared here, you can try purging the cache.
    == Date? ==

    This wasn't discussed at WT:FCDW, and we're still working out whether there is going to be WP:FCDW/December DYK contest. We can probably run this (if its finished) on December 8 unless the DYK contest materializes. Please coordinate at WT:FCDW so we don't have these last minute surprises: it was listed at the Newsroom even though it's not yet ready. SandyGeorgia (Talk) 07:17, 5 December 2008 (UTC)[reply]

    I linked in the two earlier Dispatches that discuss free and non-free images, in case any of that information is repeated here and can be deleted. It would be helpful to explain the criteria up front since they are frequently referred to. Also, the image sizes were so large that the prose was lost. I temporarily added section editing. Done for now: I've asked other people to help. SandyGeorgia (Talk) 07:58, 5 December 2008 (UTC)[reply]
    The featured picture criteria are (implicitly) assumed knowledge and it was the first link in the article anyway. I think it's more convenient to have them in separate tabs rather than scrolling through the whole lot. MER-C 11:19, 5 December 2008 (UTC)[reply]
    I've started explaining the FP criteria. Since most editors don't participate in the FPC process, you can't assume they know the criteria and most readers are lazy. They won't click through and read the criteria carefully. Awadewit (talk) 17:20, 30 December 2008 (UTC)[reply]

    Non-photographic media

    This article contains no examples of engravings, lithographs, diagrams, paintings, documents, and all the other types of featured pictures. At a minimum, I'd throw in a Lady of Hats diagram, an engraving, and one of Durova's document restorations - possibly the Islamic calligraphy. Shoemaker's Holiday (talk) 13:30, 9 January 2009 (UTC)[reply]

    I threw in a few things to try and show the variety of FPs a bit better. IT's now more about images, less about text. It might be a little over-illustrated, actually. Shoemaker's Holiday (talk) 15:20, 9 January 2009 (UTC)[reply]

    Deleted material

    This is far, far too detailed. I'll do a brief summary instead. This should probably form part of a dispatch in its own right. Shoemaker's Holiday (talk) 17:13, 9 January 2009 (UTC)[reply]

    Yes. The entire thing was too long and overbearing, too heavy on detail. SandyGeorgia (Talk) 18:27, 9 January 2009 (UTC)[reply]

    Now comes the interesting bit—assessing the technical quality (criteria 1, 3 and 8). Use a raster image editing program such as The Gimp to assess photographs. A vector graphics program may be more appropriate for diagrams. Before you get started, you should calibrate your monitor by following the directions at "Is my monitor calibrated correctly?"

    Download the full-resolution image to your hard drive and open it up with your image editor. Now start checking for problems:

    This FPC of a poplar hawk-moth was not promoted because part of the caterpillar was cut off.
    • Use the eyedropper tool or the histogram to check for clipping, often referred to as blown highlights or blacked-out shadows. An image should show detail in both the highlights and shadows. The eyedropper tool will reveal the loss of detail in flat regions of identically colored pixels. The histogram of a clipped image will count a disproportionately large number of white or black pixels.
    • Draw horizontal or vertical lines near features that should appear straight, for example the sides of buildings, to check for tilt and distortion.
    • Is the image underexposed, overexposed, or both? This is related to clipping, and the same tools will help to identify this problem. Overexposed images will appear too bright, and tend to suffer from blown highlights. Underexposed images are, conversely, too dark, and tend to lose detail in the shadows.
    • Is the subject cut off or partially obscured for no good reason? This both reduces EV and shows poor composition.
    • Are the corners of the photo darker than the center? This may be caused by vignetting and indicates poor technical quality.
    • The subject of the image should be in sharp focus, but the background may be blurred by controlling the depth of field. Blurring the background can reduce distracting elements of the scene.
    • Does it look good? Does it stand out from ordinary photography? This is often referred to as wow factor.
    This would be a great illustration for battlefield medicine, except the detail wusses out where it matters most. The resolution is simply not high enough. Another version blows out at the same location. Nomination page: Wikipedia:Featured picture candidates/FieldSurgery.

    Zoom in so that the picture is now at 100%.

    • If the image is a JPEG file, as most photographs are, you may see squarish inconsistencies. These are compression artifacts, and none should be visible.
    • Is there sufficient detail in the picture? (criterion 2) Is the subject in focus?
    • Check for chromatic aberration. Chromatic aberration generally appears as colorful fringes around bright objects, or as a rainbow effect where the image should show only white.
    • Look for noise, most often seen in the shadows.
    • Is the image aliased? Improper processing of an image can cause odd and distracting patterns to appear.
    • If the image is a panorama, check for stitching errors. Panoramas are often stitched together from several photographs, called frames. The boundaries of the individual frames should not be visible in the panorama. Objects split between different frames must fit together perfectly, without disjoint lines or blurry "ghosting". The frames should show consistent exposure, lighting, color, and focus.
    • Check for signs of over-sharpening. Photographs should realistically depict the subject instead of inventing detail. Over-sharpening can also cause halos around bright or dark objects. Do you see any other signs of overenthusiastic post-processing?
    • Is the image upsampled? Upsampling increases the size of a picture at the expense of quality. This is done automatically in many point-and-shoot cameras when digital zoom is enabled.

    Also, check scans for several additional problems. A bad reproduction can scupper an otherwise feature-worthy image.

    • Check the image for moire patterns or other odd halftoning artifacts. Halftoning was often part of the printing process used for older pictures. These problems can appear when the image is scanned at the wrong resolution.
    • The scan should be detailed enough to read any text present in the reproduced work. (criterion 2)

    For a general idea of what we are looking for, see Commons:Image guidelines. Here are some example reviews.

    Wikipedia:Featured picture candidates/Mid-Continent Tower
    Original Comments

     

    Wikipedia:Featured picture candidates/Capitol building in Palau
    Original Comments

    This example was cropped to reduce upload size.

     

    The main problem with this picture is underexposure. Nevertheless, it was brightened in an edit, which is featured. Nomination page: Wikipedia:Featured picture candidates/New York Skyline.
     
    Here's an example of a (non-obvious) stitching error: the two sea levels of this 360° panorama differ in height by some 60 pixels. The panorama was rotated and featured. Nomination page: Wikipedia:Featured picture candidates/Panorama of Dar es Salaam.

    This one is probably not useful, and criticises an actual nominator.

    Here is an exercise nomination as would be presented on Wikipedia:Featured picture candidates. Editors entering a declaration should summarize the critique in a short paragraph, stating exactly what the problems are.


    Original - A juvenile Sulawesi Crested Macaque in its habitat at the Buffalo Zoo

    Juvenile Sulawesi Crested Macaque at the Buffalo Zoo

    Reason
    Encyclopedic and beautiful image. (Featured on Commons.)
    Articles this image appears in
    Celebes Crested Macaque, Buffalo Zoo
    Creator
    Davepape
    • Support as nominator --Dweeebis (talk) 15:43, 17 November 2008 (UTC)

    The gallery at the end seems a little gratuitous to me. Do all the images in the gallery have a reason for being there? Will they help readers understand how to review candidates? Wronkiew (talk) 07:44, 10 January 2009 (UTC)[reply]

    I think it could be cut in half; the Dispatch is still enormously long, and we should remember it's a Signpost Dispatch, not an article. SandyGeorgia (Talk) 15:31, 10 January 2009 (UTC)[reply]
    I'd also like to cut the image in the lead, as it's fighting with the Signpost banner and giving a cluttered appearance. SandyGeorgia (Talk) 15:32, 10 January 2009 (UTC)[reply]
    I removed the extra formatting from the lead image, and perhaps it's not so bad now. SandyGeorgia (Talk) 15:40, 10 January 2009 (UTC)[reply]
    If we need to cut some content to reduce the length, I think these sections add the least value to the article:
    1. What types of images can become featured pictures?
    2. The list of criteria; a link to the criteria might also serve the same purpose
    3. What makes a featured image?
    The rest of the sections are essential, though they might be compressed further. Some explanation of the technical aspects of photography is needed, for example, for readers to understand the discussions at FPC. Wronkiew (talk) 00:07, 11 January 2009 (UTC)[reply]
    I'm happy with the text so far: just think the gallery is large. SandyGeorgia (Talk) 00:58, 11 January 2009 (UTC)[reply]
    I really think at least the gallery section should be cut. It does show readers what kinds of non-photographic images they might see at FPC, but it does nothing to help them understand how to review them. Wronkiew (talk) 07:10, 11 January 2009 (UTC)[reply]
    Can someone else do the pruning? As I am not involved with images, I can't decipher which are the most essential. SandyGeorgia (Talk) 17:54, 11 January 2009 (UTC)[reply]
    I'm not suggesting we prune it. I think we should cut the gallery entirely. It is a neat collection of images, and shows the diversity of FP submissions. However, I don't think it's helpful in this dispatch, which is supposed to be about helping people review FP noms. Maybe we could save it for the hypothetical future dispatch? Wronkiew (talk) 18:13, 11 January 2009 (UTC)[reply]
    Ah, I see: I wouldn't object to cutting it entirely. SandyGeorgia (Talk) 18:16, 11 January 2009 (UTC)[reply]


    I removed the following section from the dispatch to reduce the length. Wronkiew (talk) 05:23, 12 January 2009 (UTC)[reply]

    What types of images can become featured pictures?

    Any type! While this Dispatch mainly concentrated on photography, here's a sample of some of the slightly more unusual images that have been promoted to featured picture status.

    Dispatch ready

    I have finished proofreading the dispatch, and I think it's ready to go. I'm a little uncomfortable about leaving people hanging at the end. However, from reading the other dispatches it seems they also omit a conclusion.. Wronkiew (talk) 02:19, 21 January 2009 (UTC)[reply]

    It should run in the next Signpost; thanks for the work !! SandyGeorgia (Talk) 03:19, 21 January 2009 (UTC)[reply]

    Hogarth

    The FP version of Gin Lane is not Hogarth's work from "Davenport's edition", but a later copy engraved by Davenport (probably around 1806-09 for Trusler's Hogarth Moralized). Sorry to jump in at the last minute (assuming it hasn't been used yet), but comparing the two as if they are both Hogarth originals doesn't strike me as a good idea. The first image isn't a good quality scan admittedly but it is from Hogarth's original plates. Yomanganitalk 20:44, 22 January 2009 (UTC)[reply]

    With so many editors involved, I don't know who added that, or who might correct it. SandyGeorgia (Talk) 00:31, 23 January 2009 (UTC)[reply]
    The images are illustrative but they could be deleted if necessary. If they are kept, I don't think the captions need to be ultra-precise, because what is being compared is the readability of the text in the scan, not the artistic merit. Wronkiew (talk) 01:26, 23 January 2009 (UTC)[reply]
    True but having Davenport's copy attributed to Hogarth is one of the reasons I was wary of it being promoted to FP in the first place. I've reworked the captions, what do you think? Yomanganitalk 01:48, 23 January 2009 (UTC)[reply]
    While it's important to note that it is reworked by Davenport (e.g. Davenport's edition), I don't see how what's agreed to be a pretty exact copy needs much more disambiguation than saying it's Davenport's edition. Shoemaker's Holiday (talk) 01:54, 23 January 2009 (UTC)[reply]
    However, if it's really an issue, since I rewrote this story, some Gustave Doré works have been promoted which have low-res versions on the commons. The Doré is definitely authentic, so we could use him instead. Shoemaker's Holiday (talk) 10:37, 24 January 2009 (UTC)[reply]
    This gets published today. SandyGeorgia (Talk) 14:02, 24 January 2009 (UTC)[reply]

















    Wikipedia:Wikipedia Signpost/2009-01-24/Dispatches