User:Nilfanion/Galleries

I've been somewhat disatisfied with the quality of galleries on Commons for some time. It is not really that surprising that galleries are poor quality: Commons users are more interested in uploading and categorising the content. Galleries can be very useful to end-users but only if they are maintained, and as few experienced editors have any inclination to do so, they can do more harm than good. They are normally created by new good-faith editors or spammers.

The current guidance for the galleries is not really that helpful in my opinion (Commons:Galleries). This essay is a detailed examination of galleries, and attempts to provide a structure which allows high quality galleries to be produced. I've also linked to some example galleries, one that I've constructed in accordance with my "rules" (Shaldon) and some examples of current galleries that don't and suffer as a result.

Summary edit

A brief set of rules that I've applied are:

  1. Be selective in choice of images
  2. A useful structure
  3. Useful captions, with only one language displayed. (Other languages should be made available, but not visible - See this comment)
  4. Introduction, to gallery and subsections
  5. Use big images

Be selective edit

Galleries are not categories, they should not attempt to show everything. Furthermore, most "casual" viewers (people who happen to be "looking at" Commons, as opposed to hunting for an image for a purpose) will only have a limited attention span. No gallery should ever have in excess of ~200 images, and rarely should have over 100, as the viewer will have switched off long before they get to the end. This lack of concentration will mean the last part of a lengthy gallery will be ignored - no matter how good its content.

If a subject is so broad that ~100 images will not begin to give adequate cover (for example - countries and major cities), use sub-galleries liberally. Its ok for sub-galleries to have no encyclopedic equivalent, the point is to link related media together. For instance, Night in London could be a very useful gallery.

If two images show a similar aspect of the topic, choose one. If you can't decide which is best doesn't matter which you choose - but don't include both because you can't make your mind up.

If some of the content of the category has only tangential relevance to the subject, ignore it. As an exampl, you don't need a picture of a sparrow in London to illustrate London. Likewise you don't need a picture of a sparrow in London to illustrate sparrows. However, the pigeons in Trafalgar Square may be significant enough to be included. Use common sense here...

If you fail to be selective, the gallery will be a failure. You'd be better off just pointing them to the category.

Structure edit

A gallery should have some sort of logical structure. Ordering by filename like a category does is completely arbitrary and is prone to spam. Ordering by upload date is similarly uninteresting. Instead consider the topic, and break it down into logical sub-topics. These then form natural sub-sections. By providing these subsections, users after a subset of the galleries content can find it much more easily.

Within these sections, you may be able to identify a logical order for the individual images. For example, a gallery about a river could follow it from source to sea. One about a bird could follow it from the egg to adulthood.

A good structure also allows the development of a narrative thread, something that encourages viewers to view the content right through to the end.

It should go without saying that a link to the category is needed...

Captions edit

A picture without a label has its value limited by the lack of description. It may be self-explanatory to you, but there's no guarantee anyone else will know what it is. Just giving the subject's name may be sufficient, in other cases a short sentence may be necessary. Avoid longer sentences, as then the text will dominate the picture, instead of supplementing it.

If you have two images with identical captions, its an indicator that one is redundant to the other. Either expand the captions to explain the difference between them, and why it is significant to the galleries (not the images) subject, or remove one of them.

Introduction edit

The gallery should have a reasonable introduction, ideally a short paragraph, not a single sentence in the style of a caption. If the subject is encyclopedic, its appropriate to use the lead from the Wikipedia article. The point is to give the gallery enough context that a user will know what it is about, however they got there.

Sub-sections should also have introductory elements; a single sentence will probably be enough here.

All the text in a gallery should follow Wikipedia guidance on article writing: Neutral point of view and Words to watch are of particular relevance. "Famous", "Pretty" and similar should not be used. This is both to maintain a reasonable tone, and adding fluff words like that increases the text:image ratio for no practical benefit.

There is no real need to bog the gallery down with extensive sourcing for whatever the text says, but if you copy/paste from WP do ensure you at least mention that fact in the edit summary. This is to respect the attribution requirement of the text's license.

Image size edit

Commons is a media repository, therefore people come here to look at images. They don't come here to view 120px images - which are little more than thumbnails. They certainly don't benefit from the page being 20% image and 80% whitespace, which is what most galleries here look like.

A significant portion of our views will come from Wikipedia readers looking for further images of a given subject. Wikipedia thumbs default to 220px. Its more than a bit silly for users to come from an encyclopedia to a dedicated media repository, and then find smaller images here.

Therefore images should be at least 220px wide. I've used 240px as its larger than this, and as its a simple fraction of 640px, but other options could work too. 640px is the possibly most common image width on Commons (due to Geograph).

Panoramic images should also be given a similar treatment, and the size should be chosen so that they don't dominate the other images, and aren't overwhelmed by them. In the sample gallery, the panos are the same height as the normal images which gives a good balance IMO.

What this means the default <gallery> tag should be avoided. The tag itself is very useful as it enables scaling to the user's window, but the image sizes should be increased.

Notes edit

Language support edit

The gallery should contain one language only. Commons is a multilingual project, and multiple languages should be catered for. However, simple multilingual galleries are not the way to do this.

Just consider a truly bilingual gallery. It would have a bilingual introduction, bilingual section headings and bilingual captions. This more than doubles the amount of text required - you also need to add eg

and

Français : (missing text)

headings to each caption.

This means the portion of the gallery that is media is decreased, as the space needed for text massively increased. A common theme in my criteria above is to maximise the amount of screen space given to displaying the content - this undoes all this and then some.

Furthermore, more than 50% of the extra text is irrelevant to all users. They don't care about the "other language" just the one they want.

These problems get exponentially worse with additional languages.

Multilingual support should instead be offered by different means. One logical way to do this is the same as for templates: Use subpages. The "master" gallery should be the local language and/or English as these are easier to maintain, other languages should mirror the master. This will probably be awkward to maintain.

An alternative approach is Language Selection. As it stands, IMO Language Selection is not good enough. If it worked for captions it will display "English:" headers all over the place for English-language readers, it can't do anything about section headers, and will still require the browser to load all the languages (before hiding all but what is wanted). This means there is more clutter, it will take longer to load, and isn't a perfect fix anyway. It is probably easier to maintain one gallery with 10 languages than 10 galleries with one language on each - but the time consuming step is translation. Translation will be harder on the massive wikipages that result from having dozens of captions for every image, and missing translations are more obvious on a one-language gallery.

We should cooperate with the Wikipedia's to improve the user experience further. If the referring link from fr.wp is to commons.wikimedia.org/wiki/[gallery] to commons.wikimedia.org/wiki/[gallery]?uselang=fr and that calls the fr version of the gallery (if it exists), the notional French speaker gets to the right content immediately and doesn't have to wade through any English/German/Japanese to get there.

FP, QI and VI edit

I have no real opinion on if {{FP star}}, {{QI seal}} and {{VI seal}} should be included. They don't significantly detract from gallery for someone trying to just view the thing, and can be of benefit.

However, just because an image is a FP/QI/VI and relates to the galleries subject, this does not mean it has to be in the gallery. Image quality is something to consider when selecting images, but amazing quality should not guarantee inclusion. A different image may illustrate that aspect of the subject better, even though it doesn't have the quality tag.

For each of these three labels my opinion is:

  • Valued images should generally be included, if their scope is relevant to the galleries topic. VI is all about "what is the most useful image for this scope?" as opposed to quality. The only real exception is if the VI is of a minor sub-topic of the gallery or if the VI is inferior to a different file in the same scope (and should be replaced).
  • Featured Pictures and Quality Images may be included. Images are awarded FP/QI status for being great images, not for illustrating the subject. If the FP/QI does illustrate the subject well, of course it should be included, but its being included because it illustrates the subject not because of the label.

In general if you have multiple images of the same aspect of the subject: If one is a VI, choose that one. If there is no VI, but there is a FP select that ahead of the rest (if it is a good illustration of the subject). If there is no VI or FP, select a QI (if it is a good illustration of the subject). The order VI>FP>QI is most appropriate for galleries.

This last thing is important - don't include X just because it is featured, if it is redundant to another image.

Examples edit

A few examples to illustrate my points.

Shaldon edit

I have construced Shaldon following my guidance, in the order I've written them above.

  1. Being selective [1]
  2. Structure [2]
  3. Captions [3]
  4. Introduction [4]
  5. Image size [5]

IMO the gallery gets better at each stage. A lot of content has been lost from the category (of course), but I don't think anything really important is gone.

Two images have been deliberately included for reasons other than following the above guidance - the two panoramas. I have done this to illustrate how panos can be included sensibly in a gallery (I haven't used {{Wide image}} as the panos are then too tall and overwhelm the other pics).

The choice of Shaldon is arbitrary of course, it serves as a typical English village category with a lot of content (thanks to Geograph). More important subject have far too much content for me to reasonably do a demonstration of selectivity...

London edit

London is IMO a terrible gallery. However, it is a good example to illustrate some of my above points:

  • Multilingual support. Most viewers of the gallery are probably English-speaking (from the UK!). The introduction contains support for many languages (good). This is immediately followed by a huge Table of Contents (bad). The TOC is all but unreadable, because editors are trying to provide multilingual support through section headings. This is not consistently done, as different languages are available on each header. One of the images in Tower Bridge has a bilingual captions. The fact it has two languages means a simple one sentence caption takes more space than the image it describes! The gallery would be much improved if all the non-English elements outside the introduction were removed (and replaced in an alternative way).
  • Image selection. There are over 400 images in the gallery. These include several very similar compositions of buildings - there's 8 very similar ones of Big Ben. The 2-8th images add little beyond the first. Wouldn't one good illustration of Big Ben will do for a gallery about London? The choice of buildings to illustrate is arbitrary too, and doesn't help a user to find an image of a building that isn't on the list - try to find an image of Wembley Stadium from that page, or any sports venue for that matter. If all the redundancy was removed, it would be possible to see what broad areas are being ignored, which have too much weight and so on.
  • Tone - there's unnecessary words such as "The Famous Tower Bridge"... Why bother including that word?
  • And what's the point of a miscellaneous section? No one is going to bother trawling through it, it may as well not be there.
  • Image sizes, standard complaint - what's the point of 120px wide thumbs?

United States edit

I think United States is significantly better than London, as you can at least find the media contained in the gallery. Like the London gallery, its primary audience is English speaking.

  • Bilingual support is added for Spanish-speakers. This is more consistently done than on the London gallery so is less painful. It would still be better to do it in a different means, an English speaker doesn't need the Spanish, a Spanish speaker doesn't need the English etc.
  • The galleries length is reasonable, you can view to the end without getting bored.
  • Image selection. The gallery basically consists of a bunch of seals, maps and politician photos. This seems just slightly skewed to me. There's a picture of the seal of the Government Accountability Office but nothing of any natural features. Wheres the Grand Canyon? Where's the Rockies? Do they not exist? :) And why is there a picture of the Aloha Tower in Hawaii, when there's nothing of the Statue of Liberty? How about certain events - September 11 for instance? The gallery is nowhere near comprehensive. How about dropping most of the seals and maps and replacing them with a broader range to show more of what the United States is, and what a broad selection of media we have about it?
  • A bit more text would be nice?
  • Image sizes, standard complaint - what's the point of 120px wide thumbs?