Wikipedia:Caption

A caption, also known as a cutline, is text that appears below an image. Most captions draw attention to something in the image that is not obvious, such as its relevance to the text. Captions can consist of a few words of description, or several sentences. Writing good captions can be difficult, and the examples below may be helpful. Along with the title, the lead, and section headings, captions are the most commonly read words in an article, so they should be succinct and informative.

There are several criteria for a good caption. A good caption

Different people read articles in different ways. Some people start at the top and read each word until the end. Others read the first paragraph and scan through for other interesting information, looking especially at pictures and captions. Those readers, even if the information is adjacent in the text, will not find it unless it is in the caption. However, it is best not to tell the whole story in the caption, but use the caption to make the reader curious about the subject.

Another way of approaching the job: imagine you're giving a lecture based on the encyclopedia article, and you are using the image to illustrate the lecture. What would you say while attention is on the image? What do you want your audience to notice in the image, and why? Corollary: if you have nothing to say about it, then the image probably does not belong in the article.

Photographs and other graphics need not have captions if they are "self-captioning" images (such as reproductions of album or book covers) or when they are unambiguous depictions of the subject of the article. In a biography article no caption is necessary for a portrait of the subject pictured alone; but one might be used to give the year, the subject's age, or other circumstances of the portrait along with the name of the subject.

"Succinctness" means using no superfluous or needless words. It is not the same as "brevity", which means using a relatively small number of words. Succinct captions have more power than verbose ones. More than three lines of text in a caption may be distracting; sometimes increasing the pixel width of the image brings better visual balance, or further information can be provided in the article text. And remember that readers wanting full detail can click through to the image description page.

Because non-visual media imparts no visual information regarding the content of its file, it is often desirable to include a longer description than is typically acceptable with image captions. As with image captions, care should be taken to include enough relevant information in-line so the media file's relevance to the article is made explicit irrespective of the caption. As a general rule, retain broader points in the article body, including specific points in the media file's description field. For example, the statement: "'Yesterday' is one of the Beatles' best-known songs" might be more appropriate for the article body than the statement: "The string arrangement on 'Yesterday' utilises counterpoint, which complements McCartney's vocals by reinforcing the tonic", which might be more appropriate as an ogg file description, especially if the text pertains to the contents of the media file or supports its fair-use rationale.

...
Wikipedia