Note

First of all, this question is mostly technical and only tangentially related to manga. However, such image annotation is mostly used by boorus, which are used mostly by manga/anime fans, so I thought it best to ask here. If this is not the correct place to post this, then I apologize. If you happen to know a better place to post this, would you mind telling me?

First Question

How do boorus annotate images so that textual data is associated with a particular region? You can see an example here. For example, if you highlight わたしのコードネームはハーモニー, a pop-up appears that provides the translation: “My codename is Harmonie.”. It’s basically the image equivalent of “softsubs” (subtitles that are not rasterized into the video stream, but instead dynamically loaded during runtime). Technical note: It’s not important that the text appears when highlighting regions. The only important part is that text is associated with a particular region of an image. The user agent will determine what to do with this information. (If you don’t understand, please don’t worry about this part.)

Note: Wikimedia Commons does this as well. You can see this here. Please ensure JavaScript is enabled. Otherwise, the annotations won’t appear.

Second Question (Very Important)

This is what I want to know the most: Is there an image file format that natively stores regional annotation like a booru? If you try to download the image in the link above, the annotations seem to disappear. I think that they are stored separately from the file. I want an image format that can store regional annotations within the file itself, so that the text and image are never separated.

I thought that PNG files could store annotations natively, but I’m not sure. Maybe JPEG XL can do it? I don’t know. But this should exist. It would be extremely convenient for manga translation (or any translation, really).

You can read further discussion here:

  • muffinsA
    2·
    4 days ago

    Welcome to the fediverse! If you want to discuss anime/manga, then you will likely want to make an account on an instance that is not lemmy.ml since they defederated from the instance hosting the most active anime and manga communities. I can’t actually directly link to that instance or the ml admins will delete my comment. This has led to the anime/manga communities on the ml instance becoming a bit of a ghost town. So, instead, I will link lemmyverse, which will allow you to search for more active anime and manga communities.

    As for the question you have posed, I suspect that there are companion data files for each image on booru’s side. This is how some photo editing programs work, they have some kind of data file stored alongside each original image that has all the metadata that the program needs. In the case of something like that wikimedia example, it would need to have image region location information, as well as different localizations for each region. The rest is done with javascript in the browser.