did you get my ‘merge request’ on gitourious.org with the Makefile changes proposal?

Saludos.
Iñaki.

I have just created a public page at mahara.org (https://mahara.org/view/view.php?id=60234) with the script and all the details on how to use it (and posted about it in the translators group’s thread “User manual and translations”).

Saludos.
Iñaki.

So you both made my day today.

Cheers,
-dajan

Thank you very much for looking more into the Gimp plugin. It sounds like a good plan. For the callouts: were it possible to add a white fill to the circle? Otherwise they won’t stand out so well on non-white background or when you need to place the callout over white and grey for example.

If you want to look into creating the Makefile so that SVG and the original layered Gimp files are converted before they make their way into the Sphinx building process, that would be fantastic because I certainly wouldn’t know where to start.

Cheers
Kristina

I’ve found a Gimp fu-script that does (almost) exactly what Shutter does with the callouts. It’s called “Mark Number Circles” (http://registry.gimp.org/node/25086). I’ve hacked it a bit to make the callouts like Shutter (the circle is not fully filled, only the circle line is stroked, plus the callout has a transparent background).

I like Gimp plus this script better than Shutter for 2 reasons:
- Shutter is only available for GNU/Linux (no Windows or Mac OSX version), while Gimp is available in other mainstream OSes (and the script is a plain text file).
- The script creates a new layer for each callout, so you can even move them once they are created (in case you missed the right target or changed your mind later). All the layers are merged when you save the file, though (unless you save the image in the Gimp native format), so you get plain old png, gif, jpg images as usual

The installation of the script is trivial (just put the file in the right directory) and the usage is super simple too.

Regarding the compilation part and Sphinx support for xcftools, I suspect Sphinx doesn’t have any support for it. But the thing is, we don’t mind We just need to make sure we have png (or gir, or jpeg, or whatever image format Sphinx needs/supports) files by the time we execute Sphinx. And this is Makefile’s job

In fact, the SVG images you mention could be converted to png (gif, jpeg, …) automatically as part of the compilation process, so you wouldn’t even need to convert them by hand every time you edit the original SVG files. In this particular case, instead of using xcftools, we would need ‘convert’ from the ImageMagick utilities.

I can take care of the Makefile modifications if think this is worth it

Saludos.
Iñaki.

Thank you very much for your suggestions. Initially, I thought to use Inkscape (or Gimp) to produce the final images, but it takes way longer than in Shutter where I just put the numbers in and they are incremented automatically. I was not drawing circles and putting numbers in them. As the translated screenshots can have different dimensions, I was not sold on it being more efficient to move these callouts around instead of quickly creating them again.

If you know of an easy way of doing these callouts in Inkscape / Gimp, I’d be happy to revisit this for the 1.6 manual. As for the compilation part: Do you know if Sphinx supports xcftools?

I also don’t put any text in the screenshots, just the numbers to avoid too much information. In addition: All important text is displayed as text and not in an image for accessibility reasons.

Some images, like the artefact box are available in SVG format for easy adoption in Inkscape. They are in the images_originals folder in git.

This way one just needs to capture the localized screenshot and paste it into Gimp as a new layer in the existing “master” screenshot image. Then we would just need to move the pasted layer to the right place in the layer stack (so the callout layers are placed “above” the screenshot layer) and align thigs a bit. If any of the callouts needed to be translated, selecting the right layer and editing the text labeld would do it.

Given that Gimp is available in Linux, Windows and OSX, this shouldn’t be a blocker for translators

Later, when “compiling” the manual, we could use xcf2png (part of the xcftools package in Debian/Ubuntu) to create the final .png files from the .xcf originals.

Saludos.
Iñaki.