Also, some images will be common between languages, some won't (i.e. screenshots).

this certainly is a problem. it could be solved if we could use SVG (as those are just XML Files), but unfortunately this is not yet available widely enough.

i can imagine we could do some script that will check wether to show an svg or a normal file, so those that are able to view svg get a translated one, those that can only view a jpg will get a english one. but i don't think we should do any scripting in the howto - this should work without a webserver.

so what i think we should do is use english graphics, but then have small numbers in the graphic which we use in the text and also have a table with translations:

so imagine a picture with three buttons: -- thisist | 1. Send picture | 2. Cancel withthe | 3. think about it numbers |

if you want to send the query click on (1) --

As everything is always on the same place i think people should be able to find it if we just refer to it by number.

cheers