Our last week’s development work mostly evolved around carrying over attribution in GIMP, an advanced image editing tool. GIMP development branch has recently gotten metadata support for quite a number of formats, so we did some experimental code for preserving metadata in a remix cycle on top of that.

Unlike Inkscape, GIMP has very basic document structure. Additional images can be added to the document as layers, which can be duplicated and destructively merged. Handling metadata in cases like those raises a number of questions, since preserving attribution requires duplicating and mixing metadata as the image contents gets mixed. Several possible directions of metadata flow in an image remix cycle are illustrated in the following diagram:

The first question that comes to mind is what to do with EXIF data here. Does it make sense to save e.g. camera settings in an image collage completely or partially? Or should we entirely drop any metadata that’s tightly coupled to original work if that work is reused in a different context?

Second question is about how to mix Dublin Core properties notably Source, Subject, Relation from two images and converting linked RDF metadata to XMP, e.g. license. Actually, software should be able to automatically recommend a license for final work based on the licenses of source works involved, but OK, that’s a long-term goal.

These are questions that we are continuously thinking about, and in the next round of iterations, we’ll see how we can deal with them. If you’ve got an opinion or thoughts you want to share, please do so in the comments.

read original post on Artem Popov's Site