I'm confused -- are you entering this information manually or are you pulling it dynamically via software from the image stream. seems like a) this would be a complete pain to do manually for a large system of photos and b) it seems that the software would have to be really amazing to pull stuff like line orientation, rhythm regularity, etc -- stuff that even humans have a hard time defining -- and give you a usefull set of metadata. On the plus side, it seems like once this attribute tagging is done, it would be really trivial to pipe them into an automatic slideshow (or whatever), based on this data.

So what will that look like? I'd want to see two things: a processed piece of media created from random, dynamically tagged photos -- and another created by hand, so to speak, with the photos and tagging all done by you before they go in the system.

attributes are much better than your previous version, btw.

I used elements originally for structural symmetry and ease of editing/extension. Though attributes are cleaner looking and seem to be just as easy to use. So the switch is made.

At the moment, I'm creating the metadata for individual images manually, though I expect to be able to automate the metadata creation process for most of the visual components.

The hand-crafted version is in development with an ETA of Wednesday.