Korsakow System

Have returned to playing around in Florian’s database film engine, aka Korsakow System. It has been completely rebuilt (in Java as far as I know) so was quite a shock when I launched it. However the interface, once over the initial surprise, is actually easier to use and understand than the earlier iterations of the project. So, I’ve put in a few videos, some thumbnails and started making connections. What I’m looking to do is to find a system that lets me easily do something with all the little bits of video I have. Video is now a snapshot medium. Free to make and nearly free to distribute, and so we all have lots of it. But while we have sites like 12second.tv, and flickr’s support for video, we don’t really have a good way take all these little bits and do something more interesting than galleries and ’sets’ with them. So I’m interested in seeing if I can use the Korsakow System as a way to publish these fragments, but also connect them in associative, non linear ways, as I would in a decent hypertext system.

So, I’ve added clips. Each clip will have the year as a key tag (these tags are the text strings that you define for clips, the system lets you say what tags are attached to a clip, and then conducts searches for specified text strings) and a couple of other key words. Then each clip searches for at least one associated key word, and then at 1, 2, and 3 seconds it will also search for clips from the previous year, this year, and next year. I had originally only searched on year, but it was quickly apparent that some of the connections made were too disjunctive, so I’m hoping in this way things will be more poetic. Having had a quick run through of this the issue is that you get caught up a bit too easily into the one thematic group, so I need to do a bit of tweaking there, perhaps adding a random link for the third link.

One problem I’ve found is it looks like I need to maintain a media library of all that I put into the engine locally. I built the first iteration, exported to the web, deleted the thumbnails I’d created only to then find I can’t export because it can’t locate the media. So I rebuilt the prototype from scratch, putting media in appropriate places first. The workflow is pretty easy, compress and edit (or edit and compress), get a thumbnail, add both to the system, tag, export. The only issue is that media assets are shown in a single scrolling view so if I use this for all the material there will be hundreds, so I will need a naming convention to be able to find things, and also I expect I might run into some problems if I end up with thousands of clips in there. Or I’ll just go crazy trying to find them. UPDATE: you can have folders in the the media window, that is very very good.

So, rebuilt, and this time themed things a bit more strongly, but also providing links between some clouds by including antonyms. So for example if there is a tag for country and a tag city, then for a country clip I include the possibility of searching for city, and vice versa.