> This might be inspirational? What kphotoalbum appears to do, is to scan all > files below a certain path, and match against the size/md5 sum database that > it has. This is very quick - it scans my image collection of about 18000 high > resolution images in 20-30 seconds
md5sum can take a while on large files, and video files are known to be large; it may be worthwhile anyways and better as an opt-in option to do this. Would it be feasible to run md5sum against just the first and last X megabytes of the clips that are incorporated into a project, to help keep this quick for very large files? eljefe$ ls -lh mall_trip_015.dv -rw-r--r-- 1 eljefe eljefe 266M 2008-11-05 19:58 mall_trip_015.dv eljefe$ time md5sum mall_trip_015.dv bcd35a02015854813de36aa8d318cebb mall_trip_015.dv real 0m3.055s user 0m1.988s sys 0m0.532s eljefe$ The above md5sum didn't take too long but with a whole lot of clips it may take much much longer; or not. Just a thought. > icon in the GUI. In Kdenlive that would amount to e.g. being able to move > clips around in the timeline, edit titles, all that kind of stuff, without > having access to the actual clips on disk at the moment. This 'off-line editing' would be fantastic, as it would allow for less disk-access for the main files (which may be stored in another location); the kdenlive project file could then be sent to that location for rendering. I think this would require a low-quality copy of the footage to work, however. Would it be worthwhile (for RAM reasons etc) to have a tool to convert/compress the real footage to a much smaller version to work on? Or is this overkill and not needed due to the way KDEnlive already works?