To All - Thanks for all the suggestions. I particularly like Raul's though it still uses breakpoints; I may use it with upper and lower breakpoints of 6667 and 3333 to minimize false positives to the point of making them extremely unlikely.
I'm interested in the timestamp-based ideas but am not sure how to get a timestamp which persists through common modifications. To clarify this, I've looked at things like Picasa (Google's photo organization software) but find it inadequate for my purposes. One unique problem I have is the number of photos I take, many of which I discard from the camera before offloading them from the chip. The genesis of this current puzzle was that I wanted to count how many pictures I took last year versus how many I retained. So, based on the number of rollovers, I took about 80,000 and retained slightly more than 34,000. Having this many photos causes problems even with basic software like MS Explorer - I used to segregate photos by year, each batch in a sub-directory named for the date range, as David Vincent-Jones suggests but this would lead to a 5-minute wait or so for Explorer to open the sub-directory with a year's worth of photos. Now I break the years into quarters and this seems to work OK. One potential problem with a timestamp-based method is that another common thing I do with a fresh batch of photos is run my J program which looks at the dates, figures out which quarter to which they belong, creates the dated sub-directory within this quarter, and moves all the files off the chip into the sub-directory. The next thing I do is to orient the photos right-side up. Since I take most of my photos "from the hip", I usually run some J code to flip them all a quarter-turn counter-clockwise, then manually re-orient the few that were taken in a different orientation. This unfortunately messes with the "modified" timestamp. I'm interested in what information may be included in the EXIF image header, as well as other timestamp information. Does anyone have code to read this header and the other timestamps? It would be nice if the EXIF information included something about camera orientation but this might be hoping for too much. I've tried to get the different timestamps using the kernel32 API GetFileTime but am having problems with it - more to follow on this unless someone has a working example. Thanks, Devon -- Devon McCormick, CFA ^me^ at acm. org is my preferred e-mail ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
