In the latest installment of my ongoing series of posts about my current "Picasa Replacement Project", a few days ago, I managed to make a few more easy wins:
1) First, by applying some of the simpler manipulations to a "calibration" image, and trying to match the results - This has quickly revealed some starting points for working with the "Tilt" and "Color Temperature" sliders (even if the values obtained still seem a bit odd / weird, and may still need some fine-tuning)
2) It turns out that during an earlier search session, I had in fact managed to dredge up a bunch of very useful posts about the various formats used - especially for the on-disk thumbnails / preview images databases (which will prove useful for having a way to batch-test some of these filter calibrations later (i.e. the thing apparently uses the standard Windows "Thumbs.db" format, for which libraries do exist !!!)
However, to protect all this info, I'm gonna have to make sure I save off all of it on the various archive sites (in addition to having already grabbed copies for my own stash, in case they disappear before I manage to do even that)... So yeah, I'm not going to publish the list of links until I have secured them.
22/7 Update - Tilt and ColorTemp
Hah! Turns out the simplest manipulations to reverse engineer did give up their secrets quite easily, once using a very basic "calibration image". It's fascinating seeing what they were doing actually!
27/7 Update - Database Format Findings
Yay! This morning's project-work session has proven most fruitful:
Reviewing one of the older tabs I'd saved off revealed that I had in fact stumbled across the "mother lobe" of resources pointing to lots of links about the core database formats of this thing I'm reverse engineering.
The biggest find: The thumbnails format is just the standard "Windows thumbs.db" format... so, in theory a standard parser can be obtained for that (and if the rumors about data corruption in larger DB's hold, then it's probably a case of certain implementations having integer overflow bugs)
----
As for the .ini settings files that I'm parsing though for the most critical stuff, between the reference notes I've mostly fully implemented already, AND the parts of those settings for which I actually have a much more detailed breakdown (as I *really* need those working):
* There's currently still a real void of information in the really critical areas I'm most curious about (i.e. how those parameters are interpreted for the most critical manipulations)... But, with the thumbs.db problem possibly solved now, automating a parameter search / calibration routine may now be feasible...
* For the rest... they're probably all easy enough that I just need to get through deciphering them all
No comments:
Post a Comment