Would you mind coming back after you've played with it to let us know
how well it works? If it does what it claims, I'm sure others would
benefit from it, including myself.
I downloaded the 30-day trial version. I have mixed opinions about it
already. My main requirement was for the software to find duplicates
that were resized, cropped, or image-enhanced from each other. It rates
the matches as a percentage with 100% being that it's absolutely sure
that they are the same.
On the plus side, it's relatively fast, it was able to compare all of my
personal photos in less than 3 minutes, which was about 4000 photos.
Also all of the 100% matches are absolutely dead-on absolutely right,
perhaps just resizes of the same photo, or with some image enhancement
(brightness, contrast, gamma, etc.). It's also able to pick out photos
that are rotates from each other, e.g. something rotated by 90° or 180°,
but that would take a bit more processing time, and it's not enabled by
default because it takes longer.
On the minus side, it's when it's not 100% sure of a picture that it
gets frustrating. As I said, it rates its surety as a percentage up to
100%. But even at 99% sureness, it made some laughable mistakes, and by
98% surety, it just becomes comical. I wish it would come up with a
rating system that's further differentiated. The ones it rated as 98%
certain, I would've only rated them as 60% or 70% certain. This sureness
scale could actually be useful for its own purposes, such as
categorizing similar-style pictures, if only it came up with a better
differentiated rating system. Perhaps some kind of a logarithmic scale
ending at 100%, but dropping off precipitously as its surety goes down.
I think it seems to have some real problems with landscape pictures. I
don't mean landscape as in portrait vs. landscape orientation, I mean
pictures of scenery. It got confused by some pictures where it thought
pictures of cityscapes, parks, and beaches were the same picture! It was
further laughable because these scenes were even taken in different
countries.
On the other hand, some pictures rated 99% should've been 100%, while
others should've been maybe 80%. Yet both were rated 99% for whatever
reason.
At this point, I don't know if I would pay the registration price for
this software. It has way too many false positives.
Yousuf Khan