Using DCP to detect sequential image "Sets"

[GraFXman]

I’m using DC v.4.01. I have an image Directory with numerous folders containing well over 150,000 images spread out between them. I’ve been using DC for years and it’s been quite useful for detecting and removing hundreds and even thousands of duplicates that have accumulated over time after constantly adding new images to the Directory.
However, in addition to duplicates I also have another tedious challenge when working with such a massive number of files: Finding image “sets” – meaning, sequential images containing the same individual, object or scape that are part of a “set”, which can be from a few images, to dozens.
When doing a DC scan using anything except “exact match” DC usually detects several images that may appear to be DUPS but are actually part of a set. I mark those separately and use the MOVE feature to move them to another folder titled “sets” where I then go and compile the sets.
The problem is that it’s inconsistent and DC will place images that are actually part of a set into different groups, so by the time I get to Group#244 I might see an image I recognize as being the same person and probably part of a set, but by then I can’t remember what earlier Group it was in. DC also can’t seem to detect all the images which are part of that set so there are always some missing which I end up finding later when sorting manually.

Is there some trick or filter that could be used for the detection of sets containing sequential images?
I created a test folder where I put numerous sets plus hundreds of randoms and I’ve tried every option and/or filter I could think of but with no success. I would think that using the “Same Created Date” and/or the “Same Modified Date” would give accurate or at least better results on the test sets because each image in the set has the same unique date as the rest of the images in the entire set. Other filters/settings don’t seem to work because not all images have metatag/metadata or other unique identifiers.
No matter how I run the scans DC will only detect maybe 4-5 images out of a set containing 15-20 and even then it will put them in different Groups rather than one single Group.
Any recommendations on settings or filters that might yield more accurate or at least better results?

[therube]

If date modified (or date created) will get you close enough, you could do something like this...

Image Mode

Similarity -> Ignore
Same file modified date/time -> Match date only

Then on the Duplicated files tab, set it to Thumbnail view

The problem here is that Thumbnail view, while it sorts each Group by date modified, it does not maintain the sort of each Group by date modified.

So that if you took "similar" pictures of a period of days, 6/18, 6/19, & 6/20, all the 6/18's would be grouped, all the 6/19's would be grouped, & all the 6/20's would be grouped, but each of them need not be grouped sequentially (by date modified). One could be at the top of the listing, another at the bottom, & the 3rd somewhere in the middle.

(If you left it at Details view, you could then sort by date modified, in which case those 3 sets would be sequentially displayed - but, it might be harder to visualize [cause less would display in each screens worth].)

-

Could try a different program. Everything (Search), can sort by date modified (or created, or...), & also display thumbs, so it might give you a greater context view.

In one case, you dealing with same date modified groups, in another, you're simply dealing with all files, sorted in date modified order.

.

.

[GraFXman]

Thank you Sir for the great information and suggestions!

I realized the first thing I needed to do was upgrade from v.4.01 to this new v.5
THUMBS UP to James at DV support who helped me with that!
WOW! What a difference! Had no idea what I was missing!
I am loving the new architecture, GUI and functionality!
I had a blast just playing around with all the functions, so cool.

Just to clarify;
Referring to "sequential images" what I meant was that all images were taken in one shoot sequence on the same date, not different or sequential dates.
The results were just as you predicted, but to my surprise while playing around with it DC actually discovered and grouped what I believe is an entire set that I didn't even know existed!
Also correct, sorted in details view showed the sets from that particular date, but like you said, with non-sequential and weird file names it's impossible to determine what they are - and maybe not ideal or convenient, but it's easy enough to click and open the image.
The new challenge then was discovering that DC detected over 870 files with the SAME created date!
Dam, I'll have to play around with additional creative detection and sort options.
I also installed your recommended App "Everything" and yes indeed, it certainly works for that purpose, defeated only by the large amount of files that unfortunately have the same date. It's actually kind of interesting to think how many people around the World were shooting on that one particular day!

Overall, I'm extremely happy with my new DC and love playing around with it and seeing all the different results.
PS. And for some odd reason, while playing around DC has detected and grouped what seems to be entire sets I didn't even know existed -so that's like a bonus!

[MegMac]

Date Created and Date Modified are File/System dates. Uploading/downloading, moving, and editing files can change File/System dates.
The only reliable date field is Exif DateTimeOriginal, which is not part of V4.
If Image similarity is set to 'Ignore', and you match by Exif DateTimeOriginal, and NOT match subseconds, you'll get groups with every photo taken during that second.
If you set 'Group by Time Taken' to 2 seconds, you'll get you'll get groups with every photo taken within 2 seconds.

[GraFXman]

The only reliable date field is Exif DateTimeOriginal

Hi MegMac,
I have updated to the latest V.5.
Thanks for the info, I had not considered that and will continue experimenting.
It's interesting to see how the results change with different settings.