[Suggestion] Video comparison tool, difficult but some ideas

[abrasion]

Hello!

I'm thinking about a video comparison tool, I admit the work to create this would be exceedingly difficult to do accurately. HOWEVER if the onus is on the user to be distinct with the search criteria, it might help.
I might recommend a disclaimer before using this function : "warning may be VERY inaccurate, please confirm before tagging" and run with some of the criteria below to ensure the list is somewhat regimented / tight.


Here's some ideas which might assist.
1, duration of the video is precisely the same or customisable (plus / minus 20 seconds?) of each other
2, search only within same file extension or within a predefined set?

3, very loose file name search. (complex)
Example
Movie is called "day I stubbed my toe lol.mkv"
I also have a copy of the same film called "toe stubbed incident.mp4"
Now, the key thing is
a, both files are movies.
b, both files are within say 5/10 seconds length
if a and b criteria is met, then c, the file name has 2 identical words in it (excluding punctuation etc) [toe, stubbed]


I have a 20TB NAS with movie files in MP4, MKV, MOV, MPEG, MPG, AVI, M2TS, formats.
15,632 files, 15.2TB


NOTE: I would expect that adding support for DVD VOB / TS joining when doing the size calculation would be ridiculous and excessive, don't even bother with that concept.
I know you're busy but please consider this, in the very least the "extremely lossy name match" which simply matches 2 similar words and a file extension that's the same or in the same group - that could also be useful for word documents (call it "super lossy mode"?)

Real estate plans for renovation of the bathroom.doc
bathroom renovation plans.txt
plans - renovation of bathroom.rtf
bathroom renovation plans original.doc

(You could even make it a 3 word deep matching?)

[therube]

Everything is great for doing stuff like finding files named "toe + stub" & then combine that with the video: filter, well there you are.

There already exists "Similar name", though the algorithm it uses for "similar" may not necessarily match what you would consider similar, so maybe some tweaking on that end?

Move run time length would not be a good indicator, IMO, as it could be something like one copy is "complete", the other only has partial credits, so the "movie" is the same, but the 5/10 seconds would be blown out of the water.


There exists, Video Comparer.
I looked at it VERY early on, but didn't get to do anything worthwhile for me. Not familiar with it currently.

[abrasion]

I must admit I don't know the similar file names tool ability, I've used it recently and it still seemed "pedantic" and needs to be looser I think.
Movie run length depends on content of film, example could have a file exact same duration but one is rotated or one is .MKV one is .MP4 - so in that instance, movie run length would work a charm.
Another possibility, although significantly more complex and I imagine James would say "no thanks!" to, would be to use his MP3 audio comparison technology against the audio track of the video?