Skip to content

Commit

Permalink
Changes in v1.1
Browse files Browse the repository at this point in the history
  • Loading branch information
jesjimher committed Oct 16, 2014
1 parent 13a24fe commit 253157a
Showing 1 changed file with 8 additions and 4 deletions.
12 changes: 8 additions & 4 deletions README.md
Original file line number Diff line number Diff line change
@@ -1,10 +1,9 @@
imgdupes
========
#imgdupes

imgdupes is a command-line tool that finds duplicated images in a directory tree. The difference with other file-oriented utilities (like fdupes) is that imgdupes is specifically tailored to JPEG files, and compares only the image data chunk, ignoring any metadata present in the file (EXIF info, tags, titles, orientation tag...). This makes possible to find duplicated images when one of the file's metadata has been modified by imaging software, and byte-by-byte comparators fail to report them as equal. This might happen in a number of situations, for example:

- Modifying EXIF info to adjust date taken (typically when the camera has been set at a wrong date/time)
- Adjusting rotation flag
- Adjusting rotation flag or rotating the photo losslessly (f. e. using jpegtran)
- Adding tags, or setting title, description, rating...

A common scenario that could lead to this kind of duplicates is importing a file, altering its metadata in some way (tag, rotate or whatever), and then re-importing it again because the image software isn't smart enough to realize that the modified image is the same than the original, unmodified file that it's in the camera. This kind of duplicates are annoying and hard to find, because standard file duplication utilities won't report them (they're actually different files), so human checking is almost always required. imgdupes tries to automate this task.
Expand All @@ -19,7 +18,12 @@ Analyzing each image chunk of data in order to compare and find duplicates is a

As a final disclaimer, imgdupes is provided as is, and I can't be made responsible of any damages that might happen to your collection by using it. I use imgdupes myself, so I'm reasonably confident that it works, and at the same time I'm the first interested in that it's free of bugs, but I can't make any guarantee of that. Keep also in mind that, even if imgdupes reports that two files correspond to the same image, this might not necessarily mean that you have to delete one of them. It's up to you to decide which cases correspond to software mistakes (i. e. re-importing an existing image that had been already imported and tagged) and which ones are legitimate.

History:
##History

v1.1

Support for losslessly rotated image detection

v1.0

Initial release

0 comments on commit 253157a

Please sign in to comment.