https://bugs.kde.org/show_bug.cgi?id=305104
--- Comment #19 from Johannes <johannes.lists+bugs.kde....@tannenhof-imshausen.de> --- Finally I was able to correct the database. What I did trying to figure it out (using digikam 4.12 on Ubuntu 16.04): I set up a new digikam collection with only the images from dropbox and dumped the sqlite database. Then I removed the ImageUniqueID from the images were it was not unique with exiv2 -M'del Exif.Photo.ImageUniqueID' /home/user/dropbox/images/*.jpg Then I removed the database-file to let digikam create it again and dumped it, too. (Now all images were already shown and no wrong images were assigned as identical.) Then I looked at the diff of the two database dumps. By that I figured out that the table ImageTags is also involved: Images wrongly identified as identical to others have the tagid which has the name "Original Version" in tags table. With this information I were able to edit my original digikam database and now everything works right. Follow these steps to solve the problem: 0. Close digikam and make a bakup of the collection database. Make sure to have sqlite3 or SQLiteManager for Firefox or a similar tool available to check and edit the database. 1. Identifying photos which have identical values in: Exif.Photo.ImageUniqueID This is easily done by SELECT *, COUNT(*) as c FROM ImageHistory GROUP BY uuid HAVING c > 1 ORDER BY c DESC (I also saw only photos affected which have no history entries: SELECT *, COUNT(*) as c FROM ImageHistory GROUP BY uuid HAVING c > 1 AND history IS NULL ORDER BY c DESC ) I removed the entries with unique uuids after copying the imageids to a file for further usage. 2. Remove the ImageUniqueID from the images where it was not unique with exiv2 -M'del Exif.Photo.ImageUniqueID' /home/user/dropbox/images/*.jpg 3. Then DELETE the offending entries in the tables ImageRelations and ImageTags using the imageids from step 1. (Because I edited some images I couldn't simply delete all entries claiming to be an "Original Version", so I looked through them manually, comparing the name or tagid.) For example I used queries like SELECT imageid, name FROM ImageTags LEFT JOIN Images ON Images.id = ImageTags.imageid WHERE name LIKE '20160624%.jpg' AND tagid = 18 to check the database for images of a certain name and the corresponding tagid for "Original Version". Look in the table Tags for tags containing "version" in the name to find out the right tagid: SELECT * FROM Tags WHERE name LIKE'%version%' 4. Close the database and start digikam again. That's it! -- You are receiving this mail because: You are watching all bug changes.