Finding your XMP file

Am I missing something re. locating XMPs when revising photos?.. I have the raw, XMP and resulting jpg files in the same folder, and I name the XMP to match the jpg. If I come back to revise an image, I import the raw, then load the XMP since I want to be certain I’m starting out from where I got to previously. I know there’s a database but I don’t want to particularly rely on it.
So I click “load sidecar file” and it takes me to my “user folder”, i.e above downloads, photos. documents and all that. This means I now have to navigate to my specific folder. Instead I might click “recent” hoping this will have the folder I just got the raw from, but it doesn’t.

Is there something I’m missing please to get straight to the XMP file?

Stop renaming the xmp files and it will find it out of the box.

Unless you remove it the jpeg file has the xmp also included so you dont have to rename it.

Read more on side car files here:
Always keep the automatically generated sidecar file with your raw file in the same directory and don’t rename it. darktable will recognize it if it has the expected naming convention when you re-import or open the photo.

As Bbawt indicated, the xmp is in the folder holding the raw file. Beyond that, it sounds as though you are not using duplicates. I immediately make a duplicate, indicated by “_01” in the xmp for the duplicate. Develop that and make a jpg, but put the xmp file name in the name of the jpg, or use some other way to tie the jpg to the particular duplicate. You can create the file name after you have set the storage directory in “storage options.”

To further develop the first duplicate, make a duplicate of it. Or you can make a fresh duplicate of the original raw file and start over. When you are working on one duplicate, you can use the duplicate manager to compare it to other duplicates.

Basically, sidecars are redundant, all information is stored in the dt database. So you don’t have to re-import every time in any case.

In addition, the XMP metadata added by darktable to jpgs and duplicates contain a field “derived from”, pointing back to the original raw file. In my case, the xmp metadata are stored directly in the jpg file as well as in the side car.

Yes, use darktable as it is, trust the devs to have chosen a safe default and stop messing with the files on which dt is relying. In other word, do nothing and all will be well.

I haven’t studied the dt file system; I do back up the AppData files (Windows) daily, but I don’t rely upon them. When I download images from my camera, I copy the folders from my camera to my H drive and rename the camera’s folder name using the format YYYYMMDD. That format lists the folders in my file manager in correct chronological order. All those files are backed up daily.

When I want to develop a photo, I create a folder just for it. When I open the folder in dt, it puts the xmp in that photo’s folder on the H drive. xmps for all my duplicates are also put in that folder by dt.

In addition, I look at Play Raw posts all the time, and when I want to try my hand at developing an image, I create a folder for it on my H drive and put the image file in it. I also download the xmps that are posted by others. I copy the username of the person who made the xmp, and then I punch a key on my auxiliary programmable keyboard, and that changes the xmp file name in the Save dialog by putting the username in the file name. When I want to study someone’s development, I create a duplicate in dt and load that person’s xmp using load sidecar file. I keep a text file in the folder in which I record which duplicate goes with which xmp.

Each image’s folder gives me something I can look at to see where things stand. If I had to restore my files, I would know exactly how to do it.

It is true that in darkroom you can name the duplicates using duplicate manager, and I sometimes do that. A particular downside of relying on dt, however, is that I have opened a folder, done development on an image, and then dt crashed. When I opened dt up again, it did not know that I had opened the folder in the first place. Since I use the compact history function frequently, I know that I can go to my image folder on my H drive and load the xmp and be back or almost back to where I was before the crash.

Maybe there is a better way to do it, but I like the system I have worked out, and I have peace of mind.

Thanks for the comments. dt is a great piece of software, and plays a decent tune, but I’m not going to dance to it. I work on my photos in a simple way, probably started years ago with photoshop, then Canon’s software, Rawtherapee and more recently dt. The same simple structure works with most software, so I don’t get over-dependent or confused. That is, raw file, output jpeg(s) and pp3/xmp/“recipe” all clearly named and in the same folder. I don’t need or want filmstrips, image groups, duplicates or a database. If others use these, great. [btw I don’t delete the xmp dt creates, I rename a copy of it]

Clearly I’m not the only one who has simple requirements re. file storage. I think it would be good to modify dt so that when you use a folder, e.g. to import a raw, the folder appears in the “Recent” list; there’s plenty space available.

The “recent” folder you refer to in the “Import side car” dialog is most likely the Gnome Recent folders short cut, since it is a GTK dialog. I don’t think you’re missing anything, just that GTK isn’t keeping track that closely of the folders you navigate to inside darktable.

That is fine.

The XMP file is written along side the RAW file. As long as you’re not deleting the database on every run or using the :memory: option for the database, then darktable remembers, via the database, that you’ve already imported the image, so no need to load the side car file, it’s already loaded in the database from the last time you edited it. If you want to preserve your edit, use the duplicates manager to create a new XMP file (either a copy of the current edits or you can start from scratch).

Don’t use dt then if it is not made for your workflow.


When I import an image or folder, I believe the folder always does show up in the recent list and also in the recently used collections list. IIRC, if I import a sidecar file from another folder, that folder doesn’t show up in the recent list.

Use duplicates. In lighttable you select the original raw and click on the duplicate button in the selected images section. That creates another xmp, for example, DSC02876_03.ARW.xmp, in the folder where the raw is located, and an image appears in lighttable for that xmp. Now select that image and load a sidecar file. That would change, in this example, DSC02876_03.ARW.xmp to be the same as the sidecar. Now all you have to do is keep track of which sidecar file the duplicate is tied to; that’s where I use a text file in the folder with the raw, which also contains the duplicate xmp.

As I said, you can name the duplicates in the duplicate manager in darkroom. When I am examining the development work of someone’s xmp that I got from a Play Raw, I name the duplicate with the username.

You don’t need to rename xmps; use duplicates.

Edit: you can tell which duplicate an image is in lighttable by selecting, or just putting the cursor over, a duplicate and looking at the version number in image information. The original xmp is version 0. If you’ve named the duplicate, that shows up as title.

If you don’t use the database you can start darktable without it using the command option:

darktable --library :memory:

Then you just have to import the folder containing your images and darktable just uses the XMP files. No need to think about filmstrips, groups etc. as you stated before.
darktable just keeps the last used folder when you reopen it later. The XMPs are loaded automatically.

I don’t see a reason to make copies or do additional imports of the XMP files for that.

I’ve been using DT for several years and sometime do several “versions” on the same raw. The only thing I do is :

  • Select my 1st version in lightroom mode
  • click in selected section -> “duplicate” (wich create a new entry for the same raw in DT database and do not duplicate the file)
  • click on the “duplicate” raw and go to history stack -> “discard history stack” wich reset (remove the reciepe) the entry for the raw

DT will on his own create a new XMP when I start working at the new “version”, I let the XMP by the raw file and never manage these on my own, I just re access my versions through the lightroom view of DT. Alternatively if you start from your file manager, opening the raw in DT and pressing “L” to go to lightroom view will show you all your “versions” for your raw right away.

If it’s not the explanation you were looking for, sorry for your time :), I must lack perspective because never really used other raw software than DT …

Good luck with your issue !

Thanks all for further comments!

Just to keep this in perspective, I posted to see if there was a way to save some clicks navigating to my photo folders (on a separate volume). It’s not a big issue!

@Pascal_Obry, no, it’s too good not to use, especially filmic. Surely you want to have dt useful for all sorts of users, not just the more sophisticated ones?

@Underexposed, thanks. Each to his own! I personally wouldn’t want to be maintaining a text file, I’d forget!

@pphoto, yes I now see that importing a folder (as opposed to just a raw file (or files)) does indeed populate the Recent list.

@clind, thanks also, and no need to apologise. But you must say dt, not DT, else you will get into trouble!!

@RawConvert: So please read the manual, you seem to fight the way dt was designed to be used. And no it is not suitable for everyone, this just cannot be and it is why in every domain there is not a single software (and that’s good) for everyone to use the one that fits best its needs.