NLP conversions (shouldn't) die with the Lightroom database

As of now, all NLP conversions are only stored in Lightroom’s database. If that database is lost - for whatever reason there might be - all the NLP conversions are lost, together with all virtual copies, should they have been used. Well, isn’t that an important reason for regular backups!

I herewith propose to add a feature to NLP that provides another way to save NLP recipes, so that files and their conversions can be restored without Lr database or database backup.

Other apps do this by writing app specific sidecar files or by adding related info to .xmp sidecars.

1 Like

You can set LrC to write develop settings (and some other metadata) to XMP in the catalog settings.
For proprietary raw files, the settings will be written to sidecar XMP files.
For DNG files and rendered files (tiff, jpg), settings will be written to the header of the files.
Although develop settings are included, history states are not, as well as ratings, collections, virtual copies, and probably some other things that I can’t remember.


Thanks and welcome to the forum @PerB

Testing if XMP sidecars would help, I found that they don’t. That is why I posted this proposal. If all I had were the RAW and XMP files read by Lightroom without prior knowledge of the conversions, the imported files just came up as negatives…


  • Conversion data is stored in XMP sidecars, if the master image was converted.
  • Conversions are lost, if they have been done with virtual copies exclusively.

Why: As of January 2024, Lightroom Classic versions 12 and 13 do not save any information about virtual copies in XMP sidecars. Although those sidecars transport NLP conversion settings, they only do so for original files, but not for virtual copies.

Implications: If one e.g. copies image files to a new computer without copying the corresponding Lightroom database, virtual copies are lost, no matter if XMP sidecars are copied or not. Note that the NLP profiles need to be copied too.

Best Practice: Always copy the whole deal: Lr database, image files, sidecars, profiles…

Background: I wanted to see if my long standing Lr database was still corresponding to what was on the drive. For this, I created a new catalog and imported the complete photo archive into it. I then discovered that the new catalog contained substantially less images than the original catalog…and the difference between the numbers corresponded to the number of virtual copies. Researching the issue, I found that VCs are lost indeed in such a scenario. I also found one source who wrote that snapshots won’t be lost, but did not test if this was the case or not.

@nate while this might be a Lr topic rather than a NLP topic, it might be an idea to drop a few lines about this in the manual. Feel free to use my text as is or edited to taste.

I don’t have any raw files to test with (I only use scanners), but it works for me with Tiff files, as long as Tiff support is enabled in the Camera Raw preferences.
Double-clicking an NLP converted tiff in the File explorer opens it with settings in Camera Raw.
This is on Windows 10.


Researching the issue, I found that VCs are lost indeed in such a scenario. I also found one source who wrote that snapshots won’t be lost, but did not test if this was the case or not.

I found this in The Lightroom Queen’s The Missing FAQ (from 2019)

Which of Lightroom’s data isn’t stored in
Flags, virtual copies, collection membership,
uncommitted location data, Develop history,
stacks, Develop module panel switches and
zoomed image pan positions are currently
only stored in the catalog itself, and not the
XMP sections of the files.

I’m on MacOS, but I believe similar exists for Windows - I have an external HD and Bombich Carbon Copy Cloner on the Mac. It backs up the whole drive one sets it to back up, so everything is cloned and nothing gets lost. I also have a subscription to Backblaze which runs in the background and provides a frequently updated back-up of specified drives to their servers, hence on off-site back-up as well. With these two utilities one can rescue the last good Lightroom catalog and image files in case anything gets deleted by accident or corrupted.

As for working procedure, Negatives captured with my camera are stored in the hard drive and in Lightroom as DNG files automatically. When NLP converts them to positive, the result is a TIFF file which sits beside the DNG file. I would only make Virtual Copies of any of these TIFF files if I were applying a dramatically different kind of editing to the photo and wanted to keep the original conversion intact, so I don’t have a great many VCs to worry about.

Thanks for your responses, @PerB and @Mark_Segal

I’ve been working with DxO PhotoLab (as a complement to Lr) for a while…and PhotoLab’s own sidecars transport all information about virtual copies and I got spoilt by this feature which is very welcome, specially when testing on more than one computer. Copying a few files is so convenient in comparison to copying a multi-GB database and having to adjust locations after the move.

I was testing a different scenario though: Assume you Lightroom catalog gets damaged to a degree where it’s still recognised by Lightroom, but catalog contents is beyond what Lightroom can repair. Such a situation might build up over time and who can know, which catalog backup is intact? That’s where the sidecars could come in handy as a last resort…if the were written in a way to enable such functionality.

Other than that, I’ve been using TimeMachine and Carbon Copy Cloner for years and always got backups for all the cases when I wanted one. I almost never keep exports though and usually re-develop and -export images when I want a print.

Thanks @Digitizer!

YES - The Lightroom Catalog database is very important.

I strongly encourage everyone to have Lightroom setup to automatically back up their catalog.

You can do this by going to “Lightroom Classic > Catalog Settings” and setting this there:

Screen Shot 2024-01-18 at 2.23.26 PM

To me, this is the solution if you are concerned about the Lr database integrity. In the event that something goes wrong, you have that backup which will not only ensure the NLP conversion data is correct, but also bring back virtual copies (something I would not be able to do even with some kind of XMP sidecar).

This also introduces the possibility of mistakes in the XMP files :wink:

Have you tried the “Export as Catalog” feature in Lightroom Classic?

You can do this with as few or as many files as you’d like. There’s no need to adjust file locations after move… you can either have the “negative” files themselves included in the export (great for bringing to a new computer), or you can have the catalog continue pointing to the existing location of the files.

Good idea!

1 Like