Write keywords to XMP

Insert the> (or <) character to define a hierarchy



It is not the same, I do not want to type in hierarchy keywords with all typographic errors possible. I want to browse hierarchy trees and apply existing keywords.


fully understand. That is the reason why I use a dedicated DAM software.


Can you guys please recommend me a good reading about hierarchy keywords ?
I have an idea but where do I find the best and easy to understand informations about how it works ?



et en français


i use a year\ 3months\ folder with date and event or place and my end archive has same system but more devided: holiday, family,daytrips, divers and pets
I am not a 10k/month shooter so no problems finding things.
i use a old pse13 library system which i can use to write tags inside the jpg’s exifdata by exporting.
if i could group photo’s by a topic but not a “tag” like “Home” “Holiday” which i use a search keywords. if you use keywords as “holiday in Paris” you end up with lots of keywords which you use only ones. and in my mind “Holiday” is a keyword and “Paris” is a topic/header a unique word to find things.
And that’s i think the issue in all kinds of DAM systems : personal way of thinking.
XMP or IPTC is depending on your use criterium and possible acces to a file.
Rawfiles are more closed containers then Jpeg or Tiff so apperantly only XMP “sidefiles” are easy to create to support your DAM without only depending on a internal Database.
But i think i would be using a image internal info based (IPTC) on the endproduct side.
Which can be rather comprehensive:

So the rawfile database and XMP files side by side on the RAWfiles. wile the export can be embedded inside the image properties. But you need/want a possible export without any IPTC data for web based upload and thus create a web folder imagedatabase where the info isn’t spread out and readable for others.

And here is a Watermark and export module come in handy: place your watermark and checkbox the IPTC/XMF data which can be export with the jpeg image for the web.
“remove EXIF data” is much to a sledgehammer and a more specified export module which you can use to choose which exif data and xmp data will be embedded and if you like to “sign” with watermark will be a big step forward to use PL as centre of your workflow.
for now we need to relay on external applications for managing this kind of things.
Using different apps for small parts of there possible use to get a puzzle resolved as hole.


To keep it clean: IPTC has a location area with Location, City, State, Country. So Paris shouldn’t be a keyword but it should be a City in IPTC under Image Information.

If you refer to “Holiday Paris” as an event, then indeed the keyword should be “Holiday Paris”, although I prefer to abuse other not useful metadata fields for that so I can rebuild my directory structure time after time from metadata.

Yes correct. So a editable IPTC field in your workflow is a key feature. (as export module and internal DB inside PL only is maybe to fragile and closed in so a file side by side of the Rawfile to be flexible in the future and have a solid backup non application dependable is a pre.
That i understand from DAM’s but i am not a experienced user or builder of DAM’s and as people already wrote the more DAM structure and managing tools you build in DxOPL the bigger/slower it gets.
So i think a XMP creator/updater for the rawfiles (preferred as command triggered not automated) and a small simple export module for jpeg/tiff’s with XMP and IPTC editing&writing when exporting edited rawfiles wil be enough for most DxOpl users.
All bigger/ heavier users shall probably relay on Photomechanic or something suitable.


Fully agree. Most important is reading all the info (keywords, color tags and ratings) from external applications and duplicating inside (e.g. add DxO specific keywords). And then a manual export to .XMP file or write to .DNG/.JPG whatever file somewhere buried in a menu.

It should be there just to aid workflow/browsing. E.g. I forgot to add a keyword. Now I have to leave DxO, open another application, add the keyword and go back to DxO. Or I wait and I probably forget to add it. So there is value, but it needs to be implemented with caution.


Definitely a +1 vote from this new PL user :slightly_smiling_face:. Writing keywords to (and importing keywords from) XMP is an essential feature if PL wants to establish itself as a serious DAM option… I for one would love to have PL as my single go-to software for the whole photo-processing workflow and be able to do all my keywording there (looking forward to the Windows update…), but also want to be able to view and use those keywords in other tools.


I agree with you here Floris. There’s no good excuse for PhotoLab not to write keywords to an external file which will survive a database reset (which is often needed to get folders to update properly). In the worst case, PhotoLab could write the keyword info to it’s own .dop file for each image. PhotoLab has the most control of the .dop (no other application else can or should use it).

A better situation would be to write the keywords into an .xmp sidecar. My images already come with an .xmp sidecar from FastRawViewer. There’s some kind of IPTC title and description support (which I don’t use for now but will consider it going forward if I can find a way to get that info into the final jpegs). The FRV sidecar title and image do not show up in PhotoLab for now (ratings work fine).

I can see this DAM issue is not becoming simpler, DAM is a veritable Pandora’s Box… The idea to add a simple keyword tool has now resulted in time consuming manually entered data ending up vulnerable to loss and deletion: data loss which DxO set out to eliminate with the .dop sidecars in the first place.

For the technically minded, here’s what the FastRawViewer .xmp sidecar data looks like with a sample title and description. There’s many other applications sharing .xmp sidecars: DxO will have to tread carefully here and play nicely not to create issues by writing into the XMP file.

<x:xmpmeta xmlns:x="adobe:ns:meta/" x:xmptk="XMP Core 5.6.0">
   <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
  <rdf:Description rdf:about=""
     <xmp:CreatorTool>NIKON D850 Ver.1.00     </xmp:CreatorTool>
           <rdf:li xml:lang="x-default">Skerlan throw-in</rdf:li>
           <rdf:li xml:lang="x-default">Sebastian Skerlan throws in the ball</rdf:li>

Unlike some people here, I find picking keywords out of a visual hierarchy to be a huge suck on my productivity. For me, Lightroom (Classic) had it just right. Here’s an example of the keywords I would be adding (probably to multiple images at once, and some of those images may gain additional ones also).

ZK-MVA < ATR-72-600 < ATR-72 < ATR-GIE < aircraft
Mount Cook Airline < airline
NZWN < Wellington < North Island < New Zealand

What I would need to type for all of that to be added would be:
ZK-MV [enter] Mou [enter] NZW [enter]

Sometimes I’d need to use a few arrow keys to pick from the auto-complete options, or type in a few more characters to zero in on the correct one.

Further to the above hierarchies, some of the keywords would have synonyms, so the complete set of keywords written, by means of my 14 keystrokes, would be:

ZK-MVA, ATR-72-600, ATR-72, ATR-GIE, Leonardo, aircraft, Mount Cook Airline, Air New Zealand Link, airline, NZWN, Wellington International Airport, Wellington, North Island, New Zealand

To do that with a mouse would require finding the three leaf keywords in three separate hierarchies, one of which is hundreds of keywords in size and 4 or 5 levels deep in places. By typing, I save a ton of time and also a ton of concentration. I’m literally looking at the photo thinking “That’s ZK-MVA, it’s operated by Mount Cook Airlines, and it’s at Wellington.” I would typically be adding keywords to at least a couple of dozen photos, sometimes up to several hundreds and in some cases having to research identities at the same time.

FWIW, PL3 has a good enough implementation at this time that I can still keyword to the same accuracy – it’s just a little hard to manage them if I need to rearrange hierarchies or investigate already tagged photos where I think there may be anomalies. It’s almost worth it to me to once again pay the subscription for LR simply to keep managing my photos, though I’d now use PL3 as an editor.

Hi everybody,
I tripped over some strange metadata issue in the new DxO PL3 metadata editor (I am using Win 10 where this feature was rolled out recently) and I found this interesting thread in DxO forum.
In essence, the new feature is useless for me.
Here’s what I found out:

  • Opened a jpg with existing metadata (no sidecar file, embedded metadata)
  • Added Keywords in Photolab 3
  • Exported the image from PL3
    Now I checked the exported file:
  • in Adobe lightroom the newly added keyword appeared correctly in the keywords box, no problem,
    opening the file from other image viewers (like freeware IrfanView or Pciture Information Extractor) didn’t show the metadata attribute that I was adding in DxO.

After some looking into the metadata in exiftool I found the following:

  • DxO correctly writes metadata into XMP, segment dx, into field “Subject”
  • DxO also correctly writes into the IPTC part, into field “Hierarchical Subject”
  • But DxO does NOT write the new tags into the IPTC part, field “Keywords”.
    I know, IPTC is pretty old, but still quite a lot of picture viewing programs do show IPTC keyword data, reading from this exact attribute. Lightroom, for example, correctly populates this field upon writing metadata into a file,
    I have issued a Ticket towards DxO, but support gave me the answer that it would “work as designed”.
    So if DxO considers this to be working correctly, the whole feature is useless for me, since I will not be able to see metadata maintained in DxO and I need to stick to my old DAM workflow.

Anybody else having a similar issue? If I am the only one, then possibly there is not the need to corrrect the existing Photolab 3 way of writing metadata :slightly_smiling_face:

1 Like

It is exactely what they announced before: All keywords are ONLY written to the PL3 database.
For me useless.
IPTC data must be available for all programs.


At least I was hoping for a minimum fuctionality to enter IPTC / metadata that could be reused in other programs … I am asking for interoperability :slightly_smiling_face:

There should be a button or a possibility in the settings for us to decide where we want the IPTC data to be written. A wish for the future.


On top of expanding metadata capabilities and writing to xmp, it would be nice if photolab also honored the pick/reject tags from lightroom and showed them as picked or rejected inside photolab. Also of course being able to write out the pick/reject tag into the xmp sidecar.

1 Like

For now the keyword facility is only internal and more “search function if you got a external DAM for your work.” You need to do the taggingwork twice: on RAW on the pre PLv3 side and again on the exported Jpeg’s in the post-PLv3 because it can’t pas through keywords by export yet.

I am pretty sure that that pass-through functionality (writing) is coming soon.(otherwise the “DAM” is that crippled that it’s only frustrating people when they find out it can’t do anything outside the DxO-DataBase.)
In what form it will be writing new data outside the DB i don’t know.
I hope it’s XMP (RAW’s ,tiff and Jpegs) and IPTC (tiff and jpegs) capable in reading and writing/updating because it’s in between your RAW archive and photolibrary/Archive and then you are able to edit already tagged rawfiles by DXOPL or an other DAM App and push this in the end step to received by the DAM you use on those.
Time will tell.

I hope so.

I am already very happy with the reading. But sometimes while I am editing a photo I want to add a keyword or two (sometimes related to workflow) but then it won’t export which is frustrating. I don’t think many applications excel in batch keyword editing so I am not expecting PhotoLab to do so. Just some basic functionality to interoperate better with my other workflow tools.

1 Like