I am having troubles with DNG files in RawTherapee


I am currently running RawTherapee RawTherapee_dev_5.8-2847-g53bb6e41b_W64_Skylake_210422.
The DNGs in question have been converted from ARW files (A7RIII) with Adobe RAW Converter,

The problem is that my adjustments - in some cases, and I can’t find out when - and when not - have no effect on the image.

I received a pp3 file from a friend, and when loaded with the same DNG file as it was created for (in this case on a Mac and in another version of RT) the only effect in my Win10 system was a darkening of the image.

Then to exclude the possible incompatibility between two versions of RT and the two OSs - I loaded the image from scratch (neutral) and made the exact same adjustments in my system. Same result - a darkening of the image - with no traces of the specific adjustments.

The same DNG file responds normally in Adobe Camera Raw and Photoshop and different plug-ins.

PS: Aren’t we supposed to be able to exchange pp3 files across versions and platforms?


Do you have the original are file? Does that work?

Its hard to say what’s wrong without a file to inspect.

If you’re loading from an older version, yes. If you’re loading from a newer version, no.

Hi Mica
No, unfortunately I have trashed this ARW file (will never happen again).
But the anomaly interests me, and maybe others have experienced the same. I would like RT to be as stabile as possible with all raw files.

If you’re loading a newer version pp3 with an older release, most probably yes. All new tools won’t appear and those tools that have suffered major revisions will usually screw the processing, but the unchanged tools present in the older version should work as usual.

If you face some odd rendering (e.g. the image turns green, or magenta, or darker, or more contrasty), turn off one tool at a time until you find the offender (one or many).

As @paperdigits suggested, it would be better having the dng (if that is possible) to see what’s happening.

Xavier Bartol,
Seems fair, and I have experienced green, magenta as as well as darker images as described.
But that does not explain why the same “foreign” adjustments applied one by one by me to my image in my own version causes an image with the same dark rendering while showing no changes in the progressing process. None of the tools were missing in my version. This can’t be a matter of incompatibility - it must be something else - and I suspect RT’s interaction with the DNG file.

That’s why we are kindly asking for the dng: maybe a developer could take a look and find out what’s happening.

Of course - here it is:

_DSC0870.dng (53.4 MB)

And this is the pp3 file created on my Win-system on the version of RT mentioned above - but without having any effect - except that the image turns darker.

_DSC0870.dng.pp3 (17.4 KB)

I just had a good look at your DNG file and the accompanying pp3.

I have multiple RawTherapee versions and I checked it in 5.8-2735 (a development appimage) and 5.8-2926 (very latest development build).

I started with applying a Neutral profile after loading your DNG and loaded your pp3 on top of that. That should give me a clean starting point.

I see the following parts that have changes:

  • Exposure. Very subtle and very hard to see, but I do see a difference.
  • L*a*b Adjustments. This one is very obvious to see.
  • Impulse Noise Reduction. You need to really zoom in to see the difference, but is is visible.
  • Noise Reduction. Only turned on, so only chroma is effected. This is visible.
  • Defringe. This one I cannot check to see if it does anything in this image.
  • White Balance - The slight difference compared to Camera default is visible.
  • Local Adjustments (Blur/Grain & Denoise)
    • Grain: Cannot see the difference (this is correct: ISO is set to 400 but strength is 0)
    • Denoise: Minimal changes, but they are visible, especially when zooming in (500/600%)

So I can (not so clearly in some cases) see that all the changes are applied and I see this in both RawTherapee versions I tested with. But…

I do see a difference in the Local Adjustments affects: version 5.8-2735 is less visible/obvious as compared to 5.8-2926. I also have preview deltaE problems in 5.8-2735 (not at all in 5.8-2926). The version you are using lies between the 2 that I used, so I don’t have any idea which way your experience goes.

This was tested on a Linux system: Debian 10.9/Buster

EDIT: Just a little heads-up.

I mention being on version 5.8 2926, this is a bit skewed though. I have merged 2 still exerimental branches which inflates the commit count. The “true” commit count at the time I posted this was 2850 Sorry for the possible confusion.

Same as @Jade_NL, here on a linux machine I can see subtle differences with the provided pp3.

There has to be some reason for that processing, but just resetting the «Lab Adjustments > Lightness» and the image is not so dark anymore.

Now, to check if a provided pp3 from another RT version just works for you, try this one:
_DSC0870.dng.pp3 (13.5 KB)

It should look on your screen close to this:

I’m not saying that you (or anybody) have to like this edit, but to prove that you can edit an image with an «external» pp3 and it still look fine.

And even more, I have checked it with RT v.5.8 (release version), and it looks the same (here).

I like it, Xavier - it’s as good as any of my own attempts.

Your pp3 seems to work here too. This is a screen copy after Neutral and then loading your pp3:

I will check again, IF … I have sent you the right pp3 and from which version it origins. My friend, working on a Mac, and being a very qualified RT-user, wanted to demonstrate the impulse noise reduction. The result was expected to look like this:

Apparently made in version RawTherapee 5.8-2813-gbb08b6bed.

This is the pp3 I received. I could have misunderstood this pp3 to be the final suggestion, but maybe it is actually showing a preliminary state:

Eigil(2).dng.pp3 (17.0 KB)

This pp3 gives me the unexpected - darker - result.

I will investigate whether this was meant to be the final result (as demonstrated in the screen copy right above), or if I accidentally received a wrong pp3 (or misunderstood the meaning of it).

Will be back with the answer.


Hi Jacques,

You are probably right. See my reply to Xavier.


I just looked at the Eigil(2).dng.pp3 and it also shows up darker then the expected result you posted. This sidecar and the previously posted _DSC0870.dng.pp3 are basically the same. White balance and Demosaic are different though.

BTW: It should not matter at all if this is shown on a Mac, Linux or Windows based machine.