Fuji Raw -> DxO PureRAW 3 DNG -> Capture One = 4 stops overexposed!

DxO PureRAW
1

SUMMARY:

When I convert my Fuji raw files with DxO PureRAW 3 to DNG, then edit them in Capture One 21, many of them (but not all) end up 4 stops overexposed in Capture One, even though the DNGs look fine in DxO and in the Windows Photos app (Windows Photos is probably displaying the JPG embedded in the DNG.).

Anyone have any idea why, or how to prevent this? Ever seen anything like this?

And I apologize for the length of this post.

DETAILS:

I shoot Fuji X-T5 raw RAF files, send them through DxO PureRAW version 3.7.28.0 for conversion to DNG + denoise + sharpen, then edit in Capture One 21 (two year old version). When I do this, some photos are very obviously 4 stops overexposed in Capture One, even though they look fine in other photo apps. (Again, I believe the other apps are displaying a JPG embedded in the DNG.) When I crank down the exposure 4 stops in C1 (-4 is the max adjustment), they look normal.

This seems to affect up to half the photos I shoot, though it varies. Weirdly, there appear to be no other differences between photos that process fine, and photos the end up overexposed in C1. In the C1 “photo browser” thumbnail screen shot below, the two photos were shot 5 seconds apart, with no changes other than to recompose slightly to the right.

Photo thumbnails in the Capture One Browser:

2 Thumbnails
2 Thumbnails361×624 46.6 KB

Yet the EXIF info is almost identical, normal left, overexposed right:
1 Properties

So even though exposures are essentially identical (f/5 @ 1/90 = f4 @ 1/140), the exposure in C1 is WAY off. See histograms:

XT5F1417-RAF_DxO.dng histogram, the normal photo:
3 XT5F1417 Hist

XT5F1418-RAF_DxO.dng histogram, the overexposed photo:
5 XT5F1418 Hist

XT5F1418-RAF_DxO.dng histogram, overexposed photo adjusted -4 stops:
4 XT5F1418 4 stop Hist

Unfortunately, the two-year-old Capture One 21 does not support Fuji X-T5 RAF raw files (though it does support older Fuji RAF files), so I need to convert Fuji RAF files to DNG to edit, plus I need DxO to denoise my high ISO astrophotography. I previously purchased the C1 version 21 perpetual license for my Fuji X-T10, and I don’t want to spend the $299 to upgrade to the current C1. (Ok, so call me cheap, but blew my budget on the X-T5, 10-24 zoom, & Rokinon 12 f/2.0!)

2

post .RAF + .DOP ( and your DNG too ) - that might help people if the answer requires digging into actual data

3

I am not sure why you did not post your raw files here, so I had to find them elsewhere – but as usual DxO PL does NOT work properly generating DNG files… for example ( and this is not just the one issue here ) if you bother to check DNG tags generated by DxO PL you see that DxO put ( in the DNG that is displayed “overexposed” in C1)

0xc62a BaselineExposure : 3.62

C1 being a nice code does respect this expo-correction

So the real question is why did DxO PL7 (in my test) puts this in linear DNG for that raw file

4

XT5F1417.RAF ( with 0xc62a BaselineExposure : -0.38 )

0x1031 PictureMode : Program AE
0x1403 DevelopmentDynamicRange : 100

and

XT5F1418.RAF ( with 0xc62a BaselineExposure : 3.62 )

0x1031 PictureMode : HDR → ???
0x1403 DevelopmentDynamicRange : 800 → ???

It seems this is why DxO PL generates 0xc62a BaselineExposure as it does differently between 2 of your RAF files

3.62 vs -0.38 = 4 stops difference that you see in C1

PS1: Adobe generates identical 0xc62a BaselineExposure tags
PS2: Iridient generates identical 0xc62a BaselineExposure tags

5

PS: the cure ( for this part of the general issue w/ DNG output ) is to put DNG after DxO through exiftool or exiv2 and nip 0xc62a BaselineExposure in the bud by zeroing it …

and/or watch what you are doing in your Fuji camera, because fat fingers clearly change shooting settings leading to tags in raw files tricking DxO into generating a lot different 0xc62a BaselineExposure tag

6

Sharing a file plus .dop can easily be done, even as a new user, by using things like google drive or wetransfer.

Maybe the original raw file is the weak link in the issue, it would therefore be helpful to get an example file. :man_shrugging:

7

original raw files were already examined and immediate reason found ( nothing to do with .DOP files btw) , however why DxO decided to go contrarian and to play with 0xc62a BaselineExposure tag in DNG output is the question left for DxO developers ( DNG output in general is not something that is done perfectly by DxO - examples are numerous , including infamous magenta tint issue , etc )

8

original raw files are perfectly dealt with by any other software out there - except by DxO in terms of generating linear DNG output

9

@parkerea, @noname, check out the differences I found in the original RAF files:

Bildschirm 2023-12-04 um 20.41.05.321
Bildschirm 2023-12-04 um 20.41.05.3213398×1850 729 KB

Bildschirm 2023-12-04 um 20.41.23.141
Bildschirm 2023-12-04 um 20.41.23.1413398×1850 774 KB

Nr. 17 seems to be a normal capture, while Nr. 18 comes up as HDR, the file size of which is about 3x the one of Nr. 17.

BTW, Lightroom Classic needs a -4 EV exposure compensation too.

10

I already posted the difference 20+ hours ago ::

11

this is just one more item ( scaling vs whitelevel, loosing unclipped data in raw, now this …) to the list why a user can’t really trust in DNG output from DxO tools w/o spending time investigating and THEN cleaning various tags in it before passing to other software ( and lost unclipped data is simply NON RECOVERABLE, period ) …

One can dream that DxO will be able to gather some resources to clean the act for PL v8 to become more operational in this matter

12

SOLVED!

I had posted this issue simultaneously to this DxO forum, and to DPReview’s Retouching forum. The DPReview forum found the issue is apparently that the underexposed photo was shot with Fuji’s HDR setting. Here is a link to that post:
DPReview Forum re Overexposed DxO DNG Output

I also neglected to repost here when I provided the original files via DropBox, so here is that link:
My Public DropBox Folder

XT5F1417.JPG | Normal JPEG, straight from camera.
XT5F1417.RAF | Normal raw, straight from camera.
XT5F1417-RAF_DxO.dng | Normal DNG, processed through DxO PureRAW 3.
XT5F1418.JPG | Overexposed JPEG, straight from camera.
XT5F1418.RAF | Overexposed RAW, straight from camera.
XT5F1418-RAF_DxO.dng | Overexposed DNG, processed through DxO PureRAW 3.

I included:

  • The “normal” file (XT5F1417) and overexposed file (XT5F1417)
  • Raw, DNG from DxO, and JPEG formats
    I assume the only files of interest are the two overexposed files, XT5F1418.RAF and XT5F1418-RAF_DxO.dng.

I have some testing to do to confirm the root cause really is the Fuji HDR setting being misinterpreted by DxO, but I won’t be able to do that until tomorrow. Assuming tests confirm this (and I would be shocked if they don’t), I will raise the issue to the DxO Help / Support system.

Thank you all,
parkerea