DNG files should all work out of the box

My new camera (Leica SL3) produces nice DNG files. Yet PL and PR don’t recognise it. (Adobe Lightroom does.). I am a little disappointed, as the lens profiles are associated with the lens, rather than the camera and the DNG files contain all the that you need to render an image.

I don’t mind if it’s described as Beta support until a human validates it, but I’m surprised that it doesn’t just work (for this, and any other camera supporting DNG raw files).

Hi, this page might help: https://www.adobe.com/creativecloud/file-types/image/comparison/dng-vs-raw.html

By Adobe’s own words, “DNG raw” cannot exist, as they are different.

There are several important differences between the two file types, especially in how to edit and store them.

Also have you checked if your camera is listed as supported: Appareils supportés - DxO

??? This will come as most surprising news to the people at Leica Camera AG headquarters, Wetzlar.

See also: DNG | dpBestflow

That is a very interesting article. Thanks for sharing it.

They do however say " DNG and RAW files are similar in that they are both versions of raw image files. However, DNG is stored in a format that’s compatible with most editing software, and the file sizes tend to be smaller than RAW."

I know, just let me open the darn DNG files! This is what DxO says about supporting new camera / lens combos and they have proved remarkably stubborn about their approach.

Hopefully, the wait, painful for sure, will be worth it.

1 Like

One may try exiftool -model=“LEICA SL2” filename.dng (better do it on a copy).
It seems to work on a sample DNG from dpreview.com but side effects may be unpredictable. In particular colors might be not quite right. Additionally you may try ACR DCP profile.

Very dirty workaround.

Hope I have contributed something.

Similar = not the same = different. So you should change your “They do however say…” to “And they say…” It’s also not “however they are smaller” but “and they are smaller… and jpgs are smaller too” - not the same (different).

very dirty :smiley: … different sensor… pixel count… processor

DNG files should all work out of the box

False.
One of DNG purposes was to include in XMP metadata specific to ACR, Lightroom, Photoshop. DNG is marketed by Adobe as “a non-proprietary file format”. This is false because of at least two reasons:

  1. Some data in DNG may still be in proprietary format, most notably the camera MakerNotes.
  2. There is no public full specification of DNG reader. For example, definition of BayerGreenSplit DNG tag starts with ‘This tag specifies, in arbitrary units, …’. You can find many more such examples, where the interpretation of data relies on the agreement between DNG maker and Adobe.

If you want more reliable sources than me, here’s a short opinion of Iliah Borg, author of libraw, copied from htps://www.dpreview.com/forums/thread/4663487?page=2
Yeah, DNG is anything but generic. Its metadata compels an Adobe-centric workflow, with its own definition of dual-illuminant color primaries and the incorporation of look processing.

Another opinion https://www.dpreview.com/forums/post/62907280 :
'It has been pushed by Adobe, which has a history of successfully making their proprietary formats “open standards” that take advantage of options to make their implementation superior to others. In other words, it’s a trap. You convert to Adobe DNGs and you’ll forever find non-Adobe alternative editing software doesn’t do better nor even as well…’

Standards have their business side too.

False.
DNG files, as many camera RAW files, are based on TIFF file format, just adding some new ‘tags’. If the DNG file contains IFD with PhotometricInterpretation set to 32803 (CFA), then it contains truly RAW data (Leica SL3 DNG samples on dpreview do have such SubIFD). If this tag value is 34892, it is considered as “LinearRaw”, a partially demosaiced image, but still in camera ‘native color space’ (‘semi-RAW’). Some other values correspond to RGB or monochrome image which is “display ready”, non-RAW. Consult DNG specs.

Yes. Often it doesn’t work, but sometimes it seemingly does. There was a post recently about Sony A9IV, where changing the ‘model’ tag solved similar problem, at least on the surface. Main concern is with camera native color space, which may be different.
Also corrections related to lens will probably work suboptimally but I have no idea to what extent. As for the image dimensions, they are read from other fields and as most other things are not incurred from the camera model string.

Note that Adobe Camera Raw still doesn’t have matching camera profile for LEICA SL3 ( Cameras supported by Camera Raw (adobe.com) ), so Adobe’s support is quite limited.

If you buy an expensive, freshly released camera, you are a good candidate for a beta-tester. Beta-testers can do great job for the rest of community, if you denoise their output.
So, if the above workaround does the job for OP, why not?

BTW, Leica SL3 uses different makernote signature (0x0a00) than SL2 (0x0900) but I think it would be too risky to change that too. Probably PhotoLab 7.5 is happy without it but I can’t tell what may go wrong because of it. Test.

semi means partially = different

Did you mean Linear DNG?

To quote DNG specs:
The enhanced image data should always use a PhotometricInterpretation of 34892 (LinearRaw). It should use the same color space as the raw image data.

Linear DNG is commonly used for DNG with LinearRAW IFD inside.

That is not my experience. In all cameras I used offering RAW or DNG, the DNG file is always the bigger one. Except you’re talking about uncompressed RAW? Lossless compressed RAW is smaller, easier to store, faster to read.

Cameras must use fast compression, so they are not very effective in terms of size reduction. Loseless camera compressions are usually Huffman based. ‘Offline’ DNG converter may use deflate, making the file smaller. So, it depends.

My cameras are no ‘offline’ DNG converters, and I would not trust any of them to not damage the RAW information if I’d let the ‘better compression’ do by them in post. I thought, we’re talking about SL3 at the moment. I’m sure your point is valid, it’s just not helping the poor early adopter to read the brand new camera’s files in PL :wink:

And about the very dirty way of sailing under false flag: Is it worth the effort? The sensors of SL2 and SL3 are very different. One can always try, sure, but maybe then also trying Sigma fp-L which also has a 60 MP sensor (I guess the same is working in Sony α 7 RV ?) Or is the Leica M11 already supported by DxO? That might also be worth a try?

Interesting…but written in 2015, close to 10 years ago, which shows occasionally.

Meanwhile, DxO has added DNG support which is limited to the “supported gear” paradigm. I can understand some reasons behind DxO’s voluntary restrictions in that area, but as long as they keep that ball and chain, they’ll never be able to take off.

Yes it is and perhaps M11 would be a better choice than SL2. To be clear, I’m not a Leica expert, so don’t trust blindly my choice.

Checked a sample Leica SL3 DNG from dpreview. It’s RAW image inside is compressed with ‘lossless Huffman JPEG’. Standard ZIP makes the DNG only 0.4% smaller, so the camera compression is quite effective.

We are many who have used HEX Editors and batch processing tools to change the model codes in our Sony files in order to get Photolab to accept the RAW-files from new cameras that lacked profile support in Photolab. We saw no other way when it took DXO six full month to add support for the A7 IV-files. Even Nikon and Canon users have done the same.

It doesn´t really solve the problem with either the A9 III files or DNG made from A9 III files more than it did for me converting the model code of my A7 IV files to the one of the predecessor A7 III.

The strange thing with some converters like Photolab is that they don´t open the DNG-files either despite the DNG Converter have suppport for these files since there also have to be a profile for these RAW-files from which the DNG-files are made.

So DXO doesn´t let us in with these files but once we have foolded Photolab with an old profile code to read these files we are totally free to pick any of the hundreds of profiles available in Photolab like the similar for A7 III but I could as well use the profile for Canon R5 without a problem if I am pleased with the effects of it. Pretty unlogical if you ask me.

I still think it has to be up to the uses to open their files without a camera profile if they wish. The only thing that changes is really that we would get a slightly different starting point for our editing. I have hard to see any differences that matters opening a Sony A7 IV-file with an A7 III profile from one with the A7 IV-profile. What is the big deal?

Doing this would also open for the import of scanned files in DNG-containers from for example Vuescan. I would be very happy if that could have been possible.

I think that one of the problems here is that DxO does a lot of key processing stages (Prime NR) as part of the demosaicing. On the one hand this gives them some technical advantages, but it also means that adding new camera support takes a lot more work and frequently never happens.

To pick an example, I use a Leica Q2 Monochrom. This has the same sensor and lens as the standard Q2, only without the colour filter array. The output files are the “same” DNG as those from the Q2. Only they are not.

The DNGs in fact use a (sensibly enough) gray-scale coding scheme that stores the pixel data differently to a Bayer colour camera. This means that if you simply change the DNG tags to mimic a plain Q2, rather than allowing the files to process you just end up with a cryptic error trying to read the file. There is no simple work-around for this, because DxO simply does not have code to parse this particular flavour of DNG.

I understand that Prime NR is made as part of demosaicing, but I do not know if any part of the lens corrections are. If not, it would be immensely useful to have a fall-back for unsupported cameras which might not support the full DxO image quality or feature set, but which would at least allow you to use damn tools while you are waiting for DxO engineering to catch up with the latest camera releases.

For the Q2M specifically, I would love to be able to process the images in DxO through a fall-back mechanism, even though it would obviously mean a slight reduction in quality and the loss of DPNR etc. The only correction in this case that is essential is the lens geometry adjustment. I can do this already by using dcraw to convert the DNGs to TIFFs and then finishing the processing in PL7, but the process is unnecessarily awkward and the lens corrections have to be dialed in manually.

Complete lack of support for a camera is far worse than providing a possibly sub-optimal use case that preserves workflow until (if) full support is released. It is a key reason why for people like myself can not use PL7 as a their only editor and suspect that it must be costing DxO revenue as if I have to pay for and use Lightroom because a camera is not supported, is it really also worth paying for PL?

1 Like

Fully agree. I’d really like some generic editing for unsupported gear. Almost every other app can do this and DxO would (just) need to overcome their self-inflicted stubbornness.

DxO software (still) makes a big difference with supported gear, for everything else, the cost is much too high imo.

BTW: The above is not meant for DNG only.