Capture One 10 & Hasselblad DNG: Controversy and Solution

[etto1972]

The arguments that they don't totally support the DNG 1.4 standard because they have blocked Hasselblad files to me is absurd.  Phase One is making folks in the market for a medium format camera system to really weigh the options of what RAW processor they can use with some other brand cameras and if it's worth it not being able to use Capture One.

Best regards,
G.A.

The software  can supports DNG 1.4, but the app blocks the DNG based on the exif manufacture name

I told you, you took the last "d" out and it works
Same with the DNG from the H5D, H6d and even from the 645Z !

[BobShaw]

etto1972
1. Phase One doesn't want anyone to buy a Hasselblad digital back, they want them to purchase a Phase One product, likewise Hasselblad doesn't want anybody purchasing a Phase One product.


2 Isn't it clear to you that the superiority of Capture One over Phocus is a selling point for Phase One and Leaf backs making a compelling reason to purchase a Phase One/Leaf back over a competing product?

1. Hasselblad actually makes cameras that support Phase backs and others as well, being the HX series. I'm not really sure why, but they do.
2. Phocus is a raw converter / tethering tool and doesn't pretend to be anything else. In this I think it far exceeds Capture One. Capture One is a jack of all trades, IMO better than Lightroom, but still master of none. I am not convinced on the reliability of the Capture One DAM, and of course it doesn't support Hasselblad files, so I don't use it. If it did support Hasselblad then I may buy it purely for the DAM. The Aperture programme I bought once ten years ago does, so that is where I stay.

[ErikKaffehr]

Hi Bart,

Colour profiles, dead pixels etc are a part of the DNG specification. Sensor tiles I don't think so. Diffraction is aperture only. Lens correction parameters are also part of the DNG spec.

If you implement reading of DNG correctly, you can process any image from any camera that has been converted to DNG. It is just another raw format, except being well defined.

Anyway, the discussion is not about Capture One supporting DNG or not but about blocking certain cameras based camera names. Just changing spelling solves that problem. It may be that inferior conversion may result using DNG, but I wouldn't think that is the case.

If a Phase One customer has extensively used Capture One and plans to use a non Team Phase One MFD he needs to ditch his experience with Capture One and his frustration is understandable.

Just to check out, I took one of the few ARW images I have from my A7rII, exported to DNG. Opened both images in Capture one, made some edits, copied the adjustments, exported both images and opened them on top of each other in Photoshop. No real difference when flipping layers on.

Than I made the ultimate test, subtracting the layers. The image came up black!  It was not as simple that C1 was using just the pixel data, as I tried to copy the white balance from the ARW image to the DNG image and that failed miserably. That means Capture One processess an A7rII image near identically weather shown it as ARW or DNG.

The first enclosed image shows the two images layered, with a mask applied so part is ARW and part is DNG. The second one is the difference image. The histogram shows they are not identical.

Best regards
Erik

Sure, and a basic conversion will be the result, but that's not a highly optimized conversion (profiles, colors, mapping of dead pixels, calibration of e.g. sensor tiles, noise reduction settings for different ISOs, etc., and whatever other secret sauce is hidden in the EXIF maker notes). Lens parameters will also need to be separately defined because Adobe uses its own models for that, and that might also affect things like diffraction correction and other/future features. That would require shooting hundreds of images and tweaking of many settings, so it does take a lot of additional work.

It would also require a significant number of potential licensees to make the investment of resources worth while, and the GFX might create a tipping point, who knows.

Not that the basic conversion is that bad, but when pushing things to the limit the additional refinements will pay off.

Cheers,
Bart

[Bart_van_der_Wolf]

Hi Bart,

Colour profiles, dead pixels etc are a part of the DNG specification. Sensor tiles I don't think so. Diffraction is aperture only. Lens correction parameters are also part of the DNG spec.

Hi Erik,

DCP profiles are not compatible with ICC profiles and vice versa. Anders Torger explained some of the technical differences in the thread about his DCamProf tool. A simple observation of the reported WhiteBalance parameters in LR and C1 alone will usually show different values of Temp and Tint, not that that matter for using the applications, but it shows that under the hood there are differences in the colourmodels used. Diffraction is wavelength, aperture, and sensel pitch dependent if you want to process at the pixel level. The DNG lens profile information is in a different format than what C1 requires, so they cannot be used. They may also not cover some of the MF Lenses that C1 might have a profile for, including shifted lens data.

If you implement reading of DNG correctly, you can process any image from any camera that has been converted to DNG. It is just another raw format, except being well defined.

Good Raw conversion is more than simply extracting the sensor's ADUs or DNs per pixel. I can also tell you that for a programmer it can be very frustrating having to redo part of your prior work, when a file format standard is changed by another party. A Raw file will remain as it was shot, but a change in the DNG specifications may require a lot of checking whether new or changed fields can wreak havok.

Anyway, the discussion is not about Capture One supporting DNG or not but about blocking certain cameras based camera names. Just changing spelling solves that problem. It may be that inferior conversion may result using DNG, but I wouldn't think that is the case.

If a Phase One customer has extensively used Capture One and plans to use a non Team Phase One MFD he needs to ditch his experience with Capture One and his frustration is understandable.

Yes, and I understand the frustration, but I also see the commercial risks involved for C1.

Just to check out, I took one of the few ARW images I have from my A7rII, exported to DNG. Opened both images in Capture one, made some edits, copied the adjustments, exported both images and opened them on top of each other in Photoshop. No real difference when flipping layers on.

Than I made the ultimate test, subtracting the layers. The image came up black!  It was not as simple that C1 was using just the pixel data, as I tried to copy the white balance from the ARW image to the DNG image and that failed miserably. That means Capture One processess an A7rII image near identically weather shown it as ARW or DNG.

The first enclosed image shows the two images layered, with a mask applied so part is ARW and part is DNG. The second one is the difference image. The histogram shows they are not identical.

Sometimes the differences are larger sometimes smaller. I wouldn't be surprised if having a camera model like the A7rII in the database would help in getting better results from the DNG as well, because there is more knowledge about specifics like non-linearities of tonecurves, CFA transmissions, noise, etc. .

Cheers,
Bart

[ErikKaffehr]

Hi Bart,

I am obviously aware that DCP profiles are not compatible with ICC profiles. On the other hand, I would presume that Phase One engineers are capable of doing a couple of matrix multiplications to convert DCP conversion matrices to C1 internal format. That jobs needs to be done only once, as DNG/DCP would use similar matrices for any camera. It does not encompass reverse engineering, as all bits are documented. Same applies to distortion correction Adobe uses different parameters and they are well documented. I am pretty sure Phase one can convert between the data.

What my little experiment with the Sony indicates is that Phase One has actually done quite a bit of that job. Their handling of a DNG converted picture gives indentical results to ARW conversion. It could be that Phase One doesn't use DCP information, but handling of WB is obviously different between the files.

Much of colour handling is simple linear algebra. Once the colour is converted to a working colour space, ICC profiling is applied to it. In the A7rII image, the very same ICC profiles seems to be applied both to ARW and DNG, which explains that the results are extremely close. But it also means that prior states match up really well.

It is quite possible that DNG conversion of Hasselblad files would not give optimal results, but it seems that just removing the "d" from Hasselblad in the EXIF is enough to have C1 to be able to process the DNG files. So it is not engineering that is the issue at hand. C1 seems gladly accept DNG from a "Hassebla" but not from a "Hasselblad". How many hourse has Phase One invested in "Hasselbla" support?

Best regards
Erik

Hi Erik,

DCP profiles are not compatible with ICC profiles and vice versa. Anders Torger explained some of the technical differences in the thread about his DCamProf tool. A simple observation of the reported WhiteBalance parameters in LR and C1 alone will usually show different values of Temp and Tint, not that that matter for using the applications, but it shows that under the hood there are differences in the colourmodels used. Diffraction is wavelength, aperture, and sensel pitch dependent if you want to process at the pixel level. The DNG lens profile information is in a different format than what C1 requires, so they cannot be used. They may also not cover some of the MF Lenses that C1 might have a profile for, including shifted lens data.

Good Raw conversion is more than simply extracting the sensor's ADUs or DNs per pixel. I can also tell you that for a programmer it can be very frustrating having to redo part of your prior work, when a file format standard is changed by another party. A Raw file will remain as it was shot, but a change in the DNG specifications may require a lot of checking whether new or changed fields can wreak havok.

Yes, and I understand the frustration, but I also see the commercial risks involved for C1.

Sometimes the differences are larger sometimes smaller. I wouldn't be surprised if having a camera model like the A7rII in the database would help in getting better results from the DNG as well, because there is more knowledge about specifics like non-linearities of tonecurves, CFA transmissions, noise, etc. .

Cheers,
Bart

[scyth]

I am obviously aware that DCP profiles are not compatible with DCP profiles. On the other hand, I would presume that Phase One engineers are capable of doing a couple of matrix multiplications to convert DCP conversion matrices to C1 internal format.

you absolutely do not need to use or support or even know about any dcp profiles to support raw conversion of DNG files in C1... just treat DNG like any other raw format and use own profiles in icc/icm format... and if DNG has some embedded dcp profile(s) - ignore it

[Bart_van_der_Wolf]

you absolutely do not need to use or support or even know about any dcp profiles to support raw conversion of DNG files in C1... just treat DNG like any other raw format and use own profiles in icc/icm format... and if DNG has some embedded dcp profile(s) - ignore it

Correct, one could ignore everything but the Raw sensor ADUs themselves, but that would then still require a serious additional effort to obtain a quality conversion. In addition, the idea that some have that a conversion from one (DCP) color system to another (ICC) is easy, is a bit too simple.

Cheers,
Bart

[scyth]

Correct, one could ignore everything but the Raw sensor ADUs themselves, but that would then still require a serious additional effort to obtain a quality conversion.

what P1 is asked to do is at least not to do anything different for /converted/ DNG raws vs non DNG raw files ... then 3rd parties will take care about the rest in terms of "camera profiles" and /pre/ processing of DNG raw data read from sensor if it is required to address certain MF related issues with sensors.

[ErikKaffehr]

Hi,

My understanding is that all raw converters use matrix multiplications to to convert RGBG channel data to an internal colour space. My understanding is that Adobe uses XYZ and it may be that capture uses some other colour space. At this stage, colour spaces are just a matrix multiplication apart. So Phase One does not need to reinvent any wheel. At this stage they can use a simple matrix multiplication to cast XYZ into their internal colour space.

My understanding is that the ICC profile is generated from and applied at this stage. But, at this stage we have pretty decent basic image. The profiles applied at this stage may be a kinf of secret sauce. Obviously it is quite possible to use the no so secret sauce from the DCP profile embedded in the DNG file. So, for generic DNG support the route to take would be:

Use DCP colour matrices for mapping RGBG data into XYZ colour space. After that apply DCP LUT tables correctly on that colour space and remap the image to internal colour space.

On another thread a poster has posted a Pentax 645Z image processed in Capture one after some manipulation of EXIF data and he has also posted the DNG image. That image may arguably look better than the same image processed in LR using AdobeStandard DCP, but if you use the DCP embedded in the file you get an image very close to the C1-conversion.

I don't know if Phase One has spent a lot of development time on a camera they never intended to support, but their default conversion on that unsupported camera is very close to Adobe Conversion using the embedded profile.

So, I guess that Phase One has made some work on properly supporting DNG.

In the attachements below Capture One is left side and LR is on right side.

Best regards
Erik

Ps. I cannot do these type of comparisons on my own images because all systems I own are officially supported by Phase One :–)

what P1 is asked to do is at least not to do anything different for /converted/ DNG raws vs non DNG raw files ... then 3rd parties will take care about the rest in terms of "camera profiles" and /pre/ processing of DNG raw data read from sensor if it is required to address certain MF related issues with sensors.

[scyth]

So, for generic DNG support the route to take would be

absolutely not - for generic DNG support read the camera model from DNG, if P1 has already their own icc/icm profile (one or many) for that camera model - then apply as usual as if it is a regular raw file... if not (unsupported camera model) - just apply some absolutely generic matrix+trc profile (actually P1 has one) and that's it - let that user deal with camera profiles themselves - create, borrow, buy, whatever - plain and simple... any attempt to support DNG files differently (by using parts or whole embedded DCP) in that aspect just complicate things

[BobShaw]

Those pushing the DNG path for Hasselblad files should read and study the images of the girl on the other thread on the similar thing.
Converting a Hasselblad file to DNG seriously alters the colour, the histogram and possibly other factors of the image compared to processing in Phocus.

[sandymc]

For those interested in going down the route of modifying the camera name in DNG files to fool Capture One, I've updated DNG Cleaner to be able to do this easily.

[pentax645z]

Those pushing the DNG path for Hasselblad files should read and study the images of the girl on the other thread on the similar thing.
Converting a Hasselblad file to DNG seriously alters the colour, the histogram and possibly other factors of the image compared to processing in Phocus.

No exactly

Yes the Hasselblad fff raw looks great on Phocus but its pretty disappointing on ACR
The converted hacked DNG in C1 is in between, not as good as the Phocus but clearly better then the original  Hasselblad raw in ACR

D

[scyth]

Converting a Hasselblad file to DNG seriously alters ... the histogram...

seriously ? histogram  ... indeed the displayed histogram is so important for raw conversion

[scyth]

Yes the Hasselblad fff raw looks great on Phocus but its pretty disappointing on ACR

make the better DCP profile then ...

[etto1972]

For those interested in going down the route of modifying the camera name in DNG files to fool Capture One, I've updated DNG Cleaner to be able to do this easily.

Unfortunately it doesn't work wit the Hasselblad X1D DNG files 
but Exif editor works !

[E.J. Peiker]

For those interested in going down the route of modifying the camera name in DNG files to fool Capture One, I've updated DNG Cleaner to be able to do this easily.

I love the concept of this piece of software.  Any plans for a Windows version?

[BobShaw]

seriously ? histogram  ... indeed the displayed histogram is so important for raw conversion

Thanks. Yes it is. The tonal range and the colour dispersion are shown there. In both the ACR and DNG conversions the highlights are noticeably clipped almost vertically.
I can't read the histogram but you can see the result is that the skin colour is green. That is more obvious in the pics on the other post.
Phocus allows you to set the levels and colour balance exactly with built in camera calibration. Making a camera profile after the shot is much more difficult.

[ben730]

Unfortunately it doesn't work wit the Hasselblad X1D DNG files 
but Exif editor works !

Yes, there is still the Brand: Hasselblad 

[scyth]

In both the ACR and DNG conversions the highlights are noticeably clipped almost vertically.

just fix the dcp profile for a start