BMPC4K DNG Highlight Recovery problem

Hello BM Team!

I have a problem with highlights recovering. It was since I got my camera more then a year ago. The problem happens in every software (ACR and Davinci Resolve) so I want to know is it my camera problem or is it some DNG encoding problem? Does anybody has similiar problem or it is just me?

Problem with highlights is that recovery process goes with some errors. Better one time to see then a hundred time to tell so this is the picture (color temp an tint set to 'As Shot'):



Looks like the scopes are curled in recovering areas. And I get a magenta artifacts on the picture. Need to say that I can workaround this issue by tweaking the Color Temperature and a Tint of the DNG file. I can smooth the scopes and artifacts are almost gone but I end up with wrong color temperature and tint of the picture. After that I need to go and tweak the lift/gamma/gain back to catch the right color temp and tint. That is very annoying.

Tweaked picture:



So can you confirm the problem or is it just my camera produces that result?

This happens in every Firmware version and in every software I used (ACR and Davinci Resolve).

DNG file here: https://www.dropbox.com/s/iacb8kwifizcf ... 4.dng?dl=0

Can you recover highlights by bringing down just the highlights? Maybe go into log controls?


Sent from my iPhone using Tapatalk

No. The recovering in DaVincie Resolve with BMD dng files works only with checking of 'Highlight Recovery' option. Without it BMPC4K can be just a 10 stop's camera. I want to use all the dynamic range of the camera. I don't want to lose 1,5 stops of DR.
In ACR I can turn highlights lower by sliding down the 'Exposure' or 'Highlights' slider. But problem is just the same.
With my default Color Temp and Tint settings problem is presents anyways.

Yep, this is a known issue with Resolve. Pretty crazy that the industry-leading color correction software can't pull down highlights as well as After Effects (ACR). All my RAW footage has to be processed in ACR first to get back the highlights without the pink halo, THEN sent to Resolve. PITA.

Strange but in ACR I have very similiar problem. Pink tint. And I can workaround it shifting the tint to higher values and lowering the color temp too. So I think that this is not a Resolve but the dng coding or decoding issue.

Interesting, never got it in ACR. But either way, ACR pulls in about a stop and a half more in the highlights than Resolve does. I wonder if Blackmagic will ever address this.

Andrew, please, check your pm.

A slightly different approach to highlight recovery is addressed here:

http://liftgammagain.com/forum/index.ph ... post-58820

It might help solve the pink highlights and dynamic range problems.

"...when the information is destroyed in the sensor itself, nothing can bring it back."

Yep, this is a known issue with Resolve. Pretty crazy that the industry-leading color correction software can't pull down highlights as well as After Effects (ACR). All my RAW footage has to be processed in ACR first to get back the highlights without the pink halo, THEN sent to Resolve. PITA.

From what I can decipher based on observations alone - the "highlight recovery" mode in the raw tab uses non clipped colour channels as a basis for extra top end range. I would guess it would desaturate and duplicate the signal onto each channel so that there isnt a tint from having one or two channels worth of missing data.

As such, depending on the state of the colour channels (probably affected by colour temperature and light sources in the scene) it will work sometimes, but not others.


It is not a standard localized highlight reduction process such as you might find in ACR - such a thing can be found in the "highlights" control on the raw tab. We are talking here about "recovering" the clipped data by comparing it to a colour channel which hasnt clipped.

Paul thanks for the link!

Tom, according to the page of 127 of the Resolve's User Manual:

CinemaDNG is an open format capable of high-resolution raw image data with a wide dynamic
range, and is one of the formats recorded by the Blackmagic Cinema Camera when you shoot in
raw mode. CinemaDNG images are decoded with full dynamic range when the Highlight Recovery
checkbox is selected.

Highlight Recovery: A checkbox that lets you include additional highlight sensor data that’s
usually clipped by the standard decoding matrix. In cases where you have extremely clipped
highlights, you may obtain additional image detail this way, although it may have unusual
color artifacts.

So looks like this is DECODING problem.

Paul thanks for the link!

Tom, according to the page of 127 of the Resolve's User Manual:

CinemaDNG is an open format capable of high-resolution raw image data with a wide dynamic
range, and is one of the formats recorded by the Blackmagic Cinema Camera when you shoot in
raw mode. CinemaDNG images are decoded with full dynamic range when the Highlight Recovery
checkbox is selected.

Highlight Recovery: A checkbox that lets you include additional highlight sensor data that’s
usually clipped by the standard decoding matrix. In cases where you have extremely clipped
highlights, you may obtain additional image detail this way, although it may have unusual
color artifacts.

So looks like this is DECODING problem.

- yes, nothing there contradicts what I have said. The fact its in the raw tab just means that its derived from the bayer data, not video data.

Just another reason to go with redcode if you can afford it. CinemaDNG seems like an awful format, and stuff like this continues to leave me very let down working with the BMD cameras and DNG. I was working with a DNG clip yesterday, and I noticed after clicking highlight recovery, I started seeing faint pink boxes on my image. Very disappointing.

Just another reason to go with redcode if you can afford it. CinemaDNG seems like an awful format, and stuff like this continues to leave me very let down working with the BMD cameras and DNG. I was working with a DNG clip yesterday, and I noticed after clicking highlight recovery, I started seeing faint pink boxes on my image. Very disappointing.

So unclick it and try not to clip your highlights when filming. It is attempting to derive lost data which would otherwise be clipped.

Its not an argument about a raw format, its a little trick which the software can do which *sometimes* works.


CinemaDNG is a great format, it allows for proper bayer data in an uncompressed, lossless compressed or lossy compressed format. Its an open format and works great.

Just another reason to go with redcode if you can afford it. CinemaDNG seems like an awful format, and stuff like this continues to leave me very let down working with the BMD cameras and DNG. I was working with a DNG clip yesterday, and I noticed after clicking highlight recovery, I started seeing faint pink boxes on my image. Very disappointing.

I disagree with you here. DNG is a great format! It just BM handling it wrong. DNG is great high dynamic range format. REDcode is fine but costs 10 times more. It's just not my train to go.

You can easy change your color temp and tint to the stage when artifacts dissapear and then return the color temp back in next node by moving the gain sliders. Not ideal solution but it works.