In implementing proper support for DNG files from the Blackmagic Pocket Cinema Camera, there was a discussion on setting proper white points. I noticed an issue that might be complicate white point accuracy, some fixed pattern noise (FPN) at or right before clipping. I am not asking for a solution to the noise as an end user, as there are many ways to solve the issue, from dual demosaicing in RT to all the blur and masking tools in Davinci Resolve and Blackmagic Fusion, as well as not being a dunce and overexposing to begin with. Instead, I am raising this issue for the awareness of the developers, in case it is needed information in refining proper white point levels/whites behavior.
Here is a link to raw files of progressively increasing overexposure, Selected overexposure raw files BMPCC – Google Drive .
https://youtu.be/FvbiqIyLJrg This links to a demo of the BMPCC fixed pattern noise problem in the highlights near clipping in Raw. There is no artificial lighting that could add uncontrolled variables in the video. This was raw processed using Rawtherapee 5.5, as it has good tools with which to amplify noise for demonstration purposes, although I have also seen the when using the Davinci Resolve Raw engine to process the files as well, allowing us to rule out an issue with Rawtherapee.
In the test, first there is the whole frame on display with no digital zoom. Then is a 400% crop of the highlights with the noise pattern that stays consistent from frame to frame, but you can make out the clouds moving, as if the noise is a transparent screen in front of the lens. Last is a 400% crop of what noise is supposed to look like for reference, with each frame having a unique noise texture.
Also, Here are two of the dng files, as well as the processing profile used in the video.
Blackmagic Pocket Cinema Camera_1_2019-05-23_0422_C0006_000001.dng (2.0 MB)
Blackmagic Pocket Cinema Camera_1_2019-05-23_0422_C0006_000000.dng (2.0 MB)
Blackmagic Pocket Cinema Camera_1_2019-05-23_0422_C0006_000000.dng.pp3 (12.0 KB)