George, ( A ) DxO PL code does not write 16bit uint digital numbers ( values in raw channels ) matching its own “0xc61d WhiteLevel tag when it creates linear DNG ( or does not write “0xc61d WhiteLevel tag with values to match how 16bit uint digital numbers were written ) and ( B ) DxO PL6 code clips data upon conversion to linear DNG where data was not clipped in the original raw - both for export to disk as DNG with “Denoise & Optical Corrections only” after applying DxO’s “No Correction” preset , it is a bug ( both A and B ), it needs to be fixed
BlackLevel tag ( there are several BlackLevel related tags used for calculations ) are used further when linear DNG is opened and processed in raw converter and then it does not matter if black level was zero or not as written in linear DNG … because for ( B ) what was not clipped in original raw was clipped by DxO and for ( A ) rescaling upon black level subtraction ( zero or non zero - does not matter ) when linear DNG is ingested will be wrong - welcome to Magenta Skies effect for example when you pull down brightness in a converter like ACR/LR that expects the linear DNG to be properly created ( and it is not )
WhiteLevel
Tag 50717 (C61D.H)
Type SHORT or LONG
Count SamplesPerPixel
Value See below
Default 2BitsPerSample - 1
Usage Raw IFD
Description
This tag specifies the fully saturated encoding level for the raw sample values. Saturation is caused either by
the sensor itself becoming highly non-linear in response, or by the camera’s analog to digital converter
clipping. The default value for this tag is 2BitsPerSample - 1 for unsigned integer images, and 1.0 for floating point images
Exactly what is in the dng from DxO. And it is 16 bit.
White level in the dng specs is specified as the max value a pixel can have: 255 for 8 bit depth, 65535 for 16 bit depth, and for floating point 1.
What makes this image so special that the white level should be different from what is specified in the DNG specs as default???
