Dear Gerhard, On Nov 26, 2012, at 9:20 AM, Gerhard Fuernkranz wrote: > > Hi Graeme, > > > Am 26.11.2012 03:02, schrieb Graeme Gill: >> But there's no guarantee that camera 1,1,1 is D50. > > Certainly not for the raw RGB data (w/o any white balancing or other color > transformations applied). > > But my feeling is that obviously many people want to build their camera > profiles from pre-processed raw files, which have already been white-balanced > by the raw converter. That is indeed the case; and due to increased noise in profiles built from linear data it is often the that before profiling some tone mapping is also applied. It is not always true that both green channels in the sensor are filtered in a similar way and given most of the profiling engines do not allow for easy support of 4->3 transforms it is the task of the raw converter to apply certain transforms to equalize the greens, usually during the demosaicking process. Running demosicking in linear space is problematic as it results in higher noise levels, more artifacts, and does not allow to separate luma and chroma well enough. > In the general case (arbitrarily non-linear device behaviour) This is always the case. Extreme highlights and shadows are non-linear for any sensor, and raw converters apply linearization before applying white balance; or white balance won't work well. For those sensors that have a "broken-line" response, like some Aptina and SONY, pre-linearisation in a raw converter is also applied. For sensors with logarithmic response curves, either pre-linearisation is applied or the white balance mechanism is totally different. > the profile needs to be applied first, of course, before white-balancing. Profiles are best applied to demosaicked data, and demosaicking needs to be run after white balance - given we want quality of course, not just speed. -- Iliah Borg ib@xxxxxxxxxxx