[argyllcms] Re: dispcal - blackpoint still corrected despite -k 0.0
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Thu, 28 Oct 2010 10:07:33 +1100
Anders Torger wrote:
So if I display an image with RGB color 0,0,0 (black) I would expect it
to be sent to the screen as 0,0,0 and not some non-zero calibrated
value, since I used the -k 0.0 parameter.
Your expectations are wrong.
1) The -k parameter changes how much the
hue of the black point should be corrected, to fall in line with
the hue of other neutral colors. If the hue of black is not close
to what is being aimed for, this will increase the black level,
since there is no way of subtracting light from black. So the
-k parameter has an indirect effect on the black level, but
setting it to zero does not force the output device values to
be zero at black, instead it makes the aim point of black be the same
as the native black point. (Calibration tried not to make assumptions
about the device response, but actually measures it!)
There may be a range of device values that give a response the
same as the native black response (ie. it may have a flat spot).
2) The vcgt is the calibration curve that makes the device appear to
respond in the desired manner. So the input to the vcgt curve
is the calibrated device values, and the output of the vcgt curve
are the actual device values needed to make it respond in the
desired fashion. The whole point of calibration is that a device
may not respond in a perfect fashion, so you simply cannot assume
that a non-zero input to the device produces some color other than black.
As I explained before, if the native device response has a flat spot
at zero, the correct calibration curve output will have to start
beyond the flat spot (ie. have non-zero values for black), so that
the resulting calibrated response is progressive.
Graeme Gill.
Other related posts:
