[argyllcms] Re: Black turning down problem (the -r trick)

  • From: Nikolay Pokhilchenko <nikolay_po@xxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Fri, 14 Jan 2011 22:03:56 +0300

Elena wrote:

> > It's clear. But... I suppose the several different, close placed in device
> > space patches may be better for profiling than the same amount but 
> > duplicated

> I think that depends on how the device/paper behavior is predictable, smooth,
> regular. If it is, then surely better to hit a big number of patches. If it's
> not, maybe half the patches but duplicated are better.

I believe that for an "ideal profiler" the big number of different patches is 
better, than the same number but duplicated.

> Recently I made a test profile starting from 3000 patches, then another
> profile with the original 3000 patches + other 3000 preconditioned ones.
> I saw no visual improvement. On the contrary, I perhaps noticed more visual
> errors in the second profile than in the first one.

In Your case behavior of the device is very sharp and broken. Higher patch 
value can only emphasize the sharp turns of device response.
When profiler have less points, it have to do more interpolation. There are 
smoother A2B curves as a result. When profiler have many points which are 
correctly measured, it have to do less interpolation and have to produce more 
sharp A2B curves. When A2B table resolution is limited, the quantization noise 
amplitude (erroneous bursts) increased.

Graeme, I see an analogy with audio analog-to-digital conversion (2D 
quantization - time and amplitude). May be the profiler must automatically 
limit the harmonic spectrum of device response bumps when the bumps spectrum is 
wider than can "digitize" the A2B. There should be the Nyquist-Shannon sampling 
theorem taken in account.
I suggest to perform low pass filtration (LPF) of device response data before 
computing of A2B table. The aim - limiting the spectrum bandwidth of device 
response to the bandwidth the A2B resolution can maintain. The filter can be 
with customizable bandwidth and rolloff.
There are obstacles - multidimensionality and non-discret patch placing in 
device space. But if You defeat multidimensional OSPF, the multidimensional 
Nyquist LPF is real.

Other related posts: