Re: Grayscale perception
Re: Grayscale perception
- Subject: Re: Grayscale perception
- From: Ray Maxwell <email@hidden>
- Date: Mon, 05 Dec 2005 22:59:46 -0800
Hi Mark,
I think there are three possible reasons why you might be seeing steps
in your film recorder.
The first is that the 12 bit LUTS may not have steps of equal visual
perception. It may be very non-linear with respect to vision. I would
guess that the output of the LUT is linear with respect to laser or
light head power. This means that it will be very non-linear with
respect to density on film.
The second is that you may be seeing banding in certain regions of range.
The third is that I qualified my remarks by specifying reflective media
which usually has a maximum density of 2.2 to 2.4. Transparent media
can reach densities of 3.0 or more. Therefore, it has a longer tonal
scale and can do a few more steps.
In the early days of Creo when they were building film recorders it
required extreme mechanical accuracy and a steep energy profile of the
laser beam to get rid of all banding. They patented the steep energy
profile and used the trademark "SquareSpot". Almost all other laser
heads or light heads used a gausian spot. It is very tricky to align
this type of head so as to not get any banding at certain power levels.
The focus on the imaging head requires temperature compensation and
servo to keep it in focus as the drum turns. If it is an internal drum
imager the same applies except you can substitute mirror rotation for
drum rotation.
Because your device has 12 bits and can produce 4096 steps does not mean
that they are equal visual steps.
There are lots of monitors around that have 8 bit LUTS that produce
posterization at certain levels. However, if you get a high-end monitor
with 12 bits going to the D/A, and you calibrate it carefully, you can
produce smooth gradients with 100 steps of 1% in the CMYK color space.
Digital quantization steps do not necessarily equal even visual steps.
Ray
Mark Rice wrote:
Ray, I have to disagree. I have been using an LVT film recorder for 20
years. It has internal 12 bit LUTs and 8 bit output. Since the beginning, we
have been plagued by quantization in the highlight regions. And the LVT is
the best film recorder - I have used the Lightjet as well, and it does not
have as smooth a gradation or as consistent calibration. I have measured the
gradations we tested to assure that there were no "jumps" in bit values. The
eye can detect this on a high quality film recorder. The only way to get rid
of it was to introduce noise, which is very similar to dithering.
One doesn't see this effect very often on printers for two reasons:
1. The dynamic range is about a factor of 10 lower than a transparency film
recorder.
2. Dithering is almost always introduced by the RIP, the driver, or the
device itself.
One of the things I have always wanted to try was to print a very
large sheet of paper on a Lambda or Lightjet with a perfect gradation in 8
bit grayscale, with no dithering, and see if any banding can be observed.
But with most RIPs and print drivers, and even inside the black box that is
the printer, it is usually impossible to tell if dithering is occurring.
Mark
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden