Re: gEDA-user: PCB produces invalid postscript

[Karel Kulhavy <clock@xxxxxxxxxxxxx>]

On Wed, Apr 19, 2006 at 06:21:39PM -0400, DJ Delorie wrote:
> 
> > I asked what the gray is supposed to be. You say that it depends on
> > the gamma setting of my X software. Supposing is an action of human
> > mind. Does it mean that the person who decided the postscript
> > generation can not only remotely read the setting of my X server,
> > but can also do it against the flow of time?
> 
> No, I meant the software can NOT tell what your X server and monitor
> are set for, so it can NOT pre-determine exactly what "grey" means.

The first is true. The second not. It can pre-determine exactly what
"grey" means if it uses a device-independent colour space.

> 
> > What do you mean with "display 50% grey"? 50% of what? 50% of voltage
> > range on the monitor input, 50% of the numerical range in the video
> > card, 50% or numerical range in the X application's output range,
> > or 50% of maximum photon flux?
> 
> I'm sorry, I didn't realize you knew so little about calibrating
> monitors.

That's a completely irrelevant answer. I asked for something else:
"What do you mean with "display 50% grey"? 50% of what? 50% of voltage
range on the monitor input, 50% of the numerical range in the video
card, 50% or numerical range in the X application's output range,
or 50% of maximum photon flux?"

> 
> Display an image that, on one side, is filled a pixel value halfway
> between white and black.  On the other side, fill it with half white

A pixel has it's value in different spaces which don't map linearly.
You have to say which space you are talking about.

> pixels and half black pixels.  Now, adjust your X/monitor/whatever
> until the two halves are visually similar in intensity.  You now have
> a linear gamma response.
> 
> > This does have to do with PCB that I am asking what grey the grey
> > was meant to be. If you want to make the grey with little black spots
> > on white paper, how many % of area are the black spots going to
> > take?
> 
> You see, the problem is, you keep saying "the grey" as if there's only
> one.  There isn't only one.  You're not giving us enough information.
> Let me put on my mystic mind reader hat and see if I can figure out
> what you *really* want the answer to...
> 
> "How do you calculate the intensity of the faded colors used in the
> assembly drawing?"
> 
> For Postscript HID output it's 20% (pigment-wise) of whatever the
> original color was, in the case of the assembly drawing.  I.e. one
> part original color mixed with four parts white.  If you need to

If you mix one part black with four parts white then you get a mixture
whose reflectance is dependent on differences between extinction rations
of the original liquid solutions. So again, this is ambiguous.

Did you mean "covering 20% of the white area with black spots and then
looking from distance"?


> change it, search for CBLEND in src/hid/ps/ps.c

I don't mind if it's 20% or 40%. I just want it to be a constant that is
defined in a device independent way.

> 
> For the old Gtk version, it appears to be 10%.
> 
> Why don't you just look at the postscript output and find out for
> yourself?  Just search for the "setgray" or "setrgbcolor" and see what

Because it's impossible. The gray level from the postscript output will
differ, depending on what gamma the display device will have. Do you
know what a gamma is?

> numbers it's using.

CL<