Digital BW, The Print

To be somewhat sloppy, 256 levels needs a 16x16 array of dots. Of 
course that is not quite true some printers emit more than one dot 
density from the jet. As well dot gain changes the result. So that 
on face value at 2880 the number of halftone patches is 180 per inch 
well except that the other dimension is not 2880. 

If you look at the NCA greyscale for the 2200 there is a flat toe of 
about 45 steps which suggests Epson is getting about 210 distinct 
ink patterns and probably less. In fact I think it has to be quite a 
bit less as the numbers do not compute or there are rather more jet 
harmonics than I thought.

So that at the level of the printer itself 8bits is by far enough. 
Even photoffset printers which were lith, ran at 2450dpi to get good 
150 (I guess I will say lines)per inch quality greyscale prints.

...John

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Roy 
Harrington" <roy@h...> wrote:
> --- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Paul D. 
DeRocco" 
> <pderocco@i...> wrote:
> > > From: Tom Husband [mailto:thusband@s...]
> > >
> > > When printing a 16 bit file does the printer convert it to 8 
bit?
> > > I've been trying to get the author of Qimage to allow 
processing of
> > > 16 bit grayscale files.  Qimage has a feature called "print to 
file"
> > > where you don't send the file to the printer but to another
> > > location.  It's a great feature but it takes a grayscale 16 
bit file
> > > and converts it to 8 bit RGB.  Anyway, Mike Chaney has 
responded with
> > > the following, "I'd also have to question why you'd want to 
print 16
> > > bits/channel when printing to file because printers cannot 
accept
> > > anything but 8 bit data (even when using a RIP). As far as I 
know,
> > > and unless things have changed recently, all printers (the 
hardware
> > > itself) are based on 8 bits/channel even when dealing with 
grayscale."
> > >
> > > I was thinking that it might happen when using the Roark 
curves and
> > > the Epson drivers but how about using QTR?
> > 
> > What goes across the cable to the printer (at least for Epson) 
is dots, not
> > 8-bit or 16-bit data. Standard printer drivers convert the image 
to dot
> > patterns, but only accept 8-bit data. Alternative software, like 
a RIP,
> > could in theory use all 16 bits, but I couldn't tell you if QTR 
does. For
> > B&W, it's possible that an extra bit or two might be useful on 
some media,
> > but it's doubtful that you'd see the difference for color.
> > 
> > --
> > 
> > Ciao,               Paul D. DeRocco
> > Paul                mailto:pderocco@i...
> 
> Both Mac and PC print systems are based on 8 bit pixel data.  So 
everything
> that gets sent through print commands will be truncated to 8bit 
pixels.
> 
> I would doubt this makes any visible difference -- our eyes just 
can't
> distinguish even 256 grays on paper.  I also think all the 
conversion to
> printer dots is the limiting factor by far.  Another thing to 
consider is that
> smooth areas of an image have lots of pixels and what you see is a 
blend
> of adjacent pixels not individual pixels -- a result of dithering, 
ink bleed,
> and eye resolution.
> 
> Roy