Digital BW, The Print

Helen Bach wrote:
> --- In DigitalBlackandWhiteThePrint@yahoogroups.com, Ernst Dinkla
> <E.Dinkla@...> wrote:
>  
>> I think it is better to start with 720 PPI right away then, 
>> instead of creating a pseudo 720 PPI test image with 2x lines 
>> and squares if the extrapolation does nearest neighbour (not 
>> bad in this case).
> 
> Ernst,
> 
> I must have explained it badly, because that is exactly my point.
> The resolution tests you gave a link to did not print a true 360 ppi
> file. They printed a 720 ppi file with psuedo-360 ppi images in it -
> which is not a true representation of a 360 ppi file.
> 
> My tests were with files at true ppi values, no resampling, no
> interpolation. I'm testing the match between the file and the printed
> image. The best match between the file and the image occurred when a
> 288 ppi file was sent to the Epson driver and at 360 ppi with OPM.
> That's not necessarily the best file resolution to use, it's just a
> test for me to try to understand what the two drivers are doing.
> 
> If 720 ppi is the native resolution of a 2200, as Epson say, how many
> dots are there per pixel? How many dot sizes? How many possible
> colours? Why does a 720 ppi file get altered - ie why isn't one file
> pixel mapped exactly to one printer pixel? These are the questions I'm
> asking myself, not rhetorical questions.
> 
> Best,
> Helen  

Helen,

Sorry, did you print several checkerboards with different PPI 
files? Your first message isn't clear about that. If you did 
send a 720 PPI file with 720 detail, 360 detail and 180 detail 
as well then no extrapolation in the driver should happen. 
Then it doesn't matter that the 360 PPI target  in it is a two 
(720) pixel size fake one.  A 1440 PPI or 360 PPI file will 
get an unknown extrapolation and all the targets get 
extrapolation to 720 PPI. So in my opinion the Qimage page has 
the right one.

The best answers on the last questions you will get in the 
Gimp/Guten print sources and discussions. Usually the droplet 
size variation numbers are reduced the higher the dpi 
resolution chosen. Where 720 and 360 dpi may have 3 droplet 
sizes the 1440 and the 2880 may have just two or one. The 
highest resolution quoted as 2880 dpi could well be a 1440 dpi 
that is shifted half a dot pitch on the next line so not a 
true 2880 dpi. Happened with the 1440 dpi setting of older 
models. So computing 256 shades back to matrix positions, 
droplet sizes, diluted inks cmk, amount of weaving, may not be 
that easy. Another obscurity is the higher bleed of smaller 
droplets with the larger circumference/area ratio compared to 
that of the larger dots. In linearising and/or profiling we 
control that bleed on 4/6 channels, that either means less 
droplets are used or a smaller droplet is used when we want to 
reduce dotgain. There has to be some slack in the total to 
deal with that all.

On a single channel I doubt the 256 shades of 8 bit are 
actually available in the print. For example on the black and 
yellow of a 6 color printer. The first isn't used on the full 
range though, the yellow not a dominant color in definition. 
All is improving though.

Ernst

                    --
           Ernst Dinkla


www.pigment-print.com
(         unvollendet         )