Digital BW, The Print

Some weeks ago, I picked up an Epson C86 and the MIS Ultratone EZN inks 
to fool around with black and white output. I'm following Paul Roark's 
recommended workflow at:

http://home1.gte.net/res09aij/C86-EZ-UT-Readme.htm

Specifically, I'm printing onto EEM with the C86 driver set to "best 
photo," with high-speed mode off.

Last night I printed an image with a snowfield and some sky. Both are 
fairly broad areas with all the pixels nearly the same tone; snowfield 
very light, sky about middle gray. I'm getting horizontal banding in the 
printout. It is clearly visible by eye, though it is small. The banding 
is more easily seen in the sky, but is definitely present in the snow as 
well. The bands are about 0.75mm apart and are about equally thick. The 
file is 320 dpi. On a casual glance, most people I show the print to do 
not notice the banding.

I ran a nozzle check (everything OK), and a head alignment check 
(everything OK). I printed the image again after this, just in case, 
with the same results.

I then downloaded and installed QTR and printed the image at the same 
size on my 2200, just to see if somehow this was inherent to the image 
and I couldn't see it on the screen. The banding does not appear in that 
print.

The C86 banding gives the impression of being the result of a bad dither 
pattern - perhaps the result of a buggy dither algorithm?

Does anyone have any recommendations on how to proceed? I'm not seeing 
this on other images I've printed with the C86, but none of the other 
images have such broad areas of a single tonality. I've wondered whether 
the output file resolution is causing problems, and I might run that 
through a spline to increase the resolution and see what happens. But I 
thought I'd check here first. Thanks!

--
Jeff Medkeff
Eagle River, Alaska

Hello Jeff,

>Does anyone have any recommendations on how to proceed? I'm not 
>seeing this on other images I've printed with the C86, but none 
>of the other images have such broad areas of a single tonality. 
>I've wondered whether the output file resolution is causing
>problems, and I might run that through a spline to increase the 
>resolution and see what happens. But I thought I'd check here 
>first. Thanks!

It seems typical for dither banding to show up in specific tonal
ranges.  From what I've gathered from looking at many prints from
print exchanges (it's pretty rare not to have any at all), the
specific zone can vary from one printer to another, even among the
same model. I'm not sure why this is.  I had an 870 once that always
had minor dither banding (required a glass to see it) at around Zone
III.  So one thing you might test is whether there is consistency
there.  Determining that this is or isn't the case may not gain you
anything, but it will be educational <g>.

Another avenue to explore is head alignment.  With my 2200 I
discovered that the least banding was obtained from a head alignment
number (2, meaning which line pair to choose from the print out) that
according to the instructions should not have been the optimal choice.
I just did an experiment one day and kept choosing a lower number each
time and making a print, and it kept getting better and better.  It's
almost imperceptible now.

I also get less banding with Hi Speed on, contrary to popular wisdom.
 Steve Karafyllakis was here a couple of weeks ago and I said this and
he was skeptical.  So I made two identical prints, one at each speed
setting, and let him examine them with a loupe, and sure enough, the
hi speed one was better (maybe if he reads this he'll jump in and say
something).  This may be unique to my printer...

Microbanding that is caused by a clogged nozzle looks really different
than dither banding.  It's easy to tell the difference when you look
at both examples together.

The only anomaly that I've seen respond to changing image resolution
is when there are wide bands (3/16") of light and dark running all
across the entire print.  If I remember correctly (it's only happened
twice), the bands run opposite to the direction of head travel.

I hope these will give some ideas.  I'd be interested in whatever you
find out.

Regards,
Clayton


Info on black and white digital printing at    
http://www.cjcom.net/digiprnarts.htm

.
> 
> I also get less banding with Hi Speed on, contrary to popular wisdom.
>  Steve Karafyllakis was here a couple of weeks ago and I said this and
> he was skeptical.  So I made two identical prints, one at each speed
> setting, and let him examine them with a loupe, and sure enough, the
> hi speed one was better (maybe if he reads this he'll jump in and say
> something).  This may be unique to my printer...
> 
My own experience has been that Hi-Speed on agravates microbanding, so 
I was a bit surprised to see better results from Clayton's printer with 
HS on. Goes to show you shoiuld never rule anything out without testing 
it for yourself. Generally, however, the highest dither pattern and HS 
off seems to do better for most people, but try the obvious- any 
combinations you haven't tested so far. 
 It might be usefull to know if the printer itself has a problem in one 
particular channel, or if it is an ink-mixing/crossover problem. Either 
print a four channel purge pattern (available on the MIS website) or  
download & install QTR and use the ink calibration mode to print out a 
test pattern. This will give you a separate scale for each head, and 
tell you if one is significantly different from the others. If you 
don't have a head problem, then you might try a different printing 
workflow, for instance Paul's curves for that inkset, or QTR itself. If 
it turns out that you do get consistent banding in one or more channels 
and nothing cures it, then your only real option is to try another 
printer. Umfortunately, there is simply a certain percentage of these 
desktop printers that get released inspite of consistent banding, as 
long as it doesn't normally show.

Hope this helps;

Steve Karafyllakis

Clayton Jones wrote:



> I hope these will give some ideas.  I'd be interested in whatever you
> find out.

Clayton, thanks for the thorough reply. It gave me several ideas about 
how to attack this issue, and I'm pleased to report I've solved the problem.

Making a long story short, after several blind alleys I got even more 
suspicious of dither artifacts when earlier today I reprinted an image 
that I had successfully printed a couple weeks ago. This print came out 
fine, which suggested to me that nothing was physically wrong with the 
printer, and that it wasn't a mysterious software change that had led to 
problems. But I did notice that this print's source file was around 500 dpi.

So I printed some test strips. The ones at 360 dpi and 720 dpi looked 
fine. The one at 320 dpi was banded in all but the darkest zones - or 
maybe it is more correct to say the banding was variably visible, being 
most visible in tones from just below middle gray almost to white.

Anyway, that seemed pretty diagnostic to me.

I resampled my troublesome image using the s spline from 320 dpi to 720 
dpi. I printed it out and it looks great.

I went back to the original 320 dpi file, and resampled it to 360 dpi. 
Printed it, and I can't tell the difference from the 720 dpi version.

I reprinted the original 320 dpi file without changing it, and it is 
banded all to hell again.

I'm concluding that this printing system appears to be sensitive to the 
source file's linear resolution.

I'm not sure in what way, because I haven't done enough tests - and I 
probably won't bother, I prefer to just raise the flag of victory and 
enjoy my ability to print. But I wonder whether 320 is a magic number 
where you get a poor print with this system, or whether the dither 
algorithm works best at multiples of 80 dpi and hence there are a lot of 
sub-optimal magic numbers, or what. I also don't know whether this is 
idiosyncratic of my system specifically, or whether it is an attribute 
of all C86's used this way. Anyway, I throw this out for those who might 
encounter the problem in the future.

Paul Roark, thanks to you also for the help you gave me off-list.

--
Jeff Medkeff
Eagle River, Alaska