Digital BW, The Print

Dear all,

Lately I have been exchanging emails with Ctein * on the subject of
rezzing-up and upsampling and resolving what Jeff Schewe and the late
Bruce Fraser have basically said: «If the native resolution of an
image fall between 180 and 480ppi, just sent the file to the printer»

After reading his emails a few times, I decided to share with you his
observations, which, IMHO, may be of interest to the members this
list. As for myself, I think I've learned a lot from his emails and
will certainly review the way I print.

"Dear Andre, 

Broadly, I agree with Schewe and Fraser. 180 PPI is a bit low for my
taste for smaller prints (up to 11 x 14). I've got pretty sharp eyes
and I can see the pixelation at that resolution. 240 PPI and above
really isn't a problem. If you're talking about larger prints, 180
would be fine. As for the high end, there's really no reason to ever
resample, unless your computer is choking on the size of the file. It
doesn't matter whether you send the printer 500 PPI, 1000 PP or even
2000 PPI; it'll simply print as much detail as it can and that's it. 

Related to that is the "base resolution" issue. That term is something
of a misnomer with modern printers. All of the "photo-quality"
printers can reproduce considerably finer detail than 300/360 PPi.
Even the Epson 2200, the least sharp printer I've tested in the past
five years, reproduced 450 PPI; the best printers I've used can
reproduce 800-1000 PPI worth of fine detail. If I have a higher
resolution file, I would never down-sample it to a mere 360/300 PPI
for output; I would just be throwing away detail. 

There are VERY subtle artifacts that turn up with "non-native" PPI's,
but I can only see them in carefully designed and very artificial test
charts and they're down near the single-pixel level. They're
completely invisible in a regular photograph. It's an entirely
ignorable effect with modern printers. 

~ pax \ Ctein" 

----------------------------------------------------------------------------------------------------
It's just Ctein, no "Mr." in front of it. We're all friends, here. 

I'm glad I could be of help. Yes, there is a lot of old, obsolete
"wisdom" out there. Plus there is some accurate yet misleading testing
going on as well. Some folks use special test targets to determine the
point where the printer starts introducing artifacts into the image
and assert that that's limiting resolution of the printer. That's only
true if you're using it for scientific purposes where you need 100%
perfect fidelity of reproduction. From a photographer's point of view,
the limiting resolution is the one where throwing finer detail at the
printer doesn't give you any more detail in the print, and that's
several times finer detail than the level at which no distortions at
all occur. 

~ pax \ Ctein 

-----------------------------------------------------------------------------------------------------
I don't believe I wrote anything particularly embarrassing [ smile ],
so feel free to quote me. 

I read Mike's column at the link you provided, and I think he gives a
pretty good assessment of the situation. There is one important point
he didn't mention; printing to a dye-sublimation printer is different
from printing to an inkjet printer. There are good reasons to
re-sample an image to the native resolution of a continuous tone
printer like dye-sublimation, because they really do operate with a
fixed pixel pitch. There is never an advantage that I can see to
downsampling an image to the "native resolution" of an inkjet printer.
Sending less detail to the printer just to match the native resolution
will always produce a less detailed print and won't make a visible
difference in terms of odd artifacts. 

~ pax \ Ctein 


* Ctein is a photographer, master printer and scientist. He has a
double degree from Caltech in English and Physics.

Mike's column refers to Mike Chaney.

I experimented with 360/720 on and off for years, and when I got my beater 3000 I send it 180. Like many suggestions here, results may vary, but it gives you a place to experiment from. 

gar

*********** REPLY SEPARATOR  ***********

On 6/3/08 at 12:04 AM Andre Moreau wrote:

>Dear all,
>
>Lately I have been exchanging emails with Ctein * on the subject of
>rezzing-up and upsampling and resolving what Jeff Schewe and the late
>Bruce Fraser have basically said: «If the native resolution of an
>image fall between 180 and 480ppi, just sent the file to the printer»
>
>After reading his emails a few times, I decided to share with you his
>observations, which, IMHO, may be of interest to the members this
>list. As for myself, I think I've learned a lot from his emails and
>will certainly review the way I print.
>
>"Dear Andre, 
>
>Broadly, I agree with Schewe and Fraser. 180 PPI is a bit low for my
>taste for smaller prints (up to 11 x 14). I've got pretty sharp eyes
>and I can see the pixelation at that resolution. 240 PPI and above
>really isn't a problem. If you're talking about larger prints, 180
>would be fine. As for the high end, there's really no reason to ever
>resample, unless your computer is choking on the size of the file. It
>doesn't matter whether you send the printer 500 PPI, 1000 PP or even
>2000 PPI; it'll simply print as much detail as it can and that's it. 
>
>Related to that is the "base resolution" issue. That term is something
>of a misnomer with modern printers. All of the "photo-quality"
>printers can reproduce considerably finer detail than 300/360 PPi.
>Even the Epson 2200, the least sharp printer I've tested in the past
>five years, reproduced 450 PPI; the best printers I've used can
>reproduce 800-1000 PPI worth of fine detail. If I have a higher
>resolution file, I would never down-sample it to a mere 360/300 PPI
>for output; I would just be throwing away detail. 
>
>There are VERY subtle artifacts that turn up with "non-native" PPI's,
>but I can only see them in carefully designed and very artificial test
>charts and they're down near the single-pixel level. They're
>completely invisible in a regular photograph. It's an entirely
>ignorable effect with modern printers. 
>
>~ pax \ Ctein" 
>
>----------------------------------------------------------------------------------------------------
>It's just Ctein, no "Mr." in front of it. We're all friends, here. 
>
>I'm glad I could be of help. Yes, there is a lot of old, obsolete
>"wisdom" out there. Plus there is some accurate yet misleading testing
>going on as well. Some folks use special test targets to determine the
>point where the printer starts introducing artifacts into the image
>and assert that that's limiting resolution of the printer. That's only
>true if you're using it for scientific purposes where you need 100%
>perfect fidelity of reproduction. From a photographer's point of view,
>the limiting resolution is the one where throwing finer detail at the
>printer doesn't give you any more detail in the print, and that's
>several times finer detail than the level at which no distortions at
>all occur. 
>
>~ pax \ Ctein 
>
>-----------------------------------------------------------------------------------------------------
>I don't believe I wrote anything particularly embarrassing [ smile ],
>so feel free to quote me. 
>
>I read Mike's column at the link you provided, and I think he gives a
>pretty good assessment of the situation. There is one important point
>he didn't mention; printing to a dye-sublimation printer is different
>from printing to an inkjet printer. There are good reasons to
>re-sample an image to the native resolution of a continuous tone
>printer like dye-sublimation, because they really do operate with a
>fixed pixel pitch. There is never an advantage that I can see to
>downsampling an image to the "native resolution" of an inkjet printer.
>Sending less detail to the printer just to match the native resolution
>will always produce a less detailed print and won't make a visible
>difference in terms of odd artifacts. 
>
>~ pax \ Ctein 
>
>
>* Ctein is a photographer, master printer and scientist. He has a
>double degree from Caltech in English and Physics.
>
>Mike's column refers to Mike Chaney.

Thanks for this post. The emails from Ctein were very interesting and 
useful.

Scott
http://www.scottjonesphoto.com

Some things I still miss in comments like that.

First, downsampling is hardly mentioned while it happens
enough. If anti-aliasing algorithms are absent or of a low
quality in the driver's extrapolation it can influence the
quality of small prints considerably. Much depends on the
image content.  If the application takes care of the
downsampling (some do that on the fly in printing, like
Qimage) and sends the image data at native resolution
to the printer the same applies to the algorithms used in
that application. So Ctein's line that 500 to 2000 PPI files
can be thrown at the printer and it will simply print as
much detail as it can is not working on downsampling. Worst
cases are when a 1200 PPI file will be printed through 600
PPI native resolution of a printer, same for 600 to 300 PPI.
or 720 to 360. It is then not the user who deliberatey
downsamples to meet the printer's native resolution but this
happens automatically in the driver or application when the
print size and the quality settings tell the driver that a
certain native PPI is requested.

http://www.xs4all.nl/~bvdwolf/main/foto/down_sample/example1.htm

The minimum of 180 PPI to send to a printer. While no
upsampling can add information to a file that has not enough
data there's a lot of difference between upsampling
algorithms on how pleasing the effect is to the picture. If
Ctein more or less says that modern printers can lay down
between 450 to 1000 PPI worth of detail (and he can still
see that) then the different qualities of upsampling methods
can not be ignored either. There are enough printer drivers
and applications around that are not up to date in their
extrapolation algorithms to proof that. The MP number of
cameras is increasing and some have even improved the pixel
quality parallel to that but at the same time the demand for
larger prints is growing too. There's more printed that
crosses the 180 PPI limit downwards than that meets any of
the 1000-2000 PPI numbers above. Like in Jeff Schewe's
comments on the Luminous Landscape pages-forum I see here
deluxe PPI numbers quoted that may be their practice
(possibly with digital MF backs, MF and LF scans,  etc) that
may make the differences in output on large prints
negligible but I don't think that is practice everywhere.

Qimage made it easy for the user to cope with "native
resolution" to a degree that he/she doesn't have to know
what the native resolution of the printer is or can even
forget the whole concept of native resolution. That's a good
thing to a degree. For the users who are interested the
preview window and the max extrapolation setting still tells
them what the current native printer resolution is + the
print queue tells them what PPI the image has at that scale
and they can adapt their printing to that if necessary
(system memory, speed of processing and printing, required
quality in the print, quality to expect, etc). I do not
think it is wise to brush the concept of native resolution
under the carpet for printer users in general and tell them
that they are save between 180 and 2000 PPI input
resolution. Especially not for users without Qimage working
for them. It doesn't improve their understanding of what
goes on in the printing process and it may deliver them
prints that are not optimal. Users that ask questions about
native resolution show at least some desire to learn and
shouldn't be send back into the bush.

Could it be that this simple "send it 180 to 2000 PPI and
don't ask more questions" has it origins in the frustration
of Mac users who are waiting so long for an application like
Qimage to appear on what they think is the only genuine
graphic platform around? Of course the Qimage's GUI must be
disgusting for a Mac user but is that enough to abandon its
concept as a whole? A concept that both makes the job easier
but also shows a bit of the underlying process. It is in my
opinion a strange omission that on the LL pages not a word
is found on Qimage or any review of that application. Often
enough the advice to use it appears in the forum there but
that's contributed by the forum members and not picked up by
the LL staff. Mike Chaney is present on the Steve's Digicam
pages but at least major new version announcements and some
reviews appear on other photography sites as well. There's
competition between sites like there is competition between
sharpening, extrapolation plug-ins/programs and Qimage like
there is competition between Ego's. It is a bit harder to
understand why that should result in ignoring some aspects
of printing that matter.

-- 
Met vriendelijke groeten, Ernst


|  Dinkla Grafische Techniek  |
|     www.pigment-print.com    |
|             ( unvollendet )            |