Digital BW, The Print

Bob Frost: I remember Kennedy McEwan and that discussion. Didn't he
eventually get kicked off the Epson printer group for abusive comments?

My real life conclusions were similar to Kennedy's: anywhere around
360 ppi seems to work as good as anything. 

Bob Michaels

 --- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Bob Frost"
<bob@f...> wrote:
> Bob,
> 
> Several of us had a long discussion on this subject on the
Epson_printers 
> list about 2 or 3 yrs ago with Kennedy McEwan (if you search the
archives 
> for him, you will find them).
> 
> Summarising these discussions:-
> 
> Question -
>     Is there a magic number eg. 360 ppi for sending images to the
printer?
> 
> Kennedy -
>     "The Epson Stylus Photo printer range all resample the image to
720ppi 
> in
> the printer driver so, in principle there is a theoretical advantage to
> setting the output ppi to an integer division of this - eg. 360ppi.
> However, unless your image contains high modulation at the limiting
> resolution (which it certainly won't if you are scanning film at
> 4000ppi) then it doesn't really matter - there just aren't enough
> harmonics present in the image to produce irrational resampling
> artifacts."
> 
> 
> Question -
>     If the printer driver is going to resample to 720 ppi (desktop
printers) 
> and PS does a better job of resampling than the Epson driver, then
why don't 
> you recommend that we all resample our images in PS to 720 ppi before 
> sending them to the printer?
> 
> Kennedy -
>     "Several reasons, some of which are just as relevant as always, but
> others become less relevant as the technology moves on.  In that latter
> category are definitely issues about file size and processing speed.  A
> couple of years ago with less than 100MB being a lot of memory and
> processor speeds of the time, resampling to 720ppi would have been an
> unnecessary burden - if at all practical for many systems to cope with.
> With todays speeds and memory allocations, that is not as much of an
> issue, but a lot of machines will still struggle.
>     In the former category are consequences of the stochastic dither
> algorithm used by Epson, which can produce unwanted fine detail
> enhancement if sent data at full resolution.  This can often result in
> unexpected granularity in the image which is similar but not the same as
> grain aliasing.  Another aspect is the printer settings itself which may
> result in a slight scaling if set incorrectly, resulting in very large
> alias patterns which are well within visible resolution limits with
> 720ppi source material."
> 
> 
> Question -
>     OK then, what ppi should we send to the printer?
> 
> Kennedy -
>     "as long as the resolution is between 240 and 480ppi,
> then I just use what I end up with.  If it is less than 240ppi or more
> than 480ppi then I use bicubic resampling to get 360ppi.  If it is also
> greater than 720ppi, then I select 720ppi as the target, but this only
> occurs for small prints with the source equipment I use."
> 
> 
> 
> Nothing in life seems simple, does it?
> 
> Bob Frost.