Digital BW, The Print

harry, I have been trying to reply to your message you sent me but your address keeps bouncing.The message you sent about printer resolution was sent to me and not to the board. That is why it never showed up. I am copying the message you sent me below.Thanks. Well, Pierre, I believe you're opening a can of worms.  But let's go fishing!
What I "know" is what I've gleaned from multiple sources, some reliable, others less so.  And all my remarks are based upon Epson printers but can easily be transcribed to other brands.
Firstly, the native resolution of  (all?) Epson printers is 1440 dpi, not 360 dpi.  The choice of printing resolution then depends upon the final print size and the question of resampling, up or down, and as we'll see, viewing distance.
It is recommended, for instance, that the chosen resolution be an integer divisor of the native resolution that results in a rational fraction.  That is: 1440/2 (720); 1440/3 (480); 1440/4 (360); 1440/5 (288); 1440/6 (240); 1440/8 (180), etc.  Note that 1440/7 = 205.57, and is, therefore, not recommended.  The rationale seems to be that these choices will, as a first step, avoid resampling by an unknown algorithm in the Epson driver.
Furthermore, the conventional wisdom is that one should choose one of the above resolutions that avoids up-sampling in favor of down-sampling for the given print size.
For instance, suppose one has a scan of a 35mm negative at 4000 ppi (not dpi!) for a pixel count of ~5670 in the long dimension, and the intention is to make an 18"x12" print.  Then 5670/18 = 315 dpi (approximately).  So, printing at a resolution of 360 dpi would require up-sampling, and the conventional wisdom cited above would suggest, instead, printing at 288 dpi.
I have personally done this test, and I was able to see a small, but real, difference in quality between 360 dpi and 288 dpi for such a print.  However this difference was visible when I viewed the print a from a distance of a few inches.  From a normal viewing distance I would not have been able to see what I saw up close.  So, viewing distance is a vital ingredient in the mix as well.
The print driver subtracts pixels in down-sampling and adds pixels in up-sampling.  How the Epson algorithm accomplishes this is not in the public domain.  But I'm aware of third-party software that purports to use algorithms that optimizes the processes.  This would seem to support the contention that re-sampling is an issue not to be ignored.
There, that is the extent of my knowledge and practice.  I'm hoping that the true experts on this forum will set the record straight.
Harry
 









   		 	   		  

[Non-text portions of this message have been removed]