At 8:43 PM -0500 11/6/00, Austin Franklin wrote:
> > *All* Epson photo printers will print best at integer divisors of
>1440dpi.
>
>That is not universally true. Some people claim they see no difference at
>some 'magic' number, but to just keep the DPI above 240 or so...
>
>The driver has a dither pattern, that is either based on random halftone
>cell size, or a fixed halftone cell size, and they don't have to be square
>cells. Unless you know exactly what the driver is doing, the 'magic'
>numbers are either pure nonsense, or pure dumb luck.
>Personally, I have found the 'magic number' 'rumor' to be unfounded for my
>printers. I have two Epson printers, the 1160 and the 3000. Neither one
>of them prints better at some integer divisor of 1440.
Austin, I'm inclined to agree. This discussion comes up
regularly on the Epson list, with people coming down firmly on both
sides of the debate. They swear they get sharper prints by resampling
to 240, or 360 ppi, or some other integer divisor of 1440 dpi .
Others swear it's better to not resample in Photoshop, but let the
Epson driver do the interpolation as it prints to the selected size.
I decided to try it for myself by taking one of my sharpest
scans (2720 ppi, 26 Mb), which has some tiny, barely visible printing
on a bottle label. I made 3 prints of a cropped portion of the image
at the same physical size (about 5x7 inches, using an 870 on Epson
Photo Paper at maximum quality). Two of the prints were resampled
from the original scan - one to 240, one to 360 ppi. The third was
simply resized ("Resample Image" unchecked), giving a resolution of
367.9 ppi when it was made the same size as the other two prints.
I examined the 3 prints using a high-quality 4x loupe. There
was a slight but visible improvement in quality from 240 to 360,
which didn't surprise me too much. What did surprise me was that
there was about the same degree of improvement from the 360 to the
non-resampled 367.9 print. It therefore appears, at least in this
case, that it is indeed better to retain the whole file size of the
scan. If you want to make a smaller print, just resize (without
resampling) to whatever print size you want, letting the file
resolution fall where it may.
There is no question the differences were slight, and would
probably be invisible at normal viewing distance (and might not even
be reproduced by a lesser printer), but it is interesting to know
that they are there.
Regards,
Roger Smith
