Re: Optimal file resolution for Epson printers
Re: Optimal file resolution for Epson printers
- Subject: Re: Optimal file resolution for Epson printers
- From: Graeme Gill <email@hidden>
- Date: Tue, 18 May 2004 17:03:47 +1000
- Organization: Color Technology Solutions
Ernst Dinkla wrote:
Correct, but quoting the number of R sensors only isn't
representing the truth either. The array delivers more
To a first order approximation, the number of full color pixels (the
only useful kind) is the number of R + G + B sensors /3. Yes, the
software fakes up 2/3 of the information for each R, G or B sensor,
thereby returning a raster that has a total number of pixels equal
to R + B + 2 * G, but in my book, this is analogous to those scanners
that upsample to any raster resolution you like.
It has never been the practice of display manufacturers to
use these type of tricks either. You will notice that LCD
screen manufacturers rate the resolution of their screen in
full color pixels, not individual R, G & B display elements.
To avoid aliasing artifacts completely in input devices, requires
the use of an optical low pass filter, to prevent frequency components above
the nyquist limit entering the sensor (this is basic sampling
theory). In practice, it is hard to manufacture high order optical
low pass filters, so often such filters are a compromise, so as not to
reduce the overall sharpness excessively.
I'd be surprised if any of the Bayer array based imaging devices (or
digital photography cameras in general) attempt to use a serious optical
low pass filter, because to suppress aliasing effectively, it would
have to suppress sharp color differences across adjacent R, G & B sensors,
thereby making a mockery of the claimed resolution.
The same applies to your inkjet analogy, it will not be 10080 DPI
but the quality can be better at 1440 dpi with 7 colors. One
Putting aside 7 colors vs 4 colors, the point is that you don't
have a colored pixel at all, unless you overlay at least 3 colors.
Saying that a pixel with only one color has "lower quality" is
a rather ridiculous understatement in my opinion.
Mediocre reviews and testing add to that. Nothing new of course,
the CPU industry has to quote frequencies first to get attention.
See AMD's CPU names etc.
Yep, which is why I thought I'd stir the pot a little :-)
suggesting quantity, that plays a role here. If quality is
expressed by a negative number, that would decrease with every
step in technology, sales probably would stagger.
The video industry is used to coping with this sort of thing. Video
cameras have long been rated by the number of horizontal lines they
can resolve, as well as signal to noise ratio (S/N). I'm intrigued
that there isn't more information in digital camera reviews on
the differences in S/N between CMOS based sensors and CCDs for instance.
Fovean started nice with the correct resolution number and the 3x
added. I got the impression that their quotings are inflated as
well now. Sigma uses 10 M x3 in their ads. No good either.
Yes it's sad reflection on the industry that a manufacturer who
has been honest about their product, is now feeling pressured into
exaggerating because all their competitors do it.
Graeme Gill.
_______________________________________________
colorsync-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/colorsync-users
Do not post admin requests to the list. They will be ignored.