Ah, but there are some errors in your reasoning. All of this might make
sense if I were using QImage and printing on a high end inkjet printer.
I do print on an inkjet printer (a cheap one) very occasionally but
any serious printing goes to a commercial (and chemical) print shop. I
output at the native resolution of the devices used in those shops and
sharpen the image at that specific resolution. Then I tell the shop
"don't touch this image, print as is". I then get better looking prints
and they cost me less since the hand/eye/brain of their operator was not
involved.
Chuck Norcutt
On 10/27/2010 1:05 AM, Moose wrote:
> On 10/26/2010 4:14 PM, Chuck Norcutt wrote:
>> ... As far as I'm concerned the image is dimensionless until it comes time
>> to print.
>
> Yup, in a sense. In another sense, it has dimensions in pixels, but not in
> physical dimensions.
>
>> Then the image is cropped if necessary and resized to produce a 250 or 300
>> dpi output print.
>
> Here, I think you may be mishearing Bob. He's saying that there is no reason
> to resize at all if the native pixel
> dimensions of the image and the chosen physical output dimensions result in a
> ppi between 180 and 400+. Although almost
> certainly empirically based, I can see where this could come from.
>
> No matter how it starts, your image has to go through a great deal of
> processing before flying print heads squirt teeny
> droplets of ink onto the paper. (Yes Veronica, inquiring folks have pretty
> convincingly demonstrated that ideas of
> "native resolution" have no discernible effect on output quality.)
>
> When you send it to the printer, it goes through at least three steps of
> processing. The first two occur in the printer
> driver or RIP*.
>
> The first is conversion to colors the printer understands, as generally with
> inkjets, RGB to CYMK, and adjustment to the
> particular printer, ink and paper characteristics to as accurately as
> possible match the image colors. This may involve
> use of proprietary data the maker has put in their printer driver or ICC
> profile matching.
>
> The second is conversion of the image data into a string of commands in the
> printer's own language.
>
> The third occurs as the printer's firmware interprets the commands into a
> series of electrical signals to the physical
> print mechanism.
>
> Now that's a lot of processing, that ends up not with an even pattern of dots
> of single colors, but in a pattern of dots
> of various colors. Four color is long gone in photo printers. The one sitting
> here has eight cartridges, although two
> are blacks and one isn't a color at all, but a surface. So CYMK now is
> translated again, to CcYMmK, or CYMKkRB, or
> whatever inks that particular printer uses
>
> What it produces on paper isn't like screen printing, where color dots in a
> fixed pitch vary in size in an analog way.
> Although we think so, do we even know if the droplets are put down in a fixed
> pitch? The printer makers sure aren't
> telling the details of how they hope to do better than their competitors. Are
> the drops all one size?
>
> According to their specs, these printers are laying down dots at a much finer
> pitch than the pixel pitch sent to them.
> So they are, in effect producing a rather subtle matrix or mist of colored
> dots that, together with the color of the
> paper, add up to the image.
>
> I go into this detail to suggest that, within a fairly broad range,the dpi of
> the input may have little effect on the
> process of printing it.
>
> Some of these RIPs/drivers are very clever. QImage, for example, is a print
> manager that includes it's own RIP. A few
> years ago, a friend needed slides of some people/things he didn't have pics
> of. He searched the web, only to come up
> with some very low res images. I tried all sorts of things to upres/enhance
> them. Nothing in PS, either its own or a
> couple of plug-ins, really did much to improve the prints.
>
> I tried QImage. Just amazing. I don't know what it did, but images printed
> with it and its sizing and sharpening
> functions were at least an order of magnitude better. They almost made me
> believe in those TV shows where eight pixels
> are "enhanced" so you can read the license plate.
>
> Am I going to mess around resizing my images to some old standard idea? Nope,
> they are going into QI at whatever
> camera/scanner native pixels I have and I'll tell QI what size prints to
> make. Clearly, it does its processing with
> prior knowledge of how the printer will render what it puts out. That's way
> smarter than me deciding what ppi to resize
> my image to.
>
> Print Me Moose
>
> * Raster Image Processor. RIP is to printer driver roughly as Giclée is to
> inkjet. They do the same thing, but one has
> more pretentions, costs more and may be better than the other.
>
>> It gets sharpened after the resizing.
>
> Again, I do resharpening for web images, but not for printing with QI.
>
>> I don't see any reason to go to 400 dpi.
>
> As above, try not resizing at all between 180 (maybe even 150) and something
> over 400 ppi. let the smart soft/firmware
> do it's job.
>
>> There aren't many Mooses around. :-)
>
> Pick one:
>
> How blessed are they who know one. :-P
> Thanks for small favors.
>
> But keep your choice to yourself, There will be a test later. :-)
>
> Moose
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|