>
>
>
> --- In DigitalBlackandWhiteThePrint@yahoogroups.com, "shileshjani" <janishilesh@> wrote:
> >
> > Thank you for clarifying that Roy.
> >
> > So let us take a hypothetical situation:
> >
> > We have a square with a value of 128 (mid value in 8 bit), and print that square. It will look up in the QTR curve for row 128 of each ink to lay down the inks.
> >
> > If we have our image in 16 bits (total discrete levels is 2^16 = 65536), mid point is 23768. We cannot select this gray point in Photoshop color picker as far as I know. But I can imagine it exists is a 16 bit gray file. If we pick +1 from mid-point (or 23768 + 1 = 23769), how will QTR assign inks for this spot? Will it interpolate all inks values between 128 and 129 in the QTR curve description?
>
> Shilesh,
>
> Photoshop lacks easy ways to set specific 16-bit values but you can see them with
> the Info palette -- select 16-bit. Note that 16-bit is really 0-32768 which is more
> like 15 bit but that's another story. But internally with 16-bit files all the arithmetic
> is done in 16-bit so you don't lose any data.
>
> Here's a way to do an experiment to show what goes on.
>
> Create a 16-bit grayscale file - say 8in wide by 2in tall. Make top left black and
> top right white, on the bottom half make a long gradient from black to white.
> Set up your Info palette for gray 16-bit and see the values.
> Now shrink all the values in the file to the range 128 to 129. Do this with the
> Levels command, setting the Output values to 128 and 129. The whole file should
> be essentially all middle gray. But use the Info palette and see top left = 16450 and
> top right = 16577, on the bottom you can see the gradient for the same range.
> On a good screen you might see the border between sides. (even the bottom may
> show it because only 8-bits is passed to the display).
>
> Print this on your favorite paper and set of curves. If you look close you should see
> a dividing line in the middle on the top but not at the bottom. If you have an accurate
> spectro all the better. The two top patches will give consist measurements in
> various places. On the bottom you should be able to see lots of intermediate values
> between the two top measurements -- with a very accurate gradient.
>
> To make it even more interesting convert that 16-bit file to 8-bit and print it
> again. You'll find the prints to be virtually identical -- gradient and all. Yes
> the 8-bit file will show all the intermediate grays in the print that the 16-bit
> file showed. In the gradient some of the pixels are 128 and some 129 - with
> the mix ratio providing the extra information that you probably think was lost.
>
> To answer your question about how QTR works. On the Mac if you send QTR the
> 16-bit data (check 16-bit in PS and in QTR), QTR will linearly interpolate the
> values from the .quad file between 128 and 129. On PCs or Mac where you don't
> set 16-bit, it'll work like the second case above. Note that the 16 to 8 bit
> conversion has to be done intelligently -- PS and QTR both do this.
>
> Try this out. It's really worth seeing for yourself.
>
> Roy
>
> >
> > I know this does not have anything to do with K3 vs K7.
> >
> > Shilesh
> >
> >
> >
> > --- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Roy" <roy@> wrote:
> > >
> > > Hi Shilesh,
> > >
> > > Sorry but I have to disagree with this -- especially: QTR curves are 16-bit not 8-bit.
> > > Yes there are 256 points but each one is a 16-bit value.
> > > > I think some people do not realize that if your image is 8 bit, there are 256 shades available,
> > > > and no interpolation between them is allowed. Even if the ink/paper system is capable of
> > > > producing more shades than 256, it is never achieved in reality. The same applies even if
> > > > your image is 16 bits, because QTR curves are described in 8 bit architecture.
> > > >
> > >
> > > I think the issue of 256 levels of gray is a misleading concept. There are all these
> > > parts of the process that have specs of 8-bit or 256 values that it is very easy to fall
> > > into the trap that the result print has to have 256 levels of gray. But in real prints -- both
> > > digital and film -- the perception of gray is always an average of lots of pixels in the
> > > source and well as lots of dots of ink. It's this averaging that fills in all the mid-grays
> > > between each of the 256 grays -- essentially making a continuous print.
> > >
> > > On a print I can pretty easily see the difference between steps that are 1 unit apart
> > > in 256 grays and 16-bit gradient across those same values. (3800, K3 OEM inks & QTR)
> > >
> > > So whether you are using K7, K3, or BO you will get more than 256 grays. The difference
> > > is mainly seeing dots in the BO case -- the dark black dots on white paper have
> > > enough contrast that you see them with the naked eye. I think the K3 LLK ink is
> > > light enough that you don't see them -- and I mean get up close and look critically.
> > > Most of my prints come from 4x5 film, tmax 100 and the grain from the film is more
> > > obvious than the grain from the dots. The K7 light inks are lighter still and can
> > > stand more magnification before seeing the dots. Tyler's article illustrates this,
> > > but the main thing I get from his scans is the dot resolution of the print is the most
> > > important factor -- 2880dpi is better than 1440dpi, and the silver dpi is still much
> > > better than that.
> > >
> > > The things I think are important are using 16-bit for all editing from 16-bit scan
> > > or raw digital. Use all the resolution you can. QTR will use up to 720ppi, the
> > > Epson driver with use this on most printers if you select Finest Detail.
> > > To get great prints most things boil down to starting with the best image and
> > > knowing now to edit and get what you want it to look like.
> > >
> > > ---------------
> > >
> > > As far as what inks to use -- they are all good but there are tradeoffs and different
> > > amounts of work to get there. Probably the biggest aesthetic decision is matte
> > > versus photo paper -- and within each category there are lots of choices.
> > >
> > > K7 is great in that most is fixed and done for you. K3 is great in that you have
> > > fine tuning control of color tone including split tones if you use QTR. Glossy
> > > papers are often problematic and need GLOP but I find Harman Gloss Baryta
> > > does really well with Epson K3 inks straight as is.
> > >
> > > Roy
> > >
> > > --- In DigitalBlackandWhiteThePrint@yahoogroups.com, "shileshjani" <janishilesh@> wrote:
> > > >
> > > > I just read an interesting post by Jon Cone here:
> > > >
> > > > http://www.piezography.com/PiezoPress/blog/piezography-technical/making-a-fine-piezography-print/
> > > >
> > > > He says on page 2 of the pdf:
> > > >
> > > > "Because Piezography with four or as many as seven distinct shades of ink can separate more tonal values than can the OEM three shade systems, it is important to start with very smooth image files that have a long tonal scale. Without going into mathematics and the virtues of using overlapping shades of Piezography inks rather than the interlinking shades of the OEM, the Piezography system produces tens of thousands more gray value combinations than can the OEM"
> > > >
> > > > Is this true? When I arbitrarily open up a QTR curve for K7 inks published by Piezography, I find exactly 256 gray values, for each there is an assignment of inks to be used. But 256 (8 bit) gray values, no more no less.
> > > >
> > > > To be provocatine, I say even a single black only (BO) system can print 256 shades gray. To be fair, the multiple ink set-up of K7 (or K3 for that matter) will produce each of the 256 shades with greater smoothness and overall finess than a BO print. Using the 8 bit QTR set-up, both BO and multi K will each produce 256 shades of gray.
> > > >
> > > > I think some people do not realize that if your image is 8 bit, there are 256 shades available, and no interpolation between them is allowed. Even if the ink/paper system is capable of producing more shades than 256, it is never achieved in reality. The same applies even if your image is 16 bits, because QTR curves are described in 8 bit architecture.
> > > >
> > > > In summary, on one single paper, K7, K3, or BO all produce AT MOST 256 shades of gray. It can be argued that using ABW on that same paper, you can actually produce many, many more shades of "near-neutral gray", but only 256 at a time. Again, ABW gets maligned in a manner that has no basis in fact.
> > > >
> > > > Did I get this all wrong? Is there something to this that eludes my mind? Or is the ".........tens of thousands more gray value ....." just poetic waxing?
> > > >
> > > > I concede there are some reasons to chose multi K systems like Piezography K7, but greater number of gray shades?
> > > >
> > > > Shilesh
> > > >
> > >
> >
>