On Mon, Jun 6, 2016 at 9:45 AM, kenlee333@... [QuadtoneRIP] <QuadtoneRIP@yahoogroups.com> wrote:
Please correct me if I am wrong.
You are mixing up various issues. And worrying about things that are not important.
Are we confusing precision with accuracy ? We all know that we can be precise to 10 decimal places, but if our readings are only accurate to 2 decimal places, our data is only good to 2 decimal places.
Yeah, but accuracy is not the important thing.
It seems that we are starring out with readings from a 21-step wedge, then smoothing them out to 256 steps. At that point, the graph may be smooth, but it's still accurate to only 21 steps (less than 5-bit precision). smoothed out to 8-bit precision. Apparently the 16-bit driver subsequently interpolates that out to 16-bits.
The 21steps are on the x-axis but the y-axis is what counts for how much ink gets output.
We could interpolate it to 64 or even 128 bits precision if we liked, but that's just precision: the data is only accurate to less-than-5-bits, no ?
Data accuracy is how good your device is -- decimal places of the measured values.
Not how many patches you measured.
I scan film and presume that my 16-bit files are actually accurate to 16-bits, not just interpolated. Am I mistaken there too ?
Lots of luck there. Film is just chunks of opaque silver or clear plastic. There's no where near
16-bits of ACCURACY. But averaging over larger patches produces perception of smoothness.
The important part of printing is smooth transitions all along -- 16-bit is actually overkill,
but it insures smoothness no matter how steep or flat your gradients are.
For each location (x,y) the printer either prints a dot of ink or not -- just 1 bit. For the most part
averaging noise dominates what you see.
Roy