I have taken Jon Cone's linearisation checker spreadsheet and extended it so that it spits out the points for a photoshop curve that is the inverse of the one you get from plotting the measured L values. I then printed the 21x4 with this curve applied and remeasured the linearity and it was pretty much dead straight along that pink line. I tested this for several papers and it worked in each case. So this provides one solution.
However it does mean remembering to apply the curve for each image. It would be a lot simpler if this curve could be applied to the .quad file. But the photoshop curve is L whereas the .quad file has ink values, which I assume are more akin to density. Undeterred I made a couple of attempts. I used a cubic spline to interpolate all 256 points from the 21 linearity measurements.
I had some success. The straight L curve doesn't provide enough of a correction, so I migrated my inverse curve calculations from L to density and applied them to the .quad file. I got fairly close, but not quite close enough. So what am I missing? For the photoshop curve, I had to apply a scaling factor to translate a dmin-to-dmax curve into the 0-255 range. I suspect that I'm missing some sort of scaling factor here as well. Until I can get this to work, I'm using the photoshop curve to linearise.
I've read Mike's post below several times and I still don't quite understand his methodology. I don't quite get his reverse lookup. Mike - if you're still around, I'm happy to exchange spreadsheets and explain my methodology in more detail. It must be possible to get this to work somehow, other than paying $99 per paper.
B.
---In QuadtoneRIP@yahoogroups.com, <drmrking@...> wrote :
After many months of wracking my brain and experimenting, I had a light bulb moment and realised the solution was staring me in the face all the time - you simply print and measure the linearization strip and that creates a response curve for your printer+ink+paper. Then you do a reverse lookup from that curve back the the quad file values for the linear steps you want to use.
So to give a simple example, if hypothetically you want to linearize at Lab Dmin,10,20,30,40,50,60,70,80,90,Dmaxyou simply extract from the quad file the values that give you these Lab values on your response curve.The only requirement is that original response curve has to be continuously increasing otherwise you can get duplicate values that this approach cannot resolve. But that's easy to avoid with a reasonable step width.It works great and I was able to use linearization steps from 21 to 128+ with ease. 128+ is a special case and requires a consistent paper surface and the average of several sets of readings, which in turn is only practical with a sheet reading spectro.To do this in practise you need to set up a spreadsheet to do the heavy lifting of searching for the values and filling in the gaps between the linearization points etc. I created one for my own use and shared a version with a few others to test - which all worked great. I was planning to develop a web based service to do this cheaply for others, but I got waylaid by other more financially pressing projects. I've been thinking for the last year or two that I should share more widely the work I did, so others can use it.When I've got some time over Christmas I will share the spreadsheets and some simple instructions.
Cheers,Mike