Digital BW, The Print

Jeff,

Wow, thanks for the details. This will keep me going for a while...

Craig Spaulding

--- In DigitalBlackandWhiteThePrint@y..., "jrandall1149" 
<jrandall@c...> wrote:
> --- In DigitalBlackandWhiteThePrint@y..., "craig_spaulding" 
> <craig_spaulding@y...> wrote:
> > Jeff,
> > 
> > I have several questions re your transfer curves technique:
> > 
> > 1) Can the black and white points be set in Photoshop instead of 
> the 
> > scanner after the step-edge has been scanned? I work with a 
digital 
> > camera and my scanner is marginal and I don't think the black 
and 
> > white points can be set in the software.
> 
> I've had better luck setting the b/w points during scanning than 
> setting in my image editing software (your results may vary).  I 
use 
> VueScan as my scanning software and have found the histogram 
feature 
> to be very good. (www.hamrick.com). I use Picture Windows as my 
image 
> editing software because I think its curves/histogram feature is 
the 
> best of any editing package (I also use it for a lot more reasons, 
> but that's another post) (www.dl-c.com). I also have and 
occassionly 
> use PhotoShop 6.0 
> > 
> > 2) Is this a technique for making a transfer curve or a curve 
that 
> > is to be applied in an adjustment layer? Once the curve is made, 
> can 
> > it be applied either place?
> 
> Transfer curve and adjustment layer curves are essentially the 
same.  
> They both map an input value (in what ever units) to an output 
> value.  Because they use different units they are not directly 
> interchangeable, but if you do the math you can create curves the 
do 
> the same thing in both worlds.
> > 
> > 3) Is there a good method for going about modifying an existing 
> > curve (like the Woolf)? Do you click on the appropriate step in 
the 
> > step-wedge to find out where it is on the curve that step is, 
then 
> > use the up/down arrow or change the numbers in the dialog box? 
If 
> > so, is there a rhyme or reason for how much to change it to 
equally 
> > space out the peaks, or, is it all trial and error. Or, do you 
just 
> > focus on changing the general shape of the curve to space out 
the 
> > peaks?
> >
> Here is an approach that I used to create my RGB Partitioned 
Workflow 
> for the 1160 and MIS FS inkset (and modify the Woolf lumped 
> workflow).  It works for me (your milage may vary)  It is not a 
> complete cookbook and every nuance is not spelled out.  
> 
> 1)  Print a quad purge image and scan using the procedure outlined 
in 
> step #3 to determine the approximate reflective values (I use the 
0-
> 100% scale) of each pure ink on the paper.   Crop the image to 
> produce approximately equal areas of the 4 inks and white space.  
The 
> ink values determine the maximum "density" that can be
> achieved by 
> each ink alone and serve as a guide to when the next darker shade 
> must be started. 
> 2)  Apply a starter RGB partitioned curve set to a 21-step wedge 
test 
> image and then print.
> [Note:  I use the step wedge posted in the Files section of the 
Yahoo 
> Piezography3000 Group.  The file is piezmultisteps.psd.  It has 
> excellent low (95-100%), medium (45-55%), and high (0-5%) 
brightness 
> test areas to allow you to visually  evaluate and set these 
benchmark 
> values. I initially used Paul Roark's curves he developed for the 
> Piezography BW inkset and the Epson driver.]
> 3)  I scan the print with my HP-5370 scanner using VueScan.  VS 
> settings are: 
> * Crop = tightly set to only scan the narrow 21 step portion of 
the 
> test image. Don't include any of the grayscale ramp or the portion 
of 
> the 21 stepwedge with the light or dark separator lines.  My goal 
is 
> to have 21 separate and narrow histogram peaks with little noise 
in 
> between.   
> * Color balance = auto levels. I use auto levels because it sets 
each 
> RGB curve to the b/w points — neutral doesn't and white balance 
> doesn't do this well.
> [Note:  I don't care about the "color" of my
> grayscale—all I'm 
> interested in is the overall/combined brightness.]
> * B/W points set to clip the 0% (255) and 100% (0) gray value 
steps 
> to expand the range to the maximum.  That is, set the 0% gray 
value 
> to the whiteness of my paper and the 100% gray value to the 
blackest 
> black I can print with the inkset. 
> * Scanning resolution = 150 dpi. 
> * Number of passes = 4.  This reduces noise and increases peak 
> separation, especially in the dark end.  The exact pixel alignment 
is 
> not an issue with a stepwedge print.
> * File color space = AdobeRGB
> * Grain reduction = heavy.  This also seems to increase peak 
> separation in the dark end.
> * Tiff file type = 24-bit RGB
> 4) I open the Tiff file in Picture Windows and apply the 
Color|Curves 
> function which combines a curve and histogram view of the pixels. 
> [Note:  The PW Color Curve and grayscale Brightness Curve 
functions 
> can both display the image value, lightness, or brightness of the 
> image in the HSV, HSL or RGB color spaces, respectively.  I use 
the 
> HSV color space because it does a good job of separating the image 
> brightness component from the color components which I don't care 
> about.  Image brightness in the RGB space is a weighted sum of 
each 
> RGB color and overall not as useful.]
> 5)  I note on a worksheet whether each of the remaining 19 peaks 
> needs to be moved lighter or darker and a relative amount.
> 6)  I then load and apply the RGB partitioned curve used to create 
> the tight crop of the 21 step wedge to my test image and adjust 
each 
> individual RGB curve (red=dark, green=medium, & blue=light gray), 
as 
> determined in Step #5. 
> 7) I then save the new RGB partitioned curve under a different 
name.
> [Note: When I get close to where I want the peaks to be, I open 
the 
> Picture Windows RGB partitioned curve file (filename.cc) with a 
text 
> editor and manually change the values (which are stored as 0-255 
> values) ]
> 8)  I then print the test image.  
> 9)  Iterate until you are satisfied (or run out of paper or ink).