Synth-DIY Yahoo! Groups Archives

Thread

Printing custom targets

2015-02-03 by jeff.grant@...

I want to create a test patch file with K of 0-10 and 90-100 which I can then measure. Using Colorport or i1P, I can make the target but it's in CMYK. If I change mode to GG, the values are corrupted.

The supplied QTR targets are in GG. Can someone please point me to the tool that I need or advise how to get the target in GG.

Re: Printing custom targets

2015-02-03 by richard@...

You can make custom targets with the RGB values in ColorPort and then convert it to greyscale afterward and then print through QTR.

You can paste this below into a text file, save it and then import it into ColorPort to create a custom target. I use excel to generate the step values and then copy that to a text file and then import to color port. .

0 0 0

2.55 2.55 2.55

5.1 5.1 5.1

7.65 7.65 7.65

10.2 10.2 10.2

12.75 12.75 12.75

15.3 15.3 15.3

17.85 17.85 17.85

20.4 20.4 20.4

22.95 22.95 22.95

25.5 25.5 25.5

38.25 38.25 38.25

51 51 51

63.75 63.75 63.75

76.5 76.5 76.5

89.25 89.25 89.25

102 102 102

114.75 114.75 114.75

127.5 127.5 127.5

140.25 140.25 140.25

153 153 153

165.75 165.75 165.75

178.5 178.5 178.5

191.25 191.25 191.25

204 204 204

216.75 216.75 216.75

229.5 229.5 229.5

232.05 232.05 232.05

234.6 234.6 234.6

237.15 237.15 237.15

239.7 239.7 239.7

242.25 242.25 242.25

244.8 244.8 244.8

247.35 247.35 247.35

249.9 249.9 249.9

252.45 252.45 252.45

255 255 255

Richard Boutwell

Re: Printing custom targets

2015-02-03 by jeff.grant@...

Thanks Richard. I used CMYK in ColorPort originally.

It's probably my lack of understanding but, once I convert to GG in PS, the histogram readout is not linear.

Using your data, for K, I get 100,99,97,95 etc. I expected to see the numbers that I had entered. The RGB is similar, it goes 0,1,1,3 for L.

Re: [QuadtoneRIP] Re: Printing custom targets

2015-02-03 by Ernst Dinkla

The created RGB target can be printed through QTR as well, it will act as a greyscale. At least that is what I recall.

For conversion you could assign AdobeRGB to the RGB target when loading in PhotoShop and convert to Greyscale with Gamma 2.2, then save without a profile attached. Both spaces have Gamma 2.2. In theory Photoshop does not interfere then, it would surprise me when the numbers change with that method.

Met vriendelijke groet, Ernst

Dinkla Grafische Techniek
Quad, piëzografie, giclée
www.pigment-print.com

Show quoted textHide quoted text

On Tue, Feb 3, 2015 at 6:04 AM, jeff.grant@... [QuadtoneRIP] <QuadtoneRIP@yahoogroups.com> wrote:

Thanks Richard. I used CMYK in ColorPort originally.

It's probably my lack of understanding but, once I convert to GG in PS, the histogram readout is not linear.
Using your data, for K, I get 100,99,97,95 etc. I expected to see the numbers that I had entered. The RGB is similar, it goes 0,1,1,3 for L.

Re: Printing custom targets

2015-02-03 by richard@...

I just looked at the target I created with the numbers in my previous post. I brought it into photoshop, assigned sRGB, and then converted to GG2.2. Both the RGB and K are as expected. I just uploaded the target and the reference file to my BW shared dropbox folder.

Here is the link:

Tif: https://www.dropbox.com/s/fh9v0nmhg00bg63/CP-0-100-w-0-10-90-100.tif?dl=0

Ref File: CP-0-100-W-0-10-90-100.txt

CP-0-100-W-0-10-90-100.txt

Shared with Dropbox

View on www.dropbox.com

Preview by Yahoo

Richard Boutwell

Re: [QuadtoneRIP] Re: Printing custom targets

2015-02-03 by Roy Harrington

I guess I have to jump in here.

There are a lot of little gotcha's in making accurate targets -- or even what "accurate" means.

The first issue is that 8-bit files are just that 8-bit -- values from 0 to 255. So talking about

percents (i.e. K-values) is just an approximation. With 5% steps like the 21-step its

close enough to not worry about. But trying to do 1% steps is full of potholes.

Richard's text reference file shows all the fractional steps -- 0, 2.55, 5.1, 7.65, ...

but when made into the 8-bit tif target file all the fractions are dropped off -- there are

no fraction values in an 8-bit file. So the sequence becomes 0, 2, 5, 7, ... which is

nowhere near even. Some jumps are 2 and some 3. That's a 50% size difference.

To get reasonable 1% steps you have to use 16-bit. Haven't used ColorPort much

but I don't think that's possible. You can doctor up the file in Photoshop -- doable, but PS

has no 16-bit input capability so you have to know enough how to calculate it.

Doing a 85-step wedge is all RGB=3 steps so that may be possible.

Bottom line: you are stepping into a quagmire trying to do this.

The second issue you are running into is changing file formats between CMYK, RGB & Gray.

Skip CMYK altogether I can't think of any use that you would likely need.

But with RGB & Gray if you convert these in Photoshop that ALWAYS involves a color management

conversion. Simple Mode change is a conversion to the new working space set in Color Settings.

Using the Ink Info palette in Photoshop also is a color management conversion for anything

that is not the actual Mode. I.e. looking at RGB for a Gray file is conversion to RGB working space.

Looking at L-values is always based on embedded profile (working one if untagged).

As Ernst mentioned Adobe RGB & GrayGamma2.2 are compatible so that is a workaround.

If you don't pay specific attention to all this you can easily end up with wrong info.

Finally given that you are on a Mac with all this, Print-Tool is really the only way to

do custom targets for curves or ICC profiles and then for actual printing. PS is wonderful

for editing images but for printing it will throw extra conversions into the mess.

Roy

Show quoted textHide quoted text

On Tue, Feb 3, 2015 at 5:38 AM, richard@... [QuadtoneRIP] <QuadtoneRIP@yahoogroups.com> wrote:

I just looked at the target I created with the numbers in my previous post. I brought it into photoshop, assigned sRGB, and then converted to GG2.2. Both the RGB and K are as expected. I just uploaded the target and the reference file to my BW shared dropbox folder.

Here is the link:
Tif: https://www.dropbox.com/s/fh9v0nmhg00bg63/CP-0-100-w-0-10-90-100.tif?dl=0
Ref File: CP-0-100-W-0-10-90-100.txt

CP-0-100-W-0-10-90-100.txt
Shared with Dropbox

View on www.dropbox.com

Preview by Yahoo

Richard Boutwell

Roy Harrington
roy@...
www.harrington.com

Re: [QuadtoneRIP] Re: Printing custom targets

2015-02-04 by richard@...

Roy is absolutely right with the 1% steps not being perfectly even when using the 0-255 numbers. Whenever I looked at the 1% steps in the info pallet in photoshop they always showed 2, 5, 7, etc., and I thought it was just "normal"—I kinda assumed that ColorPort would create the correct steps with the values entered. Since I'm only now starting to get some very basic experience with programming and "computer math" I can learn something here too...I know there are a whole lot of people a whole lot smarter than me in these areas. The only thing I really know how to do well is make a good print.

About the targets I actually use nowadays: I have made and used the perfectly spaced 86 step targets, where each step between the 0-255 values are an even 3. It didn't seem to offer any real advantage in making QTR profiles over the 21 step grayscale included in QTR—with the required difference between each L* step used for linearization it really made it worse when the install script creates its correct curve look up.

I use the 51x3 step for checking linearization and ICC creation. I am working on something now that might be better suited for the 86 step grayscale, but I don't ever use the 100 step targets—I've found that there can be too much variance in the actual readings when the scale is sliced that thin, and it is hard to know what is actually "off"‚ the print of the spectro... (even when using the 51 step target, which is why I average that one.) In any case, the 1% steps at the end of the scale might not offer any advantage in actually making profiles, but would just show you what is happening right up to that last 99% step. Those it is easy to see where the in between 1% tones would fall when graphing the 2% 51-step wedge as well.

Richard Boutwell

Re: Printing custom targets

2015-02-04 by jeff.grant@...

Thank you all for the assistance, it is most appreciated. As a recent arrival from the world of colour, the concept of using a curve for one printer on another is an alien concept. The folks at IJM told me that I could use x880 glossy curves for an old 2400.

I'm happy to take Jon at his word on that but I thought that it would be interesting to be able to print and then read a patch file to see how they look. There is no shortage of ill-informed opinion about PZ and QTR, the thought of being able to actually measure it is very appealing.

I guess that leaves me with the question of whether I can really measure this or not. Is there another way, or is it an act of faith? I ask that question in all seriousness. I'm committed to PZ and QTR. I have a brand new 3880 waiting for carts, and have been through a lot of pain with an old 2400 getting this far,

Re: Printing custom targets

2015-02-04 by richard@...

You can measure the 51 step target and then graph a chart in Excel or or Numbers or google sheets, and observe if the shadows block up at the end of the scale. You don't need to measure 1%K steps to do that, the 2%K steps are enough. I suppose if you wanted, you could make to make the first and last 5 steps with the 8bit 256 values and still see a gradual density change, but aside from averaging several readings would be hard to tell how much change is actual density and how much is variance in the measurement data.

You could also print the QTR bullseye target to see if there are any noticeable steps. I made a modified bullseye target with additional gradients so I can see exactly there the black are blocking up or where the white are dropping out. The smooth gradients will show any banding problems that might not be visible in stepped gradients. I use this for testing for problems with any new profile. The gradients at the bottom were made with the posterize adjustement from a smooth 16-bit gradient and I made 10%, 5%, 2%, 1% steps. I made this awhile ago and I think went as far as 256 steps before leaving it as the original smooth 16-bit gradient. It is a 3x4 720dpi 16-bit tiff, and I leave it as that when printing to see just how well i've done making the profile.

You can download it here: BWMASTERY-Grayscale-torture-test.tif

BWMASTERY-Grayscale-torture-test.tif

Shared with Dropbox

View on www.dropbox.com

Preview by Yahoo

Richard Boutwell