Colorvision Profiler Plus (Print Profiler)

From: Raphael Bustin

I’ve never once made a saleable
print using a profile from Profiler Plus. $169 about
four years ago, not one useful print

Ditto for this product and also for Monaco EZ Color.

The print target is scanned on an
Epson 1640SU.

Personally I think this is the problem, it’s hard to get excellent results with a consumer grade flatbed scanner. Hard to turn off ALL the internal color management options on these models (which you *have* to do or the results are skewed), and there’s no way a scanner like this as accurate as a $1,000 + spectrophotometer, which is the instrument you and I REALLY need to get a good profile, I feel.

So what could I be doing wrong?

Talk to Flycaster, he has a $1,500 Gretag solution that really works, at least on the Epson 2200.

Or buy the Epson 7600 or 4000 and use Bill Atkinson’s highly accurate custom profiles (that’s what I’m doing).

Bill

Raphael Bustin wrote:

Why do I bother?

I’ve owned a copy of this product from back before
Colorvision was bought by Pantone, and was known
as Horses LLC.

Mostly I’ve given up on this package (the print
profiler, not the Eye-1) but once in a blue moon
I trot it out again. I’ve never once made a saleable
print using a profile from Profiler Plus. $169 about
four years ago, not one useful print, but dozens
and dozens of test prints and print targets in the
waste bin.

The current version is 2.3. I’m using it with a trial copy of Photoshop CS, on a Win2K machine. My monitor is
calibrated and profiled with Gretag Eye-1 Display.
(It’s a Viewsonic PT-775.)

In past years I’ve used this product with PS 5.5, with
similarly poor results. The print target is scanned on an Epson 1640SU. I’ve tried many combinations of printers, inks and papers (Epsons back then, Canon 9000 now.)

What’s the problem? Deep blue tones get turned
to deep cyan. Otherwise, no big complaints, but
that issue alone makes these profiles unusable, IMO.

So what could I be doing wrong? I can think of
a few reasons for this behavior:

1. The cyan really "is" in the image but my
monitor can’t show it. (But if that were true, why
doesn’t the cyan show when printing without
a profile?) Without the profile, deep blue is what
shows on the screen, and what prints on the paper.

2. Some issue or problem with the scan of the
print target.

3. The product really does suck.

Any other ideas?

I’ve always been baffled that this product magically
works without needing a "reference" target to scan (as does Monaco, for example.) OTOH, I’ve seen
this same problem from profiles I’ve made with Monaco.

I second Bill Hilton’s suggestion that the scanner is what’s limiting your profiles. With the current technology, it’s easier to print a good image than it is to scan one. In particular, the darkest areas of a print are difficult to discren using a flatbed.

You can prove this to yourself relatively quickly if you own an IT8 or similar target. Scan the gray step wedge, select the bottom half dozen or so steps, and use Image>Histogram to look at the histogram. Odds are the bottom two or three peaks are blurred together, and may even be merged into one.

There are ways to deal with this. If you have a friend with a drum scanner, or can get a drum scan done, that’s probably the best alternative.

There are steps you can take using your own scanner. First, take your scanner apart and clean the underside of the glass. Second, scan several times and combine the images as layers in Photoshop to average them, using a geometric pattern of 50%, 25%, 12%, etc for the transparency. Then see if you can discern the histogram peaks.

If you still can’t, it may be time to give up, but there are a few more tricks, particularly if you are curious and want to experiment. or if you have more time than sense :-). The idea is to post-process the scan in Photoshop to make it look as you think it should. Your most important tool here is the histogram window, and the info palette which functions as a digital colorimeter.

First, if you can see a ghost of a peak in Photoshop, call that the "nominal" value that that square should have, select that square with the marquee rectangle and fill it with the nominal color. Repeat for the other squares that are blurred in the histogram.

Second, if there is an overall cast to the shadows – in your case cyan – then add cyan to the shadows using Photoshop. The amount can be approximated by scanning your print and subtracting the cyan in your neutral shadow from the image using a curve. you may have similar adjustments to the quarter, half, and three quarter tones. In other words, use curves to fiddle with the color balance to force the final profile to come out correctly. Do not adjust the values of individual patches, but use curves to change the overall color balance in a smooth manner. Otherwise you run the risk of a discontinuous curve.

Third, get rid of noise in the target patches. You may be able to do simply apply Gaussian Blur across the entire scan, or by selecting each patch individually with the marquee and blurring.

Use Curvemeister’s profile viewer action to see how smooth your final profile is. This is a free download at www.curvemeister.com . It generates what is called a "neutral axis" – the set of RGB values used to generate shades of gray. While not complete, it does give a good idea of whether the profile will produce banding (look for kinks in the curve), whether it appears to unduly favor a particular color, and whether your are losing the dark areas of your image – a common problem with consumer grade LCD scanners.

You may also use the specially prepared Lab image available at curvemeister to see the overall size of the gamut of the profile, and again look for jagged or disconnected patches of color that generally mean banding will occur.

All said, though, there are no guarantees, since the bottleneck is basically the scanner. You may go through all this and still not end up with a useable profile.
—

Mike Russell
www.curvemeister.com
www.geigy.2y.net

On Mon, 19 Jan 2004 04:43:31 GMT, "Mike Russell" wrote:

Raphael Bustin wrote:

Why do I bother?

I’ve owned a copy of this product from back before
Colorvision was bought by Pantone, and was known
as Horses LLC.

Mostly I’ve given up on this package (the print
profiler, not the Eye-1) but once in a blue moon
I trot it out again. I’ve never once made a saleable
print using a profile from Profiler Plus. $169 about
four years ago, not one useful print, but dozens
and dozens of test prints and print targets in the
waste bin.

The current version is 2.3. I’m using it with a trial copy of Photoshop CS, on a Win2K machine. My monitor is
calibrated and profiled with Gretag Eye-1 Display.
(It’s a Viewsonic PT-775.)

In past years I’ve used this product with PS 5.5, with
similarly poor results. The print target is scanned on an Epson 1640SU. I’ve tried many combinations of printers, inks and papers (Epsons back then, Canon 9000 now.)

What’s the problem? Deep blue tones get turned
to deep cyan. Otherwise, no big complaints, but
that issue alone makes these profiles unusable, IMO.

So what could I be doing wrong? I can think of
a few reasons for this behavior:

1. The cyan really "is" in the image but my
monitor can’t show it. (But if that were true, why
doesn’t the cyan show when printing without
a profile?) Without the profile, deep blue is what
shows on the screen, and what prints on the paper.

2. Some issue or problem with the scan of the
print target.

3. The product really does suck.

Any other ideas?

I’ve always been baffled that this product magically
works without needing a "reference" target to scan (as does Monaco, for example.) OTOH, I’ve seen
this same problem from profiles I’ve made with Monaco.

I second Bill Hilton’s suggestion that the scanner is what’s limiting your profiles. With the current technology, it’s easier to print a good image than it is to scan one. In particular, the darkest areas of a print are difficult to discren using a flatbed.

You can prove this to yourself relatively quickly if you own an IT8 or similar target. Scan the gray step wedge, select the bottom half dozen or so steps, and use Image>Histogram to look at the histogram. Odds are the bottom two or three peaks are blurred together, and may even be merged into one.

There are ways to deal with this. If you have a friend with a drum scanner, or can get a drum scan done, that’s probably the best alternative.
There are steps you can take using your own scanner. First, take your scanner apart and clean the underside of the glass. Second, scan several times and combine the images as layers in Photoshop to average them, using a geometric pattern of 50%, 25%, 12%, etc for the transparency. Then see if you can discern the histogram peaks.

If you still can’t, it may be time to give up, but there are a few more tricks, particularly if you are curious and want to experiment. or if you have more time than sense :-). The idea is to post-process the scan in Photoshop to make it look as you think it should. Your most important tool here is the histogram window, and the info palette which functions as a digital colorimeter.

First, if you can see a ghost of a peak in Photoshop, call that the "nominal" value that that square should have, select that square with the marquee rectangle and fill it with the nominal color. Repeat for the other squares that are blurred in the histogram.

Second, if there is an overall cast to the shadows – in your case cyan – then add cyan to the shadows using Photoshop. The amount can be approximated by scanning your print and subtracting the cyan in your neutral shadow from the image using a curve. you may have similar adjustments to the quarter, half, and three quarter tones. In other words, use curves to fiddle with the color balance to force the final profile to come out correctly. Do not adjust the values of individual patches, but use curves to change the overall color balance in a smooth manner. Otherwise you run the risk of a discontinuous curve.

Third, get rid of noise in the target patches. You may be able to do simply apply Gaussian Blur across the entire scan, or by selecting each patch individually with the marquee and blurring.

Use Curvemeister’s profile viewer action to see how smooth your final profile is. This is a free download at www.curvemeister.com . It generates what is called a "neutral axis" – the set of RGB values used to generate shades of gray. While not complete, it does give a good idea of whether the profile will produce banding (look for kinks in the curve), whether it appears to unduly favor a particular color, and whether your are losing the dark areas of your image – a common problem with consumer grade LCD scanners.

You may also use the specially prepared Lab image available at curvemeister to see the overall size of the gamut of the profile, and again look for jagged or disconnected patches of color that generally mean banding will occur.

All said, though, there are no guarantees, since the bottleneck is basically the scanner. You may go through all this and still not end up with a useable profile.

Hey, now Mike.

I’ve been toying with CurveMeister and I think you
should have worked for Adobe. Quite cool.

But I couldn’t find anything like a "profile preview." Can you give me a hint on how to get there?

I think maybe what I could use (or would like to
see) is an ICC profile editor or generator that uses
precisely the sort of GUI and interface that you’ve
built into Curvemeister.

I mean… rather than let a stupid flatbed scanner
"automagically" generate an ICC profile — often by trial and error in "tweaking" the thing afterwards — why not just have a GUI that lets you "draw" the profile from scratch — test the results, iterate, fine tune, etc…

PS: I’ve tried — in the past — a few of the tricks you suggested for tweaking the profiles by doing a
Curves adjustment of the resulting target scan.

But hey, if I’m going to have to do that, I think I’ll
just hang out with Photosphop 4 indefinitely, cuz
"rolling your own" printer profiles just doesn’t seem worth the bother.

rafe b.
http://www.terrapinphoto.com

Raphael Bustin wrote:
[re post tweaking of scans for use in generating profiles]

Hey, now Mike.

I’ve been toying with CurveMeister and I think you
should have worked for Adobe. Quite cool.

Thanks for the kind words! The price will be going up at the end of January, BTW 🙂

But I couldn’t find anything like a "profile preview." Can you give me a hint on how to get there?

It’s at www.curvemeister.com under "Free Downloads:"

I think maybe what I could use (or would like to
see) is an ICC profile editor or generator that uses
precisely the sort of GUI and interface that you’ve
built into Curvemeister.

I mean… rather than let a stupid flatbed scanner
"automagically" generate an ICC profile — often by trial and error in "tweaking" the thing afterwards — why not just have a GUI that lets you "draw" the profile from scratch — test the results, iterate, fine tune, etc…

There may be a niche there for an inexpensive profile editor, but it would a 3d interface, not just curves, and a way to close the loop by scanning or otherwise measuring the output image.

But a lot can be done just by using curves on your scanned image. For example, if the info palette gives you an RGB value of r0g12b13 in the darkest shadows of your print, and your scanned printout shows r0g0,b0 for the same shadow, add a curve endpoint that takes this to r11,g0 b1, print again, and check the result.

One problem, of course, is the relatively large amount of paper you may potentially use, with no guarantee of converging on a profile you will actually use.

PS: I’ve tried — in the past — a few of the tricks you suggested for tweaking the profiles by doing a
Curves adjustment of the resulting target scan.

But hey, if I’m going to have to do that, I think I’ll
just hang out with Photosphop 4 indefinitely, cuz
"rolling your own" printer profiles just doesn’t seem worth the bother.

Good logic to that, although staying with PS 4 would seem to be an undue amount of punishment:-) If you can print a good neutral step wedge then how far off can you be?

Take care.
—

Mike Russell
www.curvemeister.com
www.geigy.2y.net

Raphael Bustin wrote:

….

I think maybe what I could use (or would like to
see) is an ICC profile editor or generator that uses
precisely the sort of GUI and interface that you’ve
built into Curvemeister.

Raphael,

There’s a little known Diagon Alley corner in Photoshop that may be very close to what you describe.

If you haven’t already, check out the Printer Transfer Function Settings in help. It’s accessed via the "Transfer" button in Printer Setup. This allows you to use a curves type interface to change the C, M, Y, and K curves for your printer.

Although intended for use with an imagesetter, this functionality comes very close to what you describe as an ICC profile editor. And yes, it is supported in v4.0 and even earlier. 🙂

Press ALT or CMD to save/load the defaults for your printer. —

Mike Russell
www.curvemeister.com
www.geigy.2y.net

On Mon, 19 Jan 2004 02:45:14 GMT, Raphael Bustin
wrote:

Why do I bother?

I’ve owned a copy of this product from back before
Colorvision was bought by Pantone, and was known
as Horses LLC.

Mostly I’ve given up on this package (the print
profiler, not the Eye-1) but once in a blue moon
I trot it out again. I’ve never once made a saleable
print using a profile from Profiler Plus. $169 about
four years ago, not one useful print, but dozens
and dozens of test prints and print targets in the
waste bin.

The current version is 2.3. I’m using it with a trial copy of Photoshop CS, on a Win2K machine. My monitor is
calibrated and profiled with Gretag Eye-1 Display.
(It’s a Viewsonic PT-775.)

In past years I’ve used this product with PS 5.5, with
similarly poor results. The print target is scanned on an Epson 1640SU. I’ve tried many combinations of printers, inks and papers (Epsons back then, Canon 9000 now.)

What’s the problem? Deep blue tones get turned
to deep cyan. Otherwise, no big complaints, but
that issue alone makes these profiles unusable, IMO.

So what could I be doing wrong? I can think of
a few reasons for this behavior:

1. The cyan really "is" in the image but my
monitor can’t show it. (But if that were true, why
doesn’t the cyan show when printing without
a profile?) Without the profile, deep blue is what
shows on the screen, and what prints on the paper.

2. Some issue or problem with the scan of the
print target.

3. The product really does suck.

I’ve got the latest version of this, and I’d have to say that the results are somewhat variable. One thing I’ve noticed when scanning the (printed-out) PP color target is that there can be a considerable number of specks and imperfections–you might try loading it into Photoshop and going through it, carefully cloning over the white spots and other flaws. I’m not sure about this, but it would seem possible that good-sized white specks on the darker areas might not be helpful.

I just did a profile using PP for TDK Ref Glossy, and the results are quite good, certainly far better than simply using Epson’s "Glossy" setting (as recommended). The profile cleared up the slight greenish cast that was oitherwise present. Wish I could say the same for Epson HW Matte, the results when using the PP profile were overly contrasty, although the colors were OK. But then, I’ve never had really good success with this paper and the 900/1270/1280.

In reading your description, I’m suspicious that there’s something else going on here besides poor scanner performance. I’d suggest you scan in a reference target, such a a Macbeth chart, and pay special attention to those colors that are giving you trouble. I’ve asked repeatedly for the nominal reflected color density values for a "genuine" Macbeth chart, but have received no useful information as yet. Still hoping. I do have the "real McCoy" on order.

Any other ideas?

I’ve always been baffled that this product magically
works without needing a "reference" target to scan (as does Monaco, for example.) OTOH, I’ve seen
this same problem from profiles I’ve made with Monaco.

Of course it has a reference target to scan, the .pct file you generate and print when beginning ProfilerPlus. The fact that you’re experiencing the same problem with Monaco leads me to think that you’re not getting rid of a "hidden" profile somewhere that’s messing things up, and you’re winding up double-profiling. Does that make sense? What printer are you using? If Epson, are you setting it to ICM control and "no color adjustment"?

rafe b.
http://www.terrapinphoto.com

KenP wrote:
….

I’ve asked
repeatedly for the nominal reflected color density values for a "genuine" Macbeth chart, but have received no useful information as yet. Still hoping. I do have the "real McCoy" on order.

Try this web page, which includes some Lab numbers for the Macbeth ColorChecker, and other fun and interesting items related to color: http://www.brucelindbloom.com/

Timothy Vitale has a good discussion of various targets, including the Macbeth ColorChecker:
http://palimpsest.stanford.edu/byauth/vitale/color-targets/

Ian Lyons did a comparitive review of different IT8 target providers. http://www.computer-darkroom.com/it8cal/it8_page_1.htm

IT8 targets are designed for scanner calibration, and Wolf Faust is a source for good and inexpensive IT8 targets :
http://www.targets.coloraid.de/

I’ve ordered from them and even including shipping from Germany to the US they’re cheap and fast.
—

Mike Russell
www.curvemeister.com
www.geigy.2y.net