LaserSoft Imaging

I am using Ai 5.5.2r02 with a Nikon LS-2000 on a Mac running OS 9.1.2. After finally getting the calibration process to stop hanging on selection of the IT8 reference file (more on that later) I still get a lousy profile. Here's how it goes:

In the Options dialogue, I leave the General settings on the defaults, with a gamma of 2.0 to 2.2 (yes, I've tried all three settings). I set all of the CMS settings to "NONE" , except as indicated in the manual (Internal=RGB, intent=perceptual). I carefully select the indicated area in the KodakE3Q60 target (yes I cut off a little of the colors) using the magnified prescan. I run the calibration module.

I get new prescans that are dark, muddy, and shifted to blue. The face in the Kodak slide has blocked-up shadows. All of the quarter tones in the darker squares look horrible. More to the point, the new prescan bears little resemblance, color-wise, to the actual slide as it looks on a 5000k light box. When I click the Auto button, the colors saturate and shift to a just slightly blue shifted state that is much closer to the slide, but still too dark, and the flesh tones do not have the natural quality of the actual slide (too blue? too cyan?). Another slight improvement is gained by using the pipettes to set the white and black points on the appropriate squares on the grayscale, i.e., the ends within the scan frame. The image is still too dark.

When I try the calibration with an increased gamma, the highlights and midtones wash out. Ditto increasing the exposure time with the analog gain slider. When I delete the calibration, the prescan looks much more like the actual slide, except for an unacceptable magenta cast which I can't seem to neutralize without creating other casts in other parts of the image, particularly the grays and fleshtones. I assume that the numbers created by calibration are supposed to work like a series of subtle curve adjustments working in the background to produce a known state, That known state should translate into a prescan that resembles the actual slide on a lightbox at the same color temperature as a calibrated monitor. In my case the lightbox, monitor, Photoshop 6 working space (ECI-RGB v1.0), and ColorSync 3.0.3 are set to D50 (5000K), with a gamma of 1.8. I use a Colorvision Monitor Spyder and PhotoCal to calibrate, once a week or so.

Several possibilities for this dysfunction come to mind:

1. The reference file is created using patch reader that uses PC gamma and/or a different color temperature (6500K?9300K?).

2. A different color temperature is assumed somewhere else in the code for the module, or Silverfast

3. A different gamma is assumed somewhere else in the code for the module, or Silverfast.

4. It's supposed to create a profile that requires Auto correction and setting of white and black points, in which case it still didn't do a perfect job, and the fleshtones are difficult to correct.

I am at my wit's end (that didn't take long! ha, ha). I am reasonably certain that my brand new Kodak slide is well within the IT8 tolerances. There's just something I'm missing, and I'd love to know what it is. Wizards from Lasersoft, Mr. Lyons, Mr Anders, anyone who has gotten the non-Auto-corrected prescan on a D50 Mac to look like the reference slide please HEEELLLPPP!!!

My profound gratitude, and kudoes to Lasersoft for the 48-bit tools! 5.5.2 is fantastic!

P.S. As for the calibration hangs, disabling all the Nikon native drivers but one, (whichever one you need) and running Tech Tool Pro on the newly installed 5.5.2 before running IT8, as well as discarding all but the needed IT8 reference files did the trick. Somehow the installer seemed to mess up the files. Also, the presence of another Silverfast folder/module seems to mess up IT8. Still haven't solved the problem for my Silverfast (Epson Expression 636). It still hangs on selection of the IT8 reference file. Once again, HEELLPP!!!!

Thanks again.

Marcello Amari
------------------------------------------------------------------------


<font size=-1>[ This Message was edited by: munchello on 2002-02-28 21:01 ]</font>

Munchello,

I'm not at home where I can play with Silverfast right now, but here are some possible problems/solutions:

1. I think Silverfast does the automatically, but if not, make sure that all adjustments are "RESET", i.e., no red colored tabs before you calibrate.

2. After calibrating, make sure the box that says"Activate Calibration" is checked.

3. Use the following settings:

<img border="0" src="http://www.computer-darkroom.com/using_sf5/sf5_image13.gif" alt="Image" width="400" height="451">

Substituting your own Scanner (Transparent) and Internal Profiles.

The biggest issue you're probably having is that your Rendering intent is prbably NOT set to Perceptual. Try these settings (if you haven't already) and tell us how it works for you.

Two last things, 1, I hope Ian L. doesn't mind me linking to his images, and posting in 3 places isn't considered proper forum etiquette (although I understand your frustration).

Thanks,

Ian A.



<font size=-1>[ This Message was edited by: ianders1 on 2002-02-28 21:56 ]</font>

Thanks for your help. Unfortunately, what you've shown and described is precisely what I've done.

The "activate calibration" checkbox checks itself automatically after "successful" calibration. The Scanner profile setting in the Profiles for Colorsync dialog box is greyed out, but checking ColorSync instead of Calibration allows the selection of the SF-T(LS-2000) profile. The Calibration setting automatically finds and uses this profile. Using the ColorSync engine is not an asset. At best it uses the ICC profile generated by the calibration module to produce the same result as using the Calibration setting.

But all of this is not to the point. I know I've followed all instructions and tutorials to the letter, and then proceeded with the slight variations (gamma adjustment, and exposure via analog gain slider) I described in my post, all to some avail, but not within acceptable parameters.

I maintain, perhap wrongly, that calibration should produce an extremely accurate precan of the reference slide, given that the color temp of the lightbox and monitor are the same, the working space chosen in Photoshop and SF match, and that working space corresponds in gamma and assumed color temp to the settings chosen for the monitor. I am using ECI-RGB, which is a D50 working space, gamma 1.8, with a larger gamut than Colormatch RGB so as to include colors printable on a six-color RGB (Quickdraw) printer.

Where this may break down is in the reference file. If the numbers describing the "known state" are used to create a calibration file using a transform that assumes a gamma or color temp different than what is being used, might that not result in a prescan (and scan) that visually differs from the actual film on a 5000k light box?

I realize that corrections will have to be made to different transparencies (obviously). The tools in Silverfast would otherwise be superfluous.

As I understand it, scanning the REFERENCE slide using the calibration just created from the same reference slide and corresponding reference file should produce a prescan that closely resembles the actual slide as seen on a lightbox. The "known state" described in the reference file should correct the scanner errors to produce a calibration file that when applied to the reference slide should produce a prescan (and subsequent scan) that is nearly indistinguishable from the actual slide as seen on a lightbox. Is this wrong?

Sorry about the multiple postings. I am e-mailing all of this directly to Lasersoft, along with my Silverfast log files and screenshots of my settings.

Munchello,

Sorry to hear about all of your troubles. I've thought of a few more things. A, have you tested out your monitor profile with other pictures in PhotoShop, Illustrator, etc. The new Pantone co-branded version of PhotoCal threw me off the first time I used it, since it's a little different, and I wrecked my profile. Also, if the Spyder isn't strongly attached to the screen, any stray light entering the calorimeter can really throw off the profile.

To answer your question, YES, the prescan should be 99% accurate assuming all goes well with IT8 Calibration. Matching scans to screen is usually the easy part. Matching screen to print is where things get difficult.

Also, have you been able to sucessfully calibrate in the past? Are ABSOLUTELY sure you're using the correct IT8 Reference file? You might also want to try using Adobe RGB or something like that just to see if that makes a difference. Also, are you sure that your PhotoShop settings are in-sync with Silverfast? Ian Lyons has a tutorial on that as well.

Since I don't have a MAC, I'm not sure if ColorSync could be throwing everything of, as well. And make sure that Adobe Gamma is disabled.

Since it's occuring on more than one scanner, and you've got all the SF settings correct, it seems most likely that it's your PhotoShop or ColorSync settings. I'll check on your progress later, and maybe the other Ian or someone at Lasersoft has some ideas.

-Ian A.

I get new prescans that are dark, muddy, and shifted to blue. The face in the Kodak slide has blocked-up shadows. All of the quarter tones in the darker squares look horrible. More to the point, the new prescan bears little resemblance, color-wise, to the actual slide as it looks on a 5000k light box. When I click the Auto button, the colors saturate and shift to a just slightly blue shifted state that is much closer to the slide, but still too dark, and the flesh tones do not have the natural quality of the actual slide (too blue? too cyan?). Another slight improvement is gained by using the pipettes to set the white and black points on the appropriate squares on the grayscale, i.e., the ends within the scan frame. The image is still too dark.

The prescan of the IT8 slide after calibration should APPEAR neutral. It should also be the case that using the pipette to check each of the grey patches of the IT8 gives pretty close to R=G=B for most of the mid tones (not likely the full scale). The point being that the numbers should be neutral and the appearance should also be neutral. If the appearance is wrong and the numbers are right you have a monitor calibration problem. If the numbers are wrong then the calibration is wrong.

When I try the calibration with an increased gamma, the highlights and midtones wash out. Ditto increasing the exposure time with the analog gain slider. When I delete the calibration, the prescan looks much more like the actual slide, except for an unacceptable magenta cast which I can't seem to neutralize without creating other casts in other parts of the image, particularly the grays and fleshtones. I assume that the numbers created by calibration are supposed to work like a series of subtle curve adjustments working in the background to produce a known state, That known state should translate into a prescan that resembles the actual slide on a lightbox at the same color temperature as a calibrated monitor. In my case the lightbox, monitor, Photoshop 6 working space (ECI-RGB v1.0), and ColorSync 3.0.3 are set to D50 (5000K), with a gamma of 1.8. I use a Colorvision Monitor Spyder and PhotoCal to calibrate, once a week or so.

ECI_RGB V1.0 means nothing to me. Other than your monitor being selected in ColorSync this dialog is as useful a handful of melting snow.

Several possibilities for this dysfunction come to mind:

1. The reference file is created using patch reader that uses PC gamma and/or a different color temperature (6500K?9300K?).

The reference file uses Lab and XYZ dat and so gamma is NOT a factor. The white point shouldn't be an issue.


Ian

<font size=-1>[ This Message was edited by: ilyons on 2002-03-01 00:55 ]</font>

The gamma and white point colour temperature that you calibrate to should make NO difference so far as Photoshop is concerned. The monitor compensation scheme should make sure that any transforms required are undertaken automatically an so the user should be oblivious to any differences that may exist.

Your description of monitor calibration suggests that your aim point is D50 at gamma 1.8. The choice of D50 is my view and that of many others not good. A more appropriate choice would be 6500K (D65). Making this choice will mean the monitor (outside of Photoshop) will appear less like it requires haemodiaylsis and slightly brighter. This does NOT preclude the choice of a D50 colour space in Photoshop since as I mentioned above Photoshop will handle the conversions on-the-fly.

I doubt that any of the above is the cause of your problems. However, if a problem existed with this version of Silverafst I would have expected more queries by now. It is possible, but why no more queries.? is anyone else having this problem? If so speak up !!!

IanL

<font size=-1>[ This Message was edited by: ilyons on 2002-03-01 01:06 ]</font>

I have a similar problem using Silverfast V5.5.2r13 on an and Epson 1680. I cannot obtain a calibrated profile scan that looks like the target. I scan the target with all colour management features off and I obtain a result which looks reasonably close to the target on the screen. I then go through the calibration procedure which should improve things but only results in the Cyan and Yellow columns on the target looking way off. Using the densitometer readings before and after calibration gives the following results for column 13 on the target which is the graduated Cyan column

CMY densitometer readings Before calibration:

C 2 13 23 33 43 54 64 73 79 86 92 99
M 3 2 1 1 3 5 8 13 20 26 34 49
Y 5 5 7 9 10 11 13 14 15 16 16 16

CMY densitometer readings after calibration:

C 5 25 47 59 68 77 82 84 86 85 86 88
M 5 5 9 4 0 0 0 0 0 0 0 12
Y 4 3 0 3 9 13 16 16 14 10 4 0


Similar results are obtained on Column 15 the graduated yellow column.

Now I this may not be correct but I expected an even ramp of the Cyan after calibration similar to as it was before calibration, I would not have expected calibration to make things worse.

This effect occurs on both reflective and transparency calibration but to a lesser extent on transparecy.

The procedure I have used for calibration is Tutorial 1 - Basic steps to IT-8 calibration on Ian Lyons website

Regards


Graham



<font size=-1>[ This Message was edited by: mackarel on 2002-05-12 14:34 ]</font>

Please ignore my last post on this, it was my fault as I didn't have the marque in the correct position around the target when calibrating. All is o.k. now

Graham