linear gamma profile slide for scanner?

[smilem]

I don't use i1profiler and therefore can't comment on that.

So you use vuescan to make the profile, if not what software do you use?

[Christoph C. Feldhaim]

I use Vuescan to create my scanner profiles, since I have this Velvia target which doesn't get recognized by Silverfast.
For Kodachromes I use the Silverfast target and Silverfast to create the profile.

But there's another thing I just re-read from the Colorperfect people:
They basically want you to calibrate your file in a way, that the RGB numbers are linearized.
Link: http://www.c-f-systems.com/DunthornCalibration.html

You can achieve that by converting your scanned image file (with the attached input profile you created before) to an L* Gamma Colorspace.
There are tools around which allow you linearize colorspaces to have an L*-ized Gamma.
In that case RGB values are more or less linear with L* values and thus with real world perceptual Lightness.

Cheers
~Chris

[Mark D Segal]

I use Vuescan to create my scanner profiles, since I have this Velvia target which doesn't get recognized by Silverfast.

Cheers
~Chris

Hi Chris,

I assume you've tested profiles created in both ways on several of the same photos, so I'm interested to know in what ways you find the profile created with the Velvia target delivers better results than SF8 Auto IT8 process using their own IT8 target?

[smilem]

But there's another thing I just re-read from the Colorperfect people:
They basically want you to calibrate your file in a way, that the RGB numbers are linearized.
Link: http://www.c-f-systems.com/DunthornCalibration.html

Calibrate what file? My english is not that great but this does no sound like automated process like ICC profiles?
Therefore if it provides better results the work required for it I think is not justified to any better color reproduction, they could have created a plugin for this process, but they did not, I wonder why?

[Alan Klein]

Why not just scan flat and adjust the colors in post to what you want?  If you incorrectly set the adjustments during the scan , wouldn't you have to rescan?  It seems simpler to do in post.  No?

[Christoph C. Feldhaim]

Hi Chris,

I assume you've tested profiles created in both ways on several of the same photos, so I'm interested to know in what ways you find the profile created with the Velvia target delivers better results than SF8 Auto IT8 process using their own IT8 target?

The Velvia target covers a different gamut and is is better differentiated in parts of the the extremes.
My impression is it gives better results, especially when using Velvia slides which get a bit compressed at the ends with the original SF targets.
In the highlights it gave me better greens and oranges, in the shadows better blue-magenta tones.
In other areas it was a bit reduced.
Over all I think the perfect target yet has to be invented.
I attached the Velvia based ICC profile I use with my LS-9000.

Cheers
~Chris

[Mark D Segal]

Why not just scan flat and adjust the colors in post to what you want?  If you incorrectly set the adjustments during the scan , wouldn't you have to rescan?  It seems simpler to do in post.  No?

The whole idea of profiling a scanner is to get the colour interpretation as correct as possible straight out of the scanner without having to do a whole lot of time consuming adjustment in post-processing. As well, without profiling, beyond a certain degree of poor colour rendition from the scan, correction becomes difficult and sub-optimal. The obsession with linearity is just that - however, the main thing is to assure that neither highlights are blown or shadow detail plugged-up in the unadjusted scan result. My long-running experience with film-scanning indicates that SOME amount of image adjustment is necessary before or after the scanning process, but good profiling is a big help to cut work and improve outcomes.

[Christoph C. Feldhaim]

Why not just scan flat and adjust the colors in post to what you want?  If you incorrectly set the adjustments during the scan , wouldn't you have to rescan?  It seems simpler to do in post.  No?

The method is to keep the raw scan, but you use the attached input profile to get correct colors from it.
If you make it to create a better profile later you can attach it to the file and immediately profit from better colors.
The raw data is never changed.
When converting a negative scan to a positive you make a duplicate (positive) working file.

[Mark D Segal]

Over all I think the perfect target yet has to be invented.
I attached the Velvia based ICC profile I use with my LS-9000.

Cheers
~Chris

I think you are right about there being no perfect target, but have you tried the Hutchcolor target? Thanks for providing your profile, but my scanner is a SC-5000-ED - not quite the same animal.

[Christoph C. Feldhaim]

I think you are right about there being no perfect target, but have you tried the Hutchcolor target? Thanks for providing your profile, but my scanner is a SC-5000-ED - not quite the same animal.

No, never tried it.
Reading their website now ...

[Christoph C. Feldhaim]

Calibrate what file? My english is not that great but this does no sound like automated process like ICC profiles?
Therefore if it provides better results the work required for it I think is not justified to any better color reproduction, they could have created a plugin for this process, but they did not, I wonder why?

I personally do not calibrate to their method.

I have my scanned and profiled file and squeeze it though the plugin and am happy with it.
Sometimes leave it in the original input profile, sometimes I convert it to ProPhoto or a normalized ProPhoto which has L* type gamma and not the standard gamma of 1.8.
The advantage of a normalized Prophoto is, that is has more numbers in the shadows and highlights and toning operations sometimes give better results, especially in the shadows.
So - no need to make rocket science out of it.

[smilem]

Here I read that LCMS profiler (LProf) does a good job at making scanner profiles.

http://www.tkupfer.de/imaging/Scan_Profiling.html

I downloaded latest LCMS http://sourceforge.net/projects/lprof/files/
And when I load my reference (attached) it simply does not work, while with demo file it does work.

My reference file is not read by LCMS, the rectangular check boxes does not appear on the target when you select four corners. What a nice BUG, I hope somebody fixes this.