Cyan/Blue colour shift with latest versions of Vuescan

[Graham Russell]

I've rescanned a couple of slides using Vuescan 7.6.19 previously scanned
with Vuescan 7.5.X using a Nikon LS4000. There seems a noticeable blue/cyan
cast with the new scans. Media is Kodachrome. I've tried both with and
without IR cleaning. No difference. I think the scans using earlier versions
of Vuescan are more correct.

Colour correction for both scans - black point 0.5%, white point 0.5%, white
colour balance.

I tried attaching low res jegs to illustrate the differences, but I think
the message got blocked. Happy to email privately if it helps.

I really need to re-scan these slides and I want to make Vuescan work. Am I
doing something wrong? Yes I have removed vuescan.ini - no change. Has the
introduction of scanner calibration altered some basic colour settings? I
have not calibrated my LS4000 scanner. Do I really need to?

I want to use Vuescan because it does a much better job than Nikon Scan with
IR cleaning on Kodachrome. Although Nikon Scan produces better colours with
this slide compared to the latest Vuescan image.

Thanks,

Graham.

[Bart van der Wolf]

"Graham Russell" <graham-...@dsl.pipex.com> wrote in message
news:3e613f2c$0$3410$cc9e...@news.dial.pipex.com...
SNIP

> I really need to re-scan these slides and I want to make Vuescan
work. Am I
> doing something wrong? Yes I have removed vuescan.ini - no change.
Has the
> introduction of scanner calibration altered some basic colour
settings? I
> have not calibrated my LS4000 scanner. Do I really need to?

The default (Built in) profile for all scanners has been redone. You
only need to make a profile if you want to be very close to a correct
color response. This doesn't mean your colors will be as you want
them! Profiling only tries to get the relative differences
perceptually correct.

Your problem is probably not a profiling, but rather a colorbalance
issue. Have you tried right-mouse-clicking on a neutral gray or
non-saturated white object in the preview or scan pane?

Bart

[Graham Russell]

"Bart van der Wolf" <bvd...@nospam.nl> wrote in message
news:3e6157d7$0$49102$e4fe...@news.xs4all.nl...

I've tried right clicking on a neutral grey area. It helps, but still not
quite right. I didn't need to do this with the earlier scans, I just used
white colour balance. I guess something has changed inside the colour
balance system with the latest versions.

[Greg]

FWIW, this might or might not be related, but I'm just looking at a scan now
which has a lot of sky in it, and I observe that there is a moderate shift
towards cyan in processed output files from Vuescan (colour balance set to
none), if I compare the results to a *raw* scan which I tag with either a) a
custom profile made with Vuescan, or b) a custom profile made with Little
CMS.
The raw scans, in conjunction with either custom profile, produce a very
accurate reproduction of the sky hue.
The processed results look prettier, but not as accurate.

Even if I use a Media Type of Slide film (which does a bit of colour
balancing, yes?), and then do an Auto Levels in Photoshop, the sky is still
shifted towards cyan, both before and after the auto levels. (and there is
nice bright white non clipped area in the image for the auto levels to use
as a reference)

Note that when Vuescan is configured to use a custom profile, the cyan shift
in the *processed* output file seems to be worse than when it's using the
built in profile.

I haven't done any comparisons with older versions of Vuescan.

Greg.

1. Opening the raw file in Photoshop, and tagging with a custom profile that
I have made (using Vuescan), the colour accuracy looks very good.

2. Opening the *processed* file, which I configured Vuescan to create in the
Ekta Space, the sky has shifted very noticably towards cyan. This processed
image looks prettier, but it does not look nearly as accurate.

3. Opening the same raw file (as in 1), and tagging it with a custom profile
I have made with Little CMS, the sky looks exactly the same colour as it
does when using the Vuescan profile in 1.

Note that I locked the exposure on the same setting I used during profiling.

4. Setting Vuescan to a more automatic/default setting

"Bart van der Wolf" <bvd...@nospam.nl> wrote in message
news:3e6157d7$0$49102$e4fe...@news.xs4all.nl...
>

[Greg]

The stuff after I have signed my name was not supposed to included - sorry
about that. Ignore it.
(I was going to include more detail about the testing steps, but it's dinner
time)

Greg.

"Greg" <sulliva...@bigpond.net.au> wrote in message
news:Pqk8a.4196$Xh1....@news-server.bigpond.net.au...