How to profile with lcms and scan with Vuescan ?

[Jean Delmas]

I try to to find the best workflow for profiling with lcms and scan
slides with Vuescan.
Can you help me to correct these two workflows and to tell me which is
the best ?

WORKFLOW 1

1 – 1 SCAN CHART AND MAKE PROFILE
- Preview Wolf Faust Ekta IT8 chart with Vuescan.
- Select "Input/Lock exposure" and remember "RGB exposure"
- Scan chart with Vuescan to a RAW TIFF file
Question : Raw 48 bits RGB or 24 bits ? 24 bits are gamma corrected
(2.2) but 48 bits have gamma=1 which is very bad for lcms. How to
gamma correct 48 bits TIFF chart file ?
- Produce scanner profile with lcms…

1 – 2 SCAN SLIDE
- Scan slide to Raw TIFF file (24 or 48 bits RGB ?) with Vuescan and
same RGB exposure as for chart
- Attach lcms scanner profile to this slide TIFF file in Photoshop…

WORKFLOW 2

2 – 1 SCAN CHART AND MAKE PROFILE
- Preview Wolf Faust Ekta IT8 chart with Vuescan.
- Select "Input/Lock exposure" and remember "RGB exposure"
- Scan chart with Vuescan to a TIFF file with no profile embeded (48
bits RGB, Exposure clipping=0, Color balance=None, Brightness=1,
Scanner ICC profile=blank, Output color space=Device RGB)
Questions : are these options OK ? Is strange Scanner ICC
profile=blank OK ?
- Produce scanner profile with lcms

2 – 2 SCAN SLIDE
- Scan slide to 48 or 24 bits RGB TIFF file with no profile embeded
(same RGB exposure and options as for scanning the chart).
- Attach lcms scanner profile to this slide TIFF file in Photoshop…


Jean Delmas

 

[nikita]

Jean,

search this forum (scanners) for the name nikita. I've written a few
posts related to your questions. They might answer them. 24/48
shoulden't do much of a difference in the profiling stage. The
gammabump can be taken care of quite easily.

Workflow 1 is the most optimal when using thirdparty solutions. For a
gamma adjustment of the RAW High Bit target files as well as the final
Hightbit RAW scans ( before assigning the finished profile ). Go to
the following link/site and choose in the leftcolumn: Linear Intensity
space. Scroll down the page and find readymade gammacurves to load in
Photoshop curves dialog.

Don't digg to deep into the rest on this site as it will eat you, and
the rest of the summer, up. It just complicates your life and a very
knowledgeble person at the Adobe Photoshop Team seriously writes
warnings for this site. Great battles have seen daylight through the
years between these two combats. However, the gammacurves are very
useful in any case. You can put them in a few actions in Photoshop and
quickly load them by pushing buttons. Put an ASSIGNING the profile and
Convert to workspace into those actions too when the profiling is
done.

The gamma of choice doesn't need to be "2.2". Make some RAW scans and
read the patch GS 10 or 11 (depending on reflective or transparent
targets) and use a gamma curve that deliver a patch value of 100-110.
You can go higher too. Then profile and be happy.

http://www.aim-dtp.net/aim/calibration/index.htm

nikita

 

[Jean Delmas]

Nikita, thanks a lot for your interesting answer.

(e-mail address removed) (nikita) wrote >

Workflow 1 is the most optimal when using thirdparty solutions.

Can you tell me why the two workflows are not equivalent ?

The gamma of choice doesn't need to be "2.2". Make some RAW scans and
read the patch GS 10 or 11 (depending on reflective or transparent
targets) and use a gamma curve that deliver a patch value of 100-110.
You can go higher too. Then profile and be happy.

Excellent advises, specially the last one. I immediately apply it.

Jean

 

[nikita]

Jean,

the RAW is the purest thing. The Vuescan NON with locked settings
might be a good alternative, but it goes through more of Vuescan than
the pure RAW. Exacly what it goes through I can't answer and it may
not be much. As we know how many things in Vuescan that can effect
another, I would definitely chose the RAW. Vuescan doesn't always work
as intended and bugs are showing up frequently. That said, a scanner
*can* be profiled in any locked and known state. What you lose on that
journey regarding gamut etc is another question. Vuescan is a
fantastic little piece of scannerdriver, but you shoulden't need more
features than the RAW from it when profiling positives externally. The
same goes for Silverfast which I prefer very much these days. The RAW
delivered from that application is outstanding and worth the whole
pricetag itself. Even on the negside Nikons and negRAWs (48 bit HDR
output) is in it's own class. Both in B&W negs and colornegs. But the
positive output from it is legendary. The Toy-look-a-like interface is
something I can live with in Silverfast as Vuescan is a mess of
confusing options that interacts in a unpredicable way many times. As
far as possible I try to just be a RAW grabber to eliminate
unpredicable behaviours. In both Silverfast AND Vuescan. It pays off
in the end of the day.

If we're profiling positives or want to work with traditional tools,
Photoshop is there. In cases when we're using autoflows thanks to huge
masses of work, the timesaving for an autocropping for example, is
eaten up when trying to ballance a missbehavour in the Autosettings.
If any auto is to be used, go for a flow that gives you a fully
controllable semiautoconcept in Photoshop instead. I promise – it will
deliver faster in the end. You just have to get it into habits like
pulling the gearshift in the car. Well, you Americans don't know what
that is, but trust me anyway.

If the "Brightness" (compositechannel) in Vuescan NON mode adjusts
gamma for the patch GS10/11 in a better way than the downloaded
gammacurves on a RAW, hence works better with that particular freeware
profiler, by all means use that route. It would be stupid not to just
because RAW mania....on the other hand...reading this scannerforum
show up a bit of Vuescan mania, if I can speak frankly here. One
doesn't have to pull every little switch just because they're there on
the dashboard of an aircraft, so to speak. Simplify. Regardless of
software and any features it holds.

I would suggest that you profile using both alternatives and trust
your eyes and common sense to decide which of the alternatives that
delivers the gods.

Good luck and if possible; try a commersial thirdparty solution to
like Profilemaker or Monaco Profiler Platinum/Ez. They will build
another kind of profiles. They doesn't have to better, but they're the
industry standard applications...

nikita

 

[Jean Delmas]

(e-mail address removed) (nikita) wrote :

The RAW
delivered from that application is outstanding and worth the whole
pricetag itself.

But, if RAW is RAW (with no correction by software) how can Silverfast
RAW be better than Vuescan RAW ?

If we're profiling positives or want to work with traditional tools,
Photoshop is there. In cases when we're using autoflows thanks to huge
masses of work, the timesaving for an autocropping for example, is
eaten up when trying to ballance a missbehavour in the Autosettings.
If any auto is to be used, go for a flow that gives you a fully
controllable semiautoconcept in Photoshop instead. I promise ? it will
deliver faster in the end. You just have to get it into habits like
pulling the gearshift in the car. Well, you Americans don't know what
that is, but trust me anyway. How do you know that ?

I agree with your philosophy. I am not american, but I am the only
french person in France who can only drive automatic cars. How do you
know that ?

Thanks for your advises.

Jean

 

[nikita]

Jean,

A RAW is effected and depending on a few things regarding exposure,
focusing etc. It is always the scannerdriver/software that makes those
adjustments. With Silverfast you can also use for example true ICE
dustcleaning directly on the RAW files. You can also use Vuescan with
dustcleaning on the RAWs, if set up correctly. BUT that isn't true
ICE. Silverfast uses true ICE. Vuescan uses it's own which isn't bad
at all, but different.

The negRAWs when set to negafix HDR 48 bit output is a prepaired kind
of RAW. Not a completely true RAW, but sligtly tuned to be a better
base for the Silverfast Negafix system that can be found in the
ordinary Ai6 as well as the HDR application. The HDR is used to rerun
the RAW in the same way Vuescan can reprocess RAW files. The
Silverfast neg RAWs on the Nikons are so good that I can't find words
strong enough to describe it. Yes, they come with the ordinary mask,
yes they are cyantinted, yes they are not inverted to a positive...we
have to work on them by inverting, setting endpoints and tune gamma
but the respons to that work is better than using anything else for
negworks. Check the "use gamma for HDR output" in the options and have
the RAW gammacorrected on the fly. You still have to tune the gamma
with curves in Photoshop with some smaller moves. That is a piece of a
cake. You have to clean up the mask by setting endpointballance in
each separate componenchannel. The Photoshop Autocolor used with Bruce
Fraser technics will get you vey close. Or use the manual gearshift
all the way...that is Alt+dragging endpointsliders in each shannel. In
any way you chose you will find that you have complete controll over
endpointclippings. You don't have to scratch your head over clipvalues
adjustments in scannersoftwares that clips the ends all by itself or
is hard to judge in the preview. You have it all in the RAW and can
use sophisticated tools and technics in Photoshop to do the endpoints
in a blink. If you still want to tune as much as possible in a
scannersoftware, the HDR and negafix have tools that can deliver
identical result with the Photoshop Autocolor or manually. There are
features in Negafix that can clean up the greyramp in the negs and
finetune saturations etc allover the scale. This is really unique and
a kind of negative profiler as you're saving presets from it. But with
the Nikons it works directly in Photoshop as is. The HDR can build
batchflows and special scanning concepts for larger volumescans that
some may prefer to use Photoshop. It all depends on what you intend to
work on.

Try this; output RAWs positive and negatives from both Vuescan and
Silverfast. Work on them completely by hand or semiauto in
Photoshop.You will DEFiNITELY find the negRAWs in Silverfast/Nikon
high above Vuescans when it comes to manual gearshift handling. The
positiveRAW will show you differences in Dustcleaning ICE and perhaps
other things. If you find that Vuescan deals with that better there is
no question about it – use it. Remember to make the same tests with
next scanner you buy. Both Silverfast and Vuescan can adjust analog
gain and exposure within their own concepts on the RAWs to max out
each channels CCD response. Both can lock this values and by that be
used as a zoombie RAW scangrabber for profiling or by-hand-work.
Remember that the negRAW in Silverfast works best when letting Silver
use autoexpousre On all the time without any analog gain adjustments.
With my Nikon 4000 a gamma set to 2.2 for the RAW output works better
than higher or lower values.

It may seem that I degrading Vuescan. I am not. I have had a long
scanninglife using Vuescan and I admire the work the programmer is
putting into this application. However, I'm not more stupid than I
will be openminded enough to see advantages in others when they cross
my journey. Because there are alternatives in both these apps and
userconcepts. Vuescan has always been a very lowpriced application.
Nowdays it is more in the right pricelevel. It's really worth the
higher pricetag. That pricetag also means that it comes closer to
Silverfast which always had a reputation to be set to high in price on
the other hand. Silverfast has been mudslinged through the years by
different reasons. Some have been justified, but today it's a very
solid thing. The price is still in the higher class, it may feel to
high if not in a need for many of the exclusive tools it offers. It's
a very subjective choice to make if the RAW quality *only* is worth
the money or not. When I consider the timesaving and direct workflow
on the negativework (when got used to the gearshift moves) I
personally feel no doubts.

About you beeing the only french person who can only drive automatic
cars....I did'nt know that you french guys could drive cars at
all..... But you're very good at honking the horns.....You're are
also very good at journalism, films and conversations. All of them
very nice features. But you're lousy at winemaking. What a crap. )

I wish you the very best in your search for the optimal scanning
workflow.

kind regards,

nikita

 

[Jean Delmas]

(e-mail address removed) (nikita) wrote

If the "Brightness" (compositechannel) in Vuescan NON mode adjusts
gamma for the patch GS10/11 in a better way than the downloaded
gammacurves on a RAW, hence works better with that particular freeware
profiler, by all means use that route. It would be stupid not to just
because RAW mania....on the other hand...reading this scannerforum
show up a bit of Vuescan mania, if I can speak frankly here. One
doesn't have to pull every little switch just because they're there on
the dashboard of an aircraft, so to speak. Simplify. Regardless of
software and any features it holds.

I would suggest that you profile using both alternatives and trust
your eyes and common sense to decide which of the alternatives that
delivers the gods.

Following your advises, I have done this :

- It is impossible to change gamma on RAW files produced by Vuescan
(gamma=1 if 16 bitsRGB, gamma=2.2 if 8 bits RGB)
- BUT when you produce TIFF (not RAW) files with all "options to
none", Vuescan gives EXACTLY the same file... and permit to change
output gamma (option Brightness = coefficient which multiply 2.2
default gamma)
- So, following this last method, I have made scans of my Wolf Faust
IT8 chart with 2.2 2.4 2.6 2.8 and 3.0 gamma correction.
- First method to have "optimum scanner gamma" : what is the scan
where the gray GS 11 patch is 100-110 ? The answer is gamma 2.4
- Second method : which scan of the chart gives the best lcms profile
(minimum dE) ? The answer is also gamma 2.4

I am very happy with this result...

Now, my flow chart is :

SCAN CHART AND MAKE PROFILE
- Put all Vuescan corrections options to "None" BUT Brightness=1.0909
which means gamma=1.2x1.0909=2.4)
- Preview Wolf Faust Ekta IT8
- Select "Input|Lock exposure" and remember "RGB exposure"
- Scan chart
- Produce scanner profile with lcms

SCAN SLIDE
- Scan slide (no profile embeded) with same options as for scanning
target, including "RGB exposure" and "Brightness=1.0909"
- Attach lcms scanner profile to this file in Photoshop.

Jean

 

[Erik Krause]

The negRAWs when set to negafix HDR 48 bit output is a prepaired kind
of RAW. Not a completely true RAW, but sligtly tuned to be a better
base for the Silverfast Negafix system that can be found in the
ordinary Ai6 as well as the HDR application. The HDR is used to rerun
the RAW in the same way Vuescan can reprocess RAW files. The
Silverfast neg RAWs on the Nikons are so good that I can't find words
strong enough to describe it. Yes, they come with the ordinary mask,
yes they are cyantinted, yes they are not inverted to a positive

Strange. Just a view minutes ago I scanned some Fuji Reala Negs on my
nikon with vuescan 8.0.2 to RAW files. I dragged a RAW file to
photoshop, set gamma to 2.2 and inverted. What a suprise: Very little
cyan cast, pretty colors a bit washed out (I use Reala because of it's
incredible dynamic range). Setting black point, wonderful.

The secret: Fixed CCD exposure to a non clipping value. And vuescan
seems to adjust the per channel CCD Exposure to compensate for almost
all orange mask. If I let vuescan do the processing the result is even
better, if I used fixed and normalized film base color...

 

[nikita]

Erik,

Vuescan is using a fixed maskremoving in the negative RAWs and that
will be the same whatever negfiltype is scanned. That's at least what
has been the case till now. Version 8 may be different, but I doubt
that. The "floating" filmbase clean up is in the processstep of
Vuescan, not in the variation of channel CCD of the RAW. (Please,
correct me if I'm wrong). The locked exposure of the negRAWs to avoid
clipping close to the filmbase is not an issue. It never clips on
autoexposure of the RAW if not tweaked to do so. But locking filmbase
+ mask for processing and keep a constant state is something else.

The negative RAWs from Vuescan and Silverfast is very different from
each other. The cyan tint in Silver is not heavy even if that sounded
that way...If you compare a handworked RAW from that Reala to these
Silver non-so-dense RAWs, you will see differences. Regardless if we
compare them in the state after just endpointballancing in all three
channels – or after more work, you will see differences in saturation
and hues. Look at it and figure what looks most "true" even if this is
a very a subjective thing with negs. I've done it so many times and
with so many different films. The Silver negativeRAW is in it's own
class. Neither Vuescan negativeRAW or Silverfast negativeRAW is a true
RAW (like the positiveRAW is), both are prepared. But Silverfast
negativeRAW is prepared more than just a basic maskremoving. It's
easier to work on directly with Photoshop tools and take you to the
finished picture with less moves. By that I don't mean managing to
remove the filmbasetint only...but hues,saturation and allover
appearance. Vuescans negative RAWs are..."rawer" and that means more
work when scanning negs.

You can play with the manual analog ballancing in both Vuescan and
Silverfast to max out each channel and clear up the mask as far as
possible. The dynamic range isn't an issue so that's over the top and
will not – according to my tests – create better endresult if looking
at more than the maskclean up....what you get on that route is a
"roughness" that we don't normally associate with negative work. I've
done very timeconsuming tests on how to max out the best exposure for
the negs on nikon 4000.

Actually if Silverfast negRAW is doing the exposure automatically and
no manual tweaking of the channel exposure, it will outperform both
Silvers and Vuescans manual exposure tweakings regarding
alloverballance; hues and saturations. That's why I'm pushing this
"negRAW" thing like I do. Not just to be stubborn.

I would even take it further; please compare the finished and edited
SilvernegRAW to an autoprocessed/manually tweaked Vuescan that has
come through the normal output.....and polished in as much as possible
in Photoshop. Write down all moves on each file. Please do this at
files that isn't "idealfit" for Vuescan too. I mean those ones that we
have to tweak and try each little slider we can find in Vuescan to get
it to deliver. But doesn't succed...

Make a sum of the time in each case. If you get Vuescan to do deliver
the better topwork at the very end in these negative work from the
nikons, good for you. I maybe have other subjective prefrences
regarding how I want to get there and how the colors (and the rest)
show up. It could also be possibel that we work differently in
Photoshop.

If using an ICE aware scanner, compare the dustcleaning part in these
RAWs to and if a focusing maniac, check out the pointfocus dialog in
Silverfast that pops up with magnified before and after + sliders for
tweakings. Useful? That's up to you – but it effects the RAWs and give
control.

kind regards,

nikita

------------

Jean,

ahh, finally you have found the flow that you like. I agree with you
that it's easier to use the slider for brightness/gammatweaking than
load curves. To me it doesn't take much more time as I always have to
ASSIGN and convert to workspace when the file opens in Photoshop. I
use actions and it just takes a button to push in the actionpalette.
It's also very good to hear that NON is just delivering an identical
RAW file, inverted and gammacorrected without any interferance from
the application itself. As long as no bugs creeps into that in future
updates it's fine.

Glad to hear that the gammatuned greypatch 10/11 delivered the best
base for the profiling too.

take care,

nikita

 

[Ed Hamrick]

Vuescan is using a fixed maskremoving in the negative RAWs and that
will be the same whatever negfiltype is scanned.

No, this isn't correct. The film base color is determined
for each image, unless the "Input|Lock film base color" option
is set.

The negative RAWs from Vuescan and Silverfast is very different from
each other.

It's quite possible that Silverfast is using a different ratio of
red, green and blue CCD exposure times than VueScan, and it's
possible that Silverfast isn't using gamma 1.0 (although I have
no idea, not having looked at these files). However, raw files
are raw files - they're just data from the CCD of the scanner.

Regards,
Ed Hamrick

 

[gerhard.fuernkranz]

It's quite possible that Silverfast is using a different ratio of
red, green and blue CCD exposure times than VueScan, and it's
possible that Silverfast isn't using gamma 1.0 (although I have
no idea, not having looked at these files). However, raw files
are raw files - they're just data from the CCD of the scanner.

Ed,
at least for my Epson flatbed scanner, I've the impression that Vuescan
sends the command "color correction = CRT" to the scanner (and not
"color correction = no correction"), instructing the scanner to perform
some color adjustments internally, instead of delivering the actual raw
CCD data. Is my impression wrong?

Regards,
Gerhard

 

[Ed Hamrick]

at least for my Epson flatbed scanner, I've the impression that Vuescan
sends the command "color correction = CRT" to the scanner (and not
"color correction = no correction"), instructing the scanner to perform
some color adjustments internally, instead of delivering the actual raw
CCD data. Is my impression wrong?

Yes, I had to do with with Epson scanners to get the color space
set to sRGB. Otherwise, I would have had to calibrate each Epson
scanner model separately.

However, it uses no correction when scanning negatives, since this
seems to separate the dye layers better than using sRGB.

Regards,
Ed Hamrick

 

[Erik Krause]

Vuescan is using a fixed maskremoving in the negative RAWs and that
will be the same whatever negfiltype is scanned. That's at least what
has been the case till now.

If you mean the typical adjustment for channel RGB to compensate orange
mask I will agree. This is most likely fixed in vuescan.

Version 8 may be different, but I doubt
that. The "floating" filmbase clean up is in the processstep of
Vuescan, not in the variation of channel CCD of the RAW. (Please,
correct me if I'm wrong). The locked exposure of the negRAWs to avoid
clipping close to the filmbase is not an issue. It never clips on
autoexposure of the RAW if not tweaked to do so.

You think? Some versions ago Ed introduced "Exposure clipping" to
satisfy those who wanted to avoid clipping based on CCD exposure.
Standard setting is 0.1%, so exposure is adjusted in order to have 0.1%
of all high value pixels clipped. If you set it to 0.0% you get more
shadow detail for your negs (what you will need only, if you want to
use the entire dynamic range of color negs).

 

[nikita]

Ed,

are you telling me that Vuescan are adjusting the CCD exposure (for
the RAW scan) of the three channels inependently from each other...
and for the negative RAWs not using just the fixed ratio for a general
fixed maskreducing in the RAW as you mentioned in the manual earlier??
Are these channels floating, not just composite adjusted regarding RAW
exposure? I guess you're talking about the processed cropped tiff
now? Isn't the filmbasecolor naild in the processingstage? Otherwise I
missed a very important detail in my understanding of Vuescan through
the years. In such case I appologize very much for giving this
desinformation here at the forum. Really. I'm really sorry for that.

In the other comment you are telling me that NEGATIVE RAW from Vuescan
is the same as NEGATIVE RAW for Silverfast by saying that RAW is RAW.
The different CCD exposure of the channels isn't the only difference
as you think, go check it up.. I assure you that this isn't the case
for the HDR 48 bit output when Silverfast is set to Negative (not
negative Direct, which is a setting that behaves like Nikonscan as it
is based on a part of that Maid Module). These Nikon negative RAW
files are adapted for the rerun in Negafix in HDR. Earlier Negafix had
great problem with Nikonscanners. It simply delivered crap and it
didn't matter that much if you were a little devil of knowledge and
scanningskills. Crap was what you had from it with Nikons. Silverfast
had to eat that crap up even here at the google scannerforum. Today
and since.. say a about year back in time they "tuned" the
*negative*RAW from Nikons to be a better base for Negafix concept –
and they did that extremely well. I don't know what tuning they go
trough but it works. It's these RAW scans I'm writing my ass off here
for. The reason is that **NikonScan** always had more or less clipping
when scanning negs by default, if not making workarounds like scanning
as positive then inverting, analog gain adjustments etc etc. You name
it. Many people liked *Nikonscan* as a fast, simple and direct
everyday way of producing work. Other turned to Vuescan, as I did too,
for the Nikons. Lately I've found that Silverfast was very useful and
delivered real good stuff even when I didn't needed the toolpackage we
normally think of when Silverfast is mentioned. The simple approach to
use this Negative RAW output and the other features I already
mentioned in my posts, is what I try to share here. Someone may
benefit from it, I dunno.

Raw files are RAW files, you say. Yes that is true. But there are
things that effect the RAW files and can make differences. I already
mentioned that too. Those who feel these things makes a difference may
chose either Vuescan or Silverfast depending on their preferences.
What pokes me in the side a bit is that people here are defending
Vuescan with tests on Vuescan only....including you when you tell me
that RAW scans are RAW scans. ( Well, I'm shooting myself in the foot
a little now as I haven't tried Vuescan 8 yet ). One can get the
impression that this forum is THE Vuescan forum....based on this Vue
patriotism. There are very few Silverfast discussions or questions
here. Those that tried to keep the nose above waterlevel were quickly
killed by Vuescan fans(atics?).

I've been trying to explain that the NEGATIVE RAWs are unique tuned
and not true RAWS in Silverfast Nikon. I've also been trying to tell
you all in here that even if the RAW is a CCD grab, there are things
that effects making even the RAW. Take it or leave it, pals....

kind regards,

nikita


Erik,

ok...I must confess, I don't have used Vuescan latest ( I'm running
Silverfast you know. Let me put it this way; I've never had problem
with Vuescans Exposure of negative RAWs. It's always been one of the
extremely good things with Vuescan, from my point of view; the non
clipping close to the filmbase and blackpoint (negs). Slide/positive
work might be another issue as the dynamic range is filled from one
end to the other. So, I won't argue about that. Especially as clipping
in Silverfast for the manul workflow from the RAWs is so controllable
directly in Photoshop Levels – manually Alt dragging or using Bruce
autocolor stepping from 0 uppwards......and there are always headroom.

I'll see if I would generate extra energy and download the latest
Vuescan to try the settings you gave me and look at new features too.
Hope the eventual bugs will be left out as those are one of the
anoying features I remember since I lived in that patrioism camp
myself. Each upgrade I had to go through the whole arsenal and make
very pedantic evaluations and comparisons with earlier scans. You know
what I mean. I remember the days when that Brightness slider went
trough so many phases. One day it way a true gamma slider, next update
it was a "special", next it was a combo (good luck Jean.....the
duststuffcleaning was also a source for headscratching, one day it was
just perfect at the lightest setting....next day it was inbetween
somewhere or had a very different character. Ed experimented wildly to
give us better stuff, but I was lost in the wood and didn't feel good.
The manual did never show these things in any special chapter that
said: These are the differences in practice compared to the prevoius
version. Watch out for this and that. Just one sentence only can be
found that told you things happened. Then you're on your own in the
cold again. Usually this is called BETA testing, or not even that.
BETA testers gets more info on changes like this. Do I drag this to
far? Not from my point of view. But again, my true intention isn't to
puke over Vuescan. I just want to show you alternatives.

Now, back to the RAW stuff at my desk!

all the best to you,

nikita

 

[nikita]

"Yes, I had to do with with Epson scanners to get the color space
set to sRGB. Otherwise, I would have had to calibrate each Epson
scanner model separately.

However, it uses no correction when scanning negatives, since this
seems to separate the dye layers better than using sRGB."

---------



Actually that would mean that when positive scanning using an Epson,
the real scanner RGB is degraded to s-RGB regardless of how large the
native scannergamut eventually would be? Any profiling on top of that
would then be limited to the size of s-RGB if we cant't get the RAW
pure as scanner RGB.

If so, I can understand the problem and the concept behind it, but I
would prefer to have an option (there may be a switch already?) to
choose pure RAW too. We're talking POSITIVES now. Is this s-RGB limit
a fact with Epson 4870 too.....? If so, I would hesitate to use that
4870 as a larger format filmscanner for positives( even if profiled
with thirdparty apps ) in combo with Vuescan.

I can accept tweakings of a RAW in Negative mode if that gives me a
better base as I can always switch over to *Positive* RAW for a pure
RAW of a negative, like the case with Silverfast and Nikonscanners.
But if Vuescan is using a fixed ratio for the CCD channels
indenpendently when using negmodeRAWs you can't go the other way
around with Epson and Vuescan. I assume that the analog gain feature
is for Nikons, not Epsons.

I've probably missed something along this road.

nikita

 

[Gerhard Fuernkranz]

"Yes, I had to do with with Epson scanners to get the color space
set to sRGB. Otherwise, I would have had to calibrate each Epson
scanner model separately.

However, it uses no correction when scanning negatives, since this
seems to separate the dye layers better than using sRGB."

---------

Actually that would mean that when positive scanning using an Epson,
the real scanner RGB is degraded to s-RGB regardless of how large the
native scannergamut eventually would be? Any profiling on top of that
would then be limited to the size of s-RGB if we cant't get the RAW
pure as scanner RGB.

Basically, that's how I understand Ed's statement as well.

But don't expect that the scanner really returns sRGB in this scan mode.
At least mine does not. The colors returned by my 1240U in this scan
mode are still far off sRGB (after adjusting gamma to 2.2). And I also
suspect, that even in this scan mode, different Epson models return
different colors.

However, (at least for my scanner, and for reflective scans) I don't
have evidence that this scan mode degrades scanning of the full gamut of
the IT8 medium significantly. When I characterize (profile) both scan
modes, then the two profiles fit the IT8 patches with the following error:

no color correction: peak = 2.974854, avg = 0.501386 (dE CIE94)
color correction CRT: peak = 3.076332, avg = 0.606608 (dE CIE94)

I.e. both give good numbers, and there is not too much difference. Such
numbers weren't possible, if the "CRT" scan mode would decrease the set
of distinguishable colors within the IT8 gamut significantly (but of
course, this may be different for a different Epson model, and I don't
have numbers for a slide target).

Btw, I do have a problem in general with the term "gamut of a scanner".
How should we define the "gamut of a scanner"? Most scanners can "see"
nearly any human visible object color (or actually, object reflectance),
and will return some RGB values for any scanned object. The actual
limitation in practice is rather "color-blindness", i.e. the inability
of the scanner to *distinguish* some object reflectances, which are seen
as different colors by the human vision (under a particular, desired
viewing illuminant).

Regards,
Gerhard