VueSCAM!

[Don]

I have seen something similar to that once, but I had something set up
badly wrong - unfortunately I can't remember what caused it now. If I
recall what it was I'll let you know.

OK, thanks, Kennedy. It does sound quite serious, though. As usual,
all I did was "turn everything off".

In the meantime I'll just rewrite my code to do a 16-bit preview.
Nevertheless, it will bug me until I find out what's wrong so lots of
testing ahead... Yet another unwelcome distraction... :-(

I just seem to be such a bug magnet! Murphy's corollary: Murphy's
first name is Don. ;o)

Don.

 

[Pavel Dvorak]

Don ([email protected]) writes:

[nuked]

I just seem to be such a bug magnet! Murphy's corollary: Murphy's
first name is Don. ;o)

Don.

You mean Don Murphy Lincoln-Mercury car dealerhsip in Ottawa? They seem
to be doing OK; a few years ago they even received an award for
environmentally friendly outdoor lighting.

(Sorry about posting something entirely outside the topic, but I just could
not resist.)

Pavel

 

[Don]

The "reversed" 8-bit histograms mystery has been solved!

An important correction first: The program that exhibits this problem
is the supplied example code (*not* NikonScan). Since my program is
based on same principles as the example it suffers from it too. But
NikonScan is fine.

BTW, I also noticed that the TWAIN version supplied with the SDK is
actually older than the version that came with my scanner. My program
displays this in the Diagnostics which is how it caught my attention.
Even though Nikon TWAIN doesn't follow the naming convention it can be
examined by looking at the properties.

Anyway, the problem was gamma. The sample program does not explicitly
initialize it. Therefore, it appears, either the TWAIN driver or the
scanner was in an "unknown state" which caused these "reversed"
histograms. Once the gamma is explicitly initialized the histogram
comes out fine.

Actually, it comes out too good! Namely, even in 8-bit there is no
"comb" appearance on either end!?

Curiously, there is no gamma setting as such but the client program
must supply a lookup table (LUT) to the driver. Although this is more
work for the client it also means more flexibility. For example,
instead of the regular "straight" gamma, the client could also
implement other methods e.g. Timo's "curved" gamma. Indeed, one could
supply any LUT! This makes implementing curves elementary.

Anyway, this is all hot off the presses so lots of things to play with
in the coming hours...

Don.

 

[Don]

Don ([email protected]) writes:

[nuked]

I just seem to be such a bug magnet! Murphy's corollary: Murphy's
first name is Don. ;o)

Don.

You mean Don Murphy Lincoln-Mercury car dealerhsip in Ottawa? They seem
to be doing OK; a few years ago they even received an award for
environmentally friendly outdoor lighting.

LOL! And a fellow Canadian, too!

(Sorry about posting something entirely outside the topic, but I just could
not resist.)

No, not at all. Humor is always welcome as we do have a tendency here
to take ourselves too seriously at times.

Don.

 

[Don]

I have seen something similar to that once, but I had something set up
badly wrong - unfortunately I can't remember what caused it now. If I
recall what it was I'll let you know.

Anyway, the problem was gamma. The sample program does not explicitly
initialize it. Therefore, it appears, either the TWAIN driver or the
scanner was in an "unknown state" which caused these "reversed"
histograms. Once the gamma is explicitly initialized the histogram
comes out fine.

Actually, it comes out too good! Namely, even in 8-bit there is no
"comb" appearance on either end!?

The plot thickens!

The scanner/TWAIN are *not* in an "unknown state". They are simply in
gamma 1.0!! (I retrieved the LUTs. They are 16-bit, BTW!)

Even NikonScan 4, 0, 2, 3000 running TWAIN (LS5000.MD3) 1, 0, 0, 3013
will produce a "right-handed comb histogram" when set to gamma 1.0!?

At gamma 2.2 the histogram is fine.

Anyway, does that ring any bells as to what the cause could be? Also,
you're running NikonScan 4.x as well, right? If so, could you try and
reproduce this on your scanner and TWAIN? I have AE on, but everything
else is off, bit depth is 8-bits.

Don.

 

[Kennedy McEwen]

The plot thickens!

The scanner/TWAIN are *not* in an "unknown state". They are simply in
gamma 1.0!! (I retrieved the LUTs. They are 16-bit, BTW!)

Even NikonScan 4, 0, 2, 3000 running TWAIN (LS5000.MD3) 1, 0, 0, 3013
will produce a "right-handed comb histogram" when set to gamma 1.0!?

At gamma 2.2 the histogram is fine.

Anyway, does that ring any bells as to what the cause could be? Also,
you're running NikonScan 4.x as well, right? If so, could you try and
reproduce this on your scanner and TWAIN? I have AE on, but everything
else is off, bit depth is 8-bits.

Well, I have tried this with NCM on, NCM off, gamma=1, gamma=2.2,
gamma=0.45 and every other combination I can think of but can't get
Nikonscan to reproduce it. Even tried changing between NCM on and off
before saving scans which the manual specifically warns can cause
problems - no luck, everything works perfectly! ;-)

However, I can force something pretty close if I scan with gamma=2.2 in
8 bits and THEN apply a gamma adjustment of 0.45 (which should cancel
out to gamma=1.0) in Photoshop afterwards. Not entirely unexpected
doing that, of course. However it might lend a clue as to what is going
on here.

It would appear that something in Nikonscan is locked at applying a
gamma of 2.2, to produce the original raw data truncated to 8-bits. This
is then converted by applying the inverse of gamma 2.2 (ie.
approximately 0.45) to get the final image that TWAIN imports, complete
with top end combing in the highlights.

BTW, although this shows in the histogram clearly, I bet you struggle to
see the highlight posterisation in the image itself. This proves that
8-bit data is more than adequate in the perception domain.

 

[Don]

Well, I have tried this with NCM on, NCM off, gamma=1, gamma=2.2,
gamma=0.45 and every other combination I can think of but can't get
Nikonscan to reproduce it. Even tried changing between NCM on and off
before saving scans which the manual specifically warns can cause
problems - no luck, everything works perfectly! ;-)

Darn! ;o) But I suspected as much.

However, I can force something pretty close if I scan with gamma=2.2 in
8 bits and THEN apply a gamma adjustment of 0.45 (which should cancel
out to gamma=1.0) in Photoshop afterwards. Not entirely unexpected
doing that, of course. However it might lend a clue as to what is going
on here.

The culprit seems to be gamma but that's as far as I got. Strangely,
8-bit scans with gamma 2.2 don't exhibit any comb effect?

It would appear that something in Nikonscan is locked at applying a
gamma of 2.2, to produce the original raw data truncated to 8-bits. This
is then converted by applying the inverse of gamma 2.2 (ie.
approximately 0.45) to get the final image that TWAIN imports, complete
with top end combing in the highlights.

The only thing I can think of (and it's a stretch) is that perhaps in
this TWAIN version Nikon did - whatever they did - intentionally (!?)
to make 8-bit histograms at gamma 2.2 "smoother". I know, it's crazy,
but I just can't think of anything else.

Next, I'll try a few 16-bit tests and see where that gets me...

BTW, although this shows in the histogram clearly, I bet you struggle to
see the highlight posterisation in the image itself. This proves that
8-bit data is more than adequate in the perception domain.

Very true. But this is for archiving.

BTW, I used a perfectly exposed *non*-Kodachrome slide for these tests
which is why I didn't even notice it was in gamma 1.0. After all those
dark Kodachromes at 2.2. this regular slide at gamma 1.0 scan looked
just fine. I need to have my eyes recalibrated! ;o)

Don.

 

[Sansame]

And I'm glad you're happy with VueScan. I have often mentioned VueScan
to people who just want a quick-and-dirty job. But they should not be
under the misapprehension that it's anything else.

Don.

I am almost happy to be classified in your "quick-and-dirty" category
because I am happy with Vuescan. This is probably à "fact" as you say.
But you didnt answer my question. Where do you find so much time and
energy to fight against Vuescan ? Another question could be why ?
Sansame

 

[Don]

I am almost happy to be classified in your "quick-and-dirty" category
because I am happy with Vuescan. This is probably à "fact" as you say.

Indeed! And I'm happy for you.

The problem arises only when people with similar *minimal*
requirements try to misrepresent such low threshold as "quality". Like
the case of another happy VueScan user pointing to a tiny web graphic
as "proof" that VueScan "works" with Minolta.

But as long you admit that you're after a quick-and-dirty scan and
VueScan satisfies your low requirements, more power to you!

But you didnt answer my question. Where do you find so much time and
energy to fight against Vuescan ?

There's nothing to answer because you seem to have selective memory.
If you check the archives you'll see message on many subjects.

So the real question is: How come you didn't ask me where I get the
time to write those messages? Why the obsession with VueScan messages
only?

Another question could be why ?

Because misguided VueScan users demand answers - just like you did!

So, on the one hand, you demand I respond to your VueScan message, but
when I do, you accuse me of writing too many VueScan messages...!?

Can you explain that "logic"?

Don.

 

[MPA]

Rumor has it that LaserSoft does not allow SilverFast

Pure speculation. For that matter, you don't find too many people
posting in this group from the drum scanning newsgroups either.

i had huge troubles with coolscan II(ls-20) very old scsi scanner. a lot
of stripes. image was not sharp. i cleaned the mirrors. later i could
manage to get rid of the stripes and got sharp images. i still dont know
what was wrong. i will report when there are teh same problems again.
what i would like to see ia a detailed manual, explaining all the features.
unfortunately i cannot use vuescan-which i like very much- with umax
astra 4000U(a scsi-usb-variant).

michael przewrocki

michael przewrocki

 

[Scooterlady]

(QUOTE=If the tone of your emails were the same as this post .... etc etc)

Finding this site was completely accidental & had nothing to do w/what I was originally searching for. However, I like to read thru pc forums so I skimmed thru this one.

I know it's from SEVERAL yrs ago, but I need to say "something"......

The original poster was justifiably very upset. I guess when he found this site, he probably thought he found a place where he cld vent his frustrations (been there - done that myself in forums).

I saw nothing wrong &/or offensive w/what he wrote. Although there might be some validity to judgmental responses, I think the accusations were very unfair.

I don't suggest that Joe or anyone for that matter, make a habit of venting in forums (altho if one needs to do so, maybe a warning & explanation 1st wld lessen the sound of his/hers venting), but don't be so quick to judge. Maybe u handled it differently at some time, or wld have, but I understand how/why Joe posted.