Vuescan questions

[vsbiting]

If "a real software company" buys out VS, it will take a huge effort for
them to figure out the ins and outs *before* they can come up with a
decent documentation. They will have to offer Ed a life time contract to
"revise" it every six months (weeks).

Why VS got such glowing reviews is beyond me.

Dan Rempel wrote
(in article

Don, I'm aware of how people feel about VS here. Hence I waited a long
time to try it out. I'll try to be objective with my comments.

[much snipping]

My reason for VS evaluation is to see if it can help in this regard. My
wishful thinking (extremely unlikely) is that the 5400 has a hidden hw
gamma control (separate from the hw exposure control) which VS can
tweak. Or VS can generate better scanner and film profiles to produce a
better gamma corrected raw scan (probably more likely).

My experience with VueScan is that the internals seem to work ok, but
the user interface and documentation are awful, and Ed Hamrick seems to
have a bit of a 'tude problem; could be wrong about the last one,
though. I haven't investigated gamma correction; perhaps someone who's
looked at it seriously can provide some hard data.

The UI simply sucks. For example, it's unpredictable: change something
here and another change occurs over there, which of course you won't
notice until you've scanned 20 or 30 images. And, Ed seems to make
small, arbitrary changes with each release, often leaving me to wonder
"what happened to X; did I somehow turn it off?" As for the
documentation, it's incomplete and outdated; nuff said.

Once you sort things out, though, it seems to generally do what it's
supposed to: generate useable profiles, do batch scannning and
conversion, perform somewhat useable OCR, etc. Whether or not that's
worth $90 US is up to you: I wish I'd saved my money, but now that I
have it I use it.

Sounds about right.

VueScan cries out for being bought out by a real software
company and overhauled. It supports (to varying degrees) a ton
of hardware, but the UI looks like it was designed by a
schizophrenic that lost his sight 4 years ago in 3 out of his 4
personalities.

It needs an "Apple makeover".

 

[vsbiting]

I changed the order of things a bit to tackle important stuff first.



OK, I see now! There are a number of issues here. If I go into too
much detail, please ignore and only focus on the relevant stuff.

If I understand correctly, you'd like to streamline your workflow and
avoid having to post-process in Photoshop afterwards. Is that right?

Not to avoid PS post-processing work, but to streamline it somewhat.
Creating and applying a better gamma profile will go a long way.

If so, editing in scanner software is not really a good idea, on
principle. There are a number of reasons for that. For example, basing
major editing decision on a tiny preview "keyhole" instead of 100% (or
magnified) full display an external editing program will provide.
Also, editing features of scanner programs are very limited because
their job is to scan, not to edit. Any editing is only provided as a
courtesy for people who want a "quick-and-dirty" scan and can't afford
to buy (or don't care for) an external editor.

I'm with you and have tracked the many raw scan discussions here. As a
raw scan practitioner for some time, my only issue seems to be the gamma
correction part in PS.

As to gamma correction, have you tried Timo's gamma curves?

http://www.aim-dtp.net/aim/download/gamma_maps.zip

His views are considered controversial by some but I found that his
gamma curves produced less banding than the Photoshop (version 6)
curves. (In the interest of full disclosure, I now don't use either
because my own scanner program uses full 16-bit look-up tables.)

Finally, what specific gamma problems are you referring to?

After applying Minolta's slide profile (i.e. gamma correction) to the
linear raw scans, they will appear quite decent in PS. But they still
look somewhat different from the Minolta scan previews with some color
shifts. The scans appear much "denser" in color and tone than the slides
themselves (but not very dark before applying the profile). It takes
quite a bit of PS work before the images can be printed without printer
profile oog problems. This is primarily the problem I'm trying to solve.

I find many of Timo's article insightful, e.g. how to set up a monitor
*before* calibrating. I'm going through his XLprofiler now. Is that
where the your zip reference come from?

One thing to note is that due to the way (most) programs generate
histograms they may look worse than the actual underlying data.
Specifically, the "recursive waves" i.e. spikes. The reason for that
is how the original 16-bit histogram data is scaled down to 8-bit.


I don't have a Minolta but, in theory, there's no hidden gamma control
because gamma is done is software.

I don't think that there is a hw gamma control either. Wish Minolta can
at least provide an *explicit* sw one.

One other thing, VueScan is particularly buggy with a Minolta (Mark 1)
i.e. the notorious "VueScan stripes". The author was unable to fix
this for over two years causing him to run away from here due to all
the Minolta users' complaints. He finally (allegedly) "fixed" it after
a knowledgeable contributor here explained how it's done in 5 mins.

Ed uses Nikons himself, and it is not surprising he would have problems
with other brands.

That's OK. There's nothing wrong with being subjective as such as long
as that's indicated. After all, each of our individual purchasing (or
any other) decisions are subjective in the end.

The only problem is when (some) people here try to project their
personal feelings and attempt to pass them off as objective "fact".


There's a qualitative difference between that and what VueScan does.

In case of reputable manufacturers, that "hiding" is a part of being
user friendly so as not to overwhelm the average user with technical
details, but (very important!) still allow advanced users full access.

Agreed. Not everyone wants to know the nuts and bolts.

In case of VueScan, we're talking about corruption of data beyond user
control in order to mask the program's failings e.g. taking raw
scanner data, mutilating it to hide the program's shortcomings, and
then passing that mess on as a "raw" scan.

But many native or third party digital sw do the same, although they
don't pass them on as "raw" anything.

There are basically two reasons for that. One is the "unusual thought
processes" of the author. As virtually everyone agrees (even the most
ardent VueScan supporters) the so-called user interface is a mess.

However, to the author it's perfectly "logical" and he insists that
the world follows him, in part because he doesn't have the integrity
or professionalism to fix a wrong concept from the ground up. Both of
those are clear signs of an amateur programmer.

Secondly, a confusing guide on top of a confusing so-called user
interface makes it easier for the author to hide behind that mess and
simply blame the users for his own failings.

That's truly deadly combination. Again, I wonder how VS users live with
it. Perhaps that's just how digital imaging is today.

Two for two!

I'm just trying to be logical!

 

[vsbiting]

/levels in the scanner software simply because

I completely agree with David, and my only regret is that it took me so
long to figure out that it's actually *faster* to do this in Photoshop
as PS is more consistent (I can have a canned levels or curves command
where I just manually tweak the white/black point for each image and
click OK).

I'm with both of you. When the histogram needs adjusting before
scanning, I don't use Minolta's sw to do that. Instead, I tweak the
composite rgb hw exposure control to either over- or under-expose during
the scanning. I find that adjusting the hw exposure control for each rgb
channel somewhat dicey. They cause color shifts.

When the histogram of a raw scan is unclipped, applying Minolta's
profile to gamma correct will keep the histogram to remain unclipped.

 

[Roger S.]

I was hoping that the VS film types will "help you get good color with
minimal work" also. But if they don't serve that purpose, or "just give
you bad color and more work correcting it", why did Ed make such a big
deal about them?

Does he make a big deal about them? I thought for the post part they
are quite dated and aren't routinely updated, vestiges of early Vuescan
versions. I've yet to find a use for them, but maybe others find them
"good enough."

 

[theo]

maybe others find them "good enough."

I'm another for "good enough" for an intended audience, which does NOT
care about gamut and histograms, but DOES care that the images carry the
story of the family history virtual scrapbook, WILL notice color casts.
An audience which will never desire to print MOMA-gallery-quality
movie-marquee enlargements of what originally were point-n-shoot even
brownie box camera photo frames. An audience which will be satisfied with
whatever consumer technology is current, NTSC to 1080i to whatever HD/DVD,
and won't care if it's VHS or BetaMax, BluRay or BluTooth or BluBayou.
In working with the proverbial shoe boxes of film strips and slides
predating WWII to 1998, films of domestic and European manufacture, all
(and with a lesser extent even Kodachromes of the '50s) exhibit some shift
or cast or fading. Clicking thru VS library and waiting for the preview
updates for eyeball comparisons is much quicker for me to get closer to
"good enough" starting point for less post-processing on TIFF'd images
from myriad legacy emulsions and formats.
As with any craftsman, my choice of tools and time in my own vocation is
directly related to the customers' willingness to pay for product and
service, whom depressed in the world of WalMarts, cheap enough is good
enough. Your means for paying the rent may differ.
Regards
Theo
=======
Pessimists remain morose precisely because they are so right too often.

 

[Don]

After applying Minolta's slide profile (i.e. gamma correction) to the
linear raw scans, they will appear quite decent in PS. But they still
look somewhat different from the Minolta scan previews with some color
shifts. The scans appear much "denser" in color and tone than the slides
themselves (but not very dark before applying the profile). It takes
quite a bit of PS work before the images can be printed without printer
profile oog problems. This is primarily the problem I'm trying to solve.

I think there may be several things here. I'm speculating in part
because I don't know all the details, but it appears that the Minolta
slide profile does more (or less?) than the Minolta preview. Also, a
preview and a full scan will differ, sometimes quite a bit.

I find many of Timo's article insightful, e.g. how to set up a monitor
*before* calibrating. I'm going through his XLprofiler now. Is that
where the your zip reference come from?

I stumbled upon his site very early on when I started scanning my film
library. It's very good reading and I personally don't find what he
writes controversial. It's just the facts. The question is how much
does editing in linear gamma differ from editing in 2.2 (or whatever).
I find that the differences are not sufficient to warrant changing my
whole system to linear gamma. But that's a personal preference which
does not negate any of his facts.

I don't think that there is a hw gamma control either. Wish Minolta can
at least provide an *explicit* sw one.

Oh, I think I see now what you mean! You're saying that Minolta SW is
hardcoded for gamma 2.2 (or whatever) and doesn't even offer you the
option of using linear gamma, right? That is bad!

Ed uses Nikons himself, and it is not surprising he would have problems
with other brands.

Well, at least two things about that. This was more than a problem.
It's was major bug which persisted for two years. During all this time
he shamelessly continued to claim Minolta as "supported" (when it
clearly wasn't!) and refused to refund the money to duped victims who
were justifiably angry. Finally, what he personally does or doesn't
use is really not relevant. Vuescan is (allegedly) a commercial
product and must be evaluated as such regardless of the author's
personal preferences.

But many native or third party digital sw do the same, although they
don't pass them on as "raw" anything.

Exactly! They don't pretend to be something they are not.

That's truly deadly combination. Again, I wonder how VS users live with
it. Perhaps that's just how digital imaging is today.

The way they live with it is to be in a permanent state of irritation
and with a hair trigger. So they often lash out at the most innocent
of statements and bite the heads off first time posters, etc. It's
known as a "having a huge chip on one's shoulder". ;o)

As to user guides, it's not just imaging but they're appalling across
the board. And have been for some time. The product cycles these days
are far too short so it's just not worth the effort of producing a
good manual. If anything, it would only delay the release and by then
the program would be obsolete.

I'm just trying to be logical!

And you hit the nail on the head! That's exactly right!

Don.

 

[Don]

If "a real software company" buys out VS, it will take a huge effort for
them to figure out the ins and outs *before* they can come up with a
decent documentation. They will have to offer Ed a life time contract to
"revise" it every six months (weeks).

There's no chance of that because a real software company would not be
wasting money on something as amateur as Vuescan.

The only way to fix Vuescan is to throw everything away and start from
scratch i.e. rewrite it from the ground up.

Why VS got such glowing reviews is beyond me.

Because ignorance is bliss and most of the so-called reviewers don't
really test the programs but just copy parts of the user guide.

I used to moonlight as a freelancer for several computer mags many
years ago and the editors had a favorite saying:

Don't let the facts get in the way of a good story.

I laughed, but the sad thing is they weren't kidding!

Don.

 

[Don]

There's no chance of that because a real software company would not be
wasting money on something as amateur as Vuescan.

And yet, oddly enough, VueScan seems to be recommended by the
most famous computer scientist in the world, Donald Knuth. It
just goes to show that just because someone writes the most
important books on computer science and writes TeX and
MetaFont, he's just an amateur and nowhere near as smart as me..

Because ignorance is bliss and most of the so-called reviewers don't
really test the programs but just copy parts of the user guide.

And of course Donald Knuth is also ignorant. Somehow, lots of other
professional photographers also use VueScan, but they just aren't
as intelligent as I am.

Don

 

[Don]

Of course, the problem with the Minolta 5400 was that some units had
low-quality CCD's, and these CCD's sometimes had some pixels that
were bad. Minolta's software filtered out the bad pixels, and Ed modified
VueScan to filter out the bad pixels.

I never owned a Minolta, and know very little about it, but I like to
talk about things I know nothing about.

Don

 

[Don]

VueScan is very buggy and unreliable so any testing you do will only
be valid for that particular version.

Of course, all the professional photographers who use VueScan are
also far less intelligent than me, and just don't have the wisdom and
experience that I do to see how much more poorly written VueScan.

Wait until I release the program I'm writing - then you'll all be amazed
at my brilliance and programming skills.

Don

 

[Don]

I don't have a Minolta but, in theory, there's no hidden gamma control
because gamma is done is software.

Of course, I know very little about scanners, and I like to talk about
things that I know very little about.

In case of VueScan, we're talking about corruption of data beyond user
control in order to mask the program's failings e.g. taking raw
scanner data, mutilating it to hide the program's shortcomings, and
then passing that mess on as a "raw" scan.

Of course, I'm just making this up - VueScan's raw files are actually
straight from the CCD, and don't have any corruption at all. Many
scientists use VueScan to make raw scans because you can't do this
with most vendor's software.

Don

 

[Don]

On Thu, 15 Jun 2006 13:06:06 GMT, (e-mail address removed) wrote:
In that case you should note that VueScan's matrix color profiling is
very primitive, not to mention buggy.

Vuescan does not create a LUT (look up table) like other profiling
packages do, but is relatively inexact requiring additional tweaking
in an image editor afterwards to mop up its shortcomings.

Of course, I'm misleading everyone here, because I'm actually quite
ignorant in this area. Matrix profiles work quite well, and are needed
when you use a scanner with a variable CCD exposure time.

Don

 

[Andre C]

I never owned a Minolta, and know very little about it, but I like to
talk about things I know nothing about.

That is like waving a red rag to a bull on this group isn't it.

 

[vsbiting]

Does he make a big deal about them? I thought for the post part they
are quite dated and aren't routinely updated, vestiges of early Vuescan
versions. I've yet to find a use for them, but maybe others find them
"good enough."

There are four pages of film types in the VS Guide. Different
generations of the some film types are listed, and Ed suggested to id
the generation by looking at "the writing on the film near the sprocket
holes". Such descriptions are certainly much more detailed than any
other feature descriptions in the Guide. A reader would assume that Ed
must have put in some great efforts on film types, and they play a
significant role in VS.

As someone noted, the Guide is not updated to reflect the changes of
each revision, and for good reasons (for Ed).

 

[Ralf R. Radermacher]

There are four pages of film types in the VS Guide. .... A reader would
assume that Ed must have put in some great efforts on film types, and
they play a significant role in VS.

He has "inherited" these profiles from Kodak, many moons ago. They were
meant for a totaly different platform of hard- and software. They don't
include anything introduced in the last five years. No Kodak Royal
Supra, no Elitecolor, none of their current colour print films.

On a more general note, profiles aren't of much use for colour print
film, anyway, since they'd have to be made for a particular combination
of film, exposure, processing (including the variations between
different manufacturers of chemicals etc. pp.), and hardware.

They're about as useful as colour analyzers in the olden days of RA-4
printing, i.e. they only work if they've been expensive enough that
you'll *want* to believe in their usefullness. ;-)

Ralf

 

[vsbiting]

This may be a very good reason for using the film types. But the VS
Guide did not spell it out as such. Did you stumble on how to use them,
or ???

Except for the masochists, we all want to only purchase and use the
tools that suit our needs and budgets. The best vendors would go out of
their way to achieve this goal for their customers, so would the best
tool reviewers here for those asking questions.

 

[vsbiting]

I think there may be several things here. I'm speculating in part
because I don't know all the details, but it appears that the Minolta
slide profile does more (or less?) than the Minolta preview. Also, a
preview and a full scan will differ, sometimes quite a bit.

Not sure.

I stumbled upon his site very early on when I started scanning my film
library. It's very good reading and I personally don't find what he
writes controversial. It's just the facts. The question is how much
does editing in linear gamma differ from editing in 2.2 (or whatever).
I find that the differences are not sufficient to warrant changing my
whole system to linear gamma. But that's a personal preference which
does not negate any of his facts.

I'm not technical enough to understand the significance between editing
in gamma 1 vs 2.2. But I do notice the bashing Timo took from those who
disagreed with him. Timo always stays with technical discussions, and
never throw a punch back. For that, he gets my respect. Digital imaging
technology, IMHO, is still in its infancy, kind of like before Darwin or
Galileo. It is not surprising for the current vendors and "experts" to
accuse others of heresy.

Oh, I think I see now what you mean! You're saying that Minolta SW is
hardcoded for gamma 2.2 (or whatever) and doesn't even offer you the
option of using linear gamma, right? That is bad!

Not quite. I think my raw scans before applying the profile is in linear
gamma. By an *explicit* gamma control, I mean that such a control will
offer options of 1 (and stated linear and hw), 2.2, etc. (sw). Without
such a control, I'm left to yank the curves/level on my own if I want
something other than a linear scan (gamma 1).

 

[vsbiting]

Careful readers would notice that we now have two Dons posting here. The
second fake Don started his postings on Wed. 10:40 AM.

Has Ed reincarnated and rejoined us as the second Don? Creative and in
character?

 

[Don]

Careful readers would notice that we now have two Dons posting here. The
second fake Don started his postings on Wed. 10:40 AM.

Has Ed reincarnated and rejoined us as the second Don? Creative and in
character?

Careful readers would notice that I have been posting
here for more than a year under several names. Take a
look at the posting history of "Golden" - he is also
a dial-up AOL user, just like I am. Take a look my
posting history in alt.test, using Don and Golden. How
many other names do you think I post under?

Don

 

[Don]

Careful readers would notice that we now have two Dons posting here. The
second fake Don started his postings on Wed. 10:40 AM.

Exactly! But - as history teaches - (rabid) Vuescan groupies are not
exactly known for reading. Let alone *careful* reading! ;o)

Has Ed reincarnated and rejoined us as the second Don? Creative and in
character?

Indeed! And not for the first time, either! The vocabulary sure fits
like a glove.

I didn't know he cared so much to go through all this trouble. It's
really kind of touching, in a way... My personal stalker.

All together now: Ahhh... How sweet! ;o)

Don.