Vuescan - new features

  • Thread starter Thread starter Robert Feinman
  • Start date Start date
Don, as a good scientist, I'm sure you realize that the quality of the
evidence is paramount. Not only do you have to identify good evidence
upon which to base your assertions, but you have to also examine the
quality of the evidence and defend it from question.

Or, as the case may require, you identify the source and clearly
indicate the context. After that it's up to the reader to accept or
reject any conclusions based on that source and context. They do that,
for example, by challenging either the source or the context, or both.

Notably (!), *nobody* has done any of that here (i.e. questioned
Bart's original statement).

Therefore, it follows, that both the original statement and context
have been accepted implicitly. And if they have done that, they can't
deny any logical conclusions which follow from that statement and
context, nor can they (as you now try) reject it all after the fact!

!=> Finally (and sadly!) the discussion wasn't even about any of that
(i.e. the substance). We couldn't get *up* to that point because the
"fans" couldn't even grasp the concept why measuring gamma 2.2 image
and applying the results to a gamma 1.0 image doesn't make any sense.

After I explained it with an example, their eyes glazed over and they
suddenly went quiet, except for one notable exception trying to change
the subject and ending up - as expected - by throwing insults after
they painted themselves in a corner.
If your evidence
fails to stand up to scrutiny you have to find other evidence or
abandon your argument. Simply citing "facts," particularly reports of
widely varying quality from users of various levels of knowledge, is
neither scientific nor objective, and it is quite irrational to then
claim that this is in any way scientific.

Again: Context! Best explained with an example. If the premise is:

Vuescan is notoriously buggy and unreliable

posting a long (and yet only partial!) Vuescan bug list from a
*variety* of sources (often with *multiple* confirmations!) is clear
evidence that the premise is not only true but justified.

The trouble is, in most cases, we are not dealing with logical
scientists here. Instead, such a *neutral* statement following from
evidence unleashes a torrent of "feelings". This is caused by some
Vuescan "fans" *misinterpreting* such an objective statement in a
number of different ways (e.g. as a subjective qualitative judgment).
Then based on that emotional overreaction they erupt with insults.
The tests I believe Bart have done of the 5400 comparing Minolta and
Vuescan software with different options enabled rise to the level of
credibility. You have yet to produce anything comparable to support
your assertions. I look forward to more tests about these new features
from people who use this software, rather from those who simply sprew
vitriol about it and resort to generalities and second-hand reports
when challenged.

As already indicated on several occasions that's a straw man for a
number of reasons. Here's two:

1. There is a vast amount of data out there already.
2. Anybody who disagrees with any of that data can do their own tests.

Why not tackle all of that first before asking for more from the one
source they do not even accept a priori!?

Therefore, any calls for "more tests" is merely grandstanding and
diversionary tactics, somewhat reminiscent of Bush vs climate! ;o)

!=> You will also note that when I did the tests I even went through
the trouble of posting an example where Vuescan did work (!!!) to show
both my impartiality and objectivity! It didn't (and doesn't) make any
difference! Some Vuescan "fans" just lash out anyway.

Not to mention that when I do post specifics some Vuescan "fans" -
instead of addressing the facts in those specifics - immediately
become abusive even though they themselves asked for it.

Damned if I do, damned if I don't...

Don.
 
Don said:
This angry (and unprovoked!) outburst full of insults without any
facts whatsoever illustrates exactly what I'm talking about.

Don, I was just mimicking your style of writing - no reason to be upset;-)
You may notice (although given the above I doubt it) that I have never
insulted you. That's because for me there is no "war". Just facts.

No, there's no war between you and me - there is a one-directional war
that you seem to be fighting against VueScan. Your 'facts' just
demonstrate that quite a number of shortcomings in VueScan have been
reported in this group. For me that's not enough to make a valid
comparison between VueScan and other scanning software, even if there
would be no complaints about any other software in this newsgroup.

For instance, I never suggested that Photoshop and VueScan can be
compared. I suggested that even if you compare Photoshop to another
imaging package and even if you have newsgroup-based 'bug reports' on
both, you cannot prove that one of the two is inferior to the other
based on that.
You are trying to prove that VueScan is inferior to its 'competitors'
based on VueScan bug reports from one source. That's not going to
convince me and it won't stop me using VueScan. There are things I won't
use VueScan for, however, because I personally think there are better
alternatives. One of these features is the curves adjustment - which
might be useful for others- and another one is printer profiling, which
doesn't seem to work (scanner profiling OTOH works very well in my
experience and it's a very useful feature).
 
Don, I was just mimicking your style of writing - no reason to be upset;-)

Don't you realize how insulting that statement is!?

If that's what you think of my "style" you appear to understand even
less of what is written here than I thought.

Your confusion about gamma at the beginning (which both Phillip and I
had to correct) confirms this as does earlier confusion about scanner
orientation as do all the contradictions in this message alone, etc.
No, there's no war between you and me - there is a one-directional war
that you seem to be fighting against VueScan.

That's what you think because of all your misunderstandings. There is
no "war" against Vuescan. Just facts.
For instance, I never suggested that Photoshop and VueScan can be
compared.

Yes, you did. You just made a direct comparison with Photoshop:

If you follow the link http://tinyurl.com/7eakq you will find a similar
"bug list" with (currently) 417 items (including some redundancy) from
"reliable sources" concerning Adobe Photoshop.


Anyway, in light of such massive confusion which you're repeatedly
demonstrating (as outlined in the all of the above items) this is
becoming pointless.

Don.
 
Don said:
this is
becoming pointless.

Indeed. Continuing this "discussion" doesn't make sense. Your
misinterpretations of what I stated before are only becoming worse.
Every time I react to what you say, you are twisting my words and the
words of others.
 
Indeed. Continuing this "discussion" doesn't make sense. Your
misinterpretations of what I stated before are only becoming worse.
Every time I react to what you say, you are twisting my words and the
words of others.
Do not feed the trolls. Use a little self control (please!)
 
Don ([email protected]) wrote in
However, Photoshop, Windows et al are massively more *complex* than
Vuescan! (A hint: Check file sizes, for starters!)

File size and complexity have little, if anything, to do with each
other. "Complexity" is about both what functionality is there AND how
that functionality is interrelated and interdependent; "file size" is
the result of number of functions (which may be entirely independent or
even redundant) and the efficiency of how that functionality is
translated to machine code (which itself is dependant on programming
language and compiler/assembler used). A small program cleverly written
in assembly language can be far more complex than a much larger program
written in a "higher" but far more inefficient language and compiled
with a not-very-optimizing compiler (and what an optimizing compiler can
do is again partially dependent on teh skill of the programmer).

You *may* be able to use file size as a clue (but no more than that)
given that two programs you are comparing are:
1) written in the same programming language
2) written by the same (team of) programmer(s)
3) written in a time frame that does not make much difference to the
experience of that (team of) programmer(s)
4) using available libraries in the same way and to the same extent
5) compiled with the same compiler/assembler with the same settings

(and I've probably forgotten to mention a few)

File size by itself means nothing at all.
 
I was trying to find the point in the thread where we talked about the
latest version of Vuescan, rather than a general meta-discussion. I
hope this will do.

I downloaded 8.3.16 and was not too impressed.

Have others had problems with the IR cleaning? There was a new and
interesting artifact I'd never seen before with my Canon FS4000US which
looked like a bunch of multicolor circles landed on part of my image.
Novel and yet disturbing. Saving the non-cleaned file confirmed this
was an IR artifact. Have others seen this? One caveat- I was using
the scanhancer to scan negative film (not that I'd ever had this
problem before...) I can post examples if anyone cares to see.

Also, changing from IR cleaning none to light after you preview makes
the preview rotate 90 deg and changes where the cropping guides are.
It actually scans correctly when you hit scan, but this is still
annoying.

After this I went back to 8.3.01 which works except that it chopps off
5% off of mounted slides if you change the cropping. I no longer get
my slides mounted so this isn't an issue. Better the devil you know
(and have debugged).
 
Don ([email protected]) wrote in


File size and complexity have little, if anything, to do with each
other. "Complexity" is about both what functionality is there AND how
that functionality is interrelated and interdependent; "file size" is
the result of number of functions (which may be entirely independent or
even redundant) and the efficiency of how that functionality is
translated to machine code (which itself is dependant on programming
language and compiler/assembler used). A small program cleverly written
in assembly language can be far more complex than a much larger program
written in a "higher" but far more inefficient language and compiled
with a not-very-optimizing compiler (and what an optimizing compiler can
do is again partially dependent on teh skill of the programmer).

You *may* be able to use file size as a clue (but no more than that)
given that two programs you are comparing are:
1) written in the same programming language
2) written by the same (team of) programmer(s)
3) written in a time frame that does not make much difference to the
experience of that (team of) programmer(s)
4) using available libraries in the same way and to the same extent
5) compiled with the same compiler/assembler with the same settings

Numbers 4 and 5 are the biggies when it comes to Windows programming
and DLLs. It takes a lot of work and knowledge to use only the
necessary DLLs and to do as much work as possible with the ones that
need to be included.

The DLLs alone can make as much as a 10:1 difference in size.
Using optimization and the best settings can make an even bigger
difference.

Roger Halstead (K8RI & ARRL life member)
(N833R, S# CD-2 Worlds oldest Debonair)
www.rogerhalstead.com
 
"Complexity" is about both what functionality is there AND how
that functionality is interrelated and interdependent

You know that and I know that. But we are programmers.

The trouble is I'm trying to convey this to someone who 1. apparently
is not a programmer and 2. has problems with fundamental logical
causality (e.g. can't even understand why sampling a gamma 2.2 image
and applying the result to a gamma 1.0 image doesn't make sense).

Therefore, digressing into such arcane programming minutiae is just
not appropriate given the target audience or pertinent to the subject
matter i.e. attempting to excuse Vuescan bugs by (mis)using Photoshop.

To avoid any misunderstanding you should also make it clear you're not
challenging that core statement but tangentially addressing a totally
different issue of interest only to us pedantic hardcore programmers.

In other words, you are not challenging the key point that Photoshop
is vastly more complex than simple Vuescan which is why attempting to
use Photoshop as an excuse for amateur Vuescan bugs is just silly.
You *may* be able to use file size as a clue (but no more than that)

Given the above context that's exactly what I was doing only I called
it a "hint" and "for starters".
File size by itself means nothing at all.

"Nothing" is going too far. At the very least, you could call it
"circumstantial evidence" or "indirect indication" but it's certainly
not nothing.

Even if that size is due to simple data (not even code!) the mere fact
that there is more of it makes it prone to more errors, even if it's
something as mundane as media errors, not to mention all of the other
things that can go wrong as size increases.

A simple rule of thumb is: Complex machines brake more easily,
regardless of the cause of this complexity. And size by itself (!) is
a (one) cause of this complexity.

Don.
 
True enough.

(Now why did the CPU fan in my computer break? It looks simple enough.
Never mind...)

This is where the second rule of thumb comes in ;o)
A chain is only as strong as its weakest link.

It's quite common for expensive things to brake down because
manufacturers tried to save a few cents on some tiny component.

The fan broke because they don't make them to the same standards as
the CPU it's supposed to be cooling. If they did, the fan would last
hundreds of years and probably cost more than the whole computer.

Case in point, the (in)famous US military toilets which were $1000 a
piece. Why? Because the specs were it had to withstand a nuclear blast
at a certain distance!

Granted, when a nuke goes off a lot of soldiers would need a toilet
badly but I don't think many would live to enjoy a nuke-proof one. ;o)

Don.
 
1) using Scanhancer on Nikon with Vuescan there are no artifacts
2) using Light IR doesn't affect preview...nothing changes except dust
4) Light IR is virtually perfectly effective except for handling oils
from fingers. Medium IR doesn't seem more effective. Heavy IR MAY
degrade detail, but not perceptably in small (eg 35 @ 11") prints.

Scanhancer DOES hinder film advance if one uses Nikonscan with the
motorized film advance, which positions film on the basis of
framelines. Not a problem for Vuescan, which doesn't operate the
motorized advance optically, rather it simply measures distances.
Vuescan functions properly but isn't a precise in automatically
positioning frames (needs regular tweaking) due to varying frame lines
with different cameras. Scanhancer seems not to cause other problems
using Nikon V with Vuescan.

Roger, I suspect your problems are similarly unique to Canon scanner
and/or Mac.
 
Roger,

I have not seen this artifact when scanning with my FS4000, but then
again I haven't ventured past 8.3.07. For reference, I think 8.3.06 is
the one which "fixed" my IR cleaning, so maybe that's where it "broke"
yours! Anyway, I would still be curious to see an example of the
artifact you're referring to, if you don't mind posting it.

Jeff
 
I see a grain situation, not dust & scratches to be remedied by IR cleaning.

Maris
 
Hi Maris,
It is 800 speed consumer film (800 MAX)- I don't use this film anymore
for many reasons, but this negative is a few years old. The scanhancer
took care of scratches elsewhere in the frame. Vuescan was supposed to
help with the rest.

The first file I posted shows what an odd job the IR cleaning did,
especially since there was little that needed to be fixed in this
particular crop. The second post shows what the negative actually
looks like. I used NoiseNinja when I prepared this roll of film for
output.

Djon,
I use WinXP home, not a Mac, but conceed that these VS issues are
likely FS4000US specific. Lucky me?
 
Roger,

That's just bizarre! I've never seen that kind of artifact before. I
think it's pretty clear that it has something to do with the IR
cleaning though. Exactly what, I'm not sure. Have you repeated the
scan to see if the effect is repeatable? Maybe try it without the
scanhancer (assuming it's not too hard to remove)? It might also be
instructive to have a look at the IR image to see if the IR scan is
really "seeing" something on the film at the location of the oddity.
You can do this (if you don't already know) by just selecting to view
IR from the color tab (COLOR:View color: infrared), instead of RGB.

When I get a chance, I'll download the latest VueScan version and see
if I see the same kind of effects.

Jeff
 
I have to update my own reply because I've just downloaded and tried
out version 8.3.17. And yep - the IR cleaning on my FS4000 is showing
something similar to what you were seeing, Roger. The negatives I was
using were pretty clean so I had to look awhile, but I was able to
identify something similar on a couple frames. It looks like it
happens with some of the largest dust particles. Smaller ones seem to
get the usual cleaning. Just for fun, I tried turning up the IR
cleaning to medium and then heavy. But the artifact doesn't go away
when you do this. If anything, MORE uncleaned particles started
popping up. Very strange. Another thing I noticed - and this may or
may not have anything to do with the problem - the IR view of the first
three frames looked different than the fourth (I was scanning strips of
4). The first three show a monochrome version of the scene on the
frame, while the fourth shows up like it used to - basically the image
is all white, with the dust showing up as dark spots, but no hint of
the image itself.

Jeff
 
Hi Jeff,
Thanks for confirming this.
Can you send Ed a raw scan that shows the defects (under 2 megs or post
it somewhere). I don't have time to test it out right now.
 
If you haven't checked lately you might want to look at
the latest Vuescan version. Ed has moved the graphs to a
panel under the options so that you can view the graph
and the image simultaneously.
He has also added a gamma curve adjustment with two
control points so you can do S-shaped adjustments.

Despite being a real "Vuescn Veteran," I fail to discover
a proper workflow with this "new user interface."

Even on a 3GHz processor movement of the sliders on the
new tiny diagrams is jerky and I did what I did so often
previously already: I reverted to an older well proven
version 8.1.30.

This new user interface is an enigma to me.

Thomas
 
Back
Top