VueScan feature request: multi-pass scan from disk

VueScan feature request: multi-pass scan from disk

Post by Nicole Vincen » Mon, 09 Sep 2002 19:19:06



Hi

I've just emailed Ed Hamrick asking whether he'd be able to add the
following feature to VueScan:  make it possible for VueScan to perform a
multi-pass scan when scanning from files stored on the hard disk.

What exactly do I mean?  Well basically I'd like to be able to have
several images in the usual scan0001.tif scan0002.tif etc format in the
source directory, and to tell VueScan how many images to scan through
when I press the "scan" button, and to integrate those separate images
into one image, just like when you tell VueScan to do a 4X multi-pass
scan to get rid of noise.

Why would this be useful?  I've just got myself a digicam, and I'd like
to be able to take longer exposure shots than the current limit of 15
seconds.  This would be possible if I could take 4 or more 15-second
shots from the same position, put them into the VueScan source
directory, and set VueScan going.  Apart from reducing CCD noise in the
eventual images, it would also make it possible to get brighter images
without amplifying noise.

Another use could also be to fake multiple exposure shots.  Rather than
cranking the shutter whilst retaining the film in its current position
and taking a second exposure, and hoping that the two images were
aligned, this could also allow (with a couple of extra settings) to
integrate the two images into one image seamlessly from our favourite
scanning application VueScan  :-)

Given that most of the functionality is already there in VueScan, why
not enable multi-pass scanning for "scan from disk" operations?

Would anybody else find this feature useful?

Cheers

Nicole

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Bart van der Wol » Mon, 09 Sep 2002 21:38:29



SNIP

Quote:> Given that most of the functionality is already there in VueScan, why
> not enable multi-pass scanning for "scan from disk" operations?

> Would anybody else find this feature useful?

Despite the fact that there is nothing wrong with the concept, the results
will be haphazard in practice. The problem is with registration. It could
only work when the images are in perfect pixel alignment (needs morphing).
Since I assume most users use VueScan for scanning film and reflection
copies, alignment of different images is everything but guaranteed. Even
multiple scan passes of the same image often lack perfect alignment, causing
irregular blur.

Bart

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Nicole Vincen » Mon, 09 Sep 2002 22:48:14


Bart

I'm not talking about holding the camera in my hand, but putting it on a
tripod and using my infra red remote control each time I want to set off
another frame.

So registration is definitely not going to be a problem.

Besides, there is another product on the market which does precisely
this, but given that all the functionality is already there in VueScan,
why not use it.  And since others have done it without problems, I'm
sure that Ed would do a fabulous job too.

Cheers

Nicole




> SNIP
> > Given that most of the functionality is already there in VueScan, why
> > not enable multi-pass scanning for "scan from disk" operations?

> > Would anybody else find this feature useful?

> Despite the fact that there is nothing wrong with the concept, the results
> will be haphazard in practice. The problem is with registration. It could
> only work when the images are in perfect pixel alignment (needs morphing).
> Since I assume most users use VueScan for scanning film and reflection
> copies, alignment of different images is everything but guaranteed. Even
> multiple scan passes of the same image often lack perfect alignment, causing
> irregular blur.

> Bart

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Ronald Bruc » Tue, 10 Sep 2002 01:00:45


[[ This message was both posted and mailed: see
   the "To," "Cc," and "Newsgroups" headers for details. ]]

Would this really work?  I assume your images are VERY dim
(telescope?).  I can easily foresee the existence of a threshold
phenomenon, where the number of photons striking the film is linear in
time, but the film response requires a certain number of seconds to
"set up", hence is SUBlinear.  Thus four 15-second exposures might not
be the same as a 60-second exposure.  Any astronomers who could comment
on this?

And how do you propose to keep the photos in registration?  Especially
if they're astronomical observations from a rotating earth.  I suppose
the telescope would take care of most of this...

--Ron Bruck



> Hi

> I've just emailed Ed Hamrick asking whether he'd be able to add the
> following feature to VueScan:  make it possible for VueScan to perform a
> multi-pass scan when scanning from files stored on the hard disk.

> What exactly do I mean?  Well basically I'd like to be able to have
> several images in the usual scan0001.tif scan0002.tif etc format in the
> source directory, and to tell VueScan how many images to scan through
> when I press the "scan" button, and to integrate those separate images
> into one image, just like when you tell VueScan to do a 4X multi-pass
> scan to get rid of noise.

> Why would this be useful?  I've just got myself a digicam, and I'd like
> to be able to take longer exposure shots than the current limit of 15
> seconds.  This would be possible if I could take 4 or more 15-second
> shots from the same position, put them into the VueScan source
> directory, and set VueScan going.  Apart from reducing CCD noise in the
> eventual images, it would also make it possible to get brighter images
> without amplifying noise.

> Another use could also be to fake multiple exposure shots.  Rather than
> cranking the shutter whilst retaining the film in its current position
> and taking a second exposure, and hoping that the two images were
> aligned, this could also allow (with a couple of extra settings) to
> integrate the two images into one image seamlessly from our favourite
> scanning application VueScan  :-)

> Given that most of the functionality is already there in VueScan, why
> not enable multi-pass scanning for "scan from disk" operations?

> Would anybody else find this feature useful?

> Cheers

> Nicole

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Nicole Vincen » Tue, 10 Sep 2002 01:42:25


Ron

I intend to use successive shots of the same scene to generate a
composite image.  The scene may be of star trails for example, or of a
moonlit scene.  In either case, the least I can expect is for the
electrical noise to be cancelled out as long as the successive shots are
not out of register (otherwise blurring will occur).  Other astronomical
applications seem possible but they may be limited by the problem you
point to.

How do I intend to keep the successive shots in registration?  By
keeping my camera on a tripod and using an infra-red remote control to
set of consecutive shots.  It works beautifully, and all I need is an
automatic way of combining them, and VueScan's multi-pass scanning would
do precisely what I want.

As to the issue of certain detail simply not registering unless the
film/ccd-plain is exposed for a certain period of time, I really can't
comment on how this would work.  I expect that we should be able to get
_some_ improvement, but not necessarily as much as we could if we could
do a single really long exposure.  Indeed it would be good to get some
feedback from someone who knows what they are talking about.  :-)

But like I said in a responce to another person's question, it's not
like this would be the first time this kind of feature would be
implemented in a computer program for precisely this sort of function.
Others have done it too and it seems to work well in their programs, but
since I already have VueScan and I know how to use it as a workhorse,
I'd rather have keep using it rather than dish out more money on someone
else's program.

Cheers

Nicole


> [[ This message was both posted and mailed: see
>    the "To," "Cc," and "Newsgroups" headers for details. ]]

> Would this really work?  I assume your images are VERY dim
> (telescope?).  I can easily foresee the existence of a threshold
> phenomenon, where the number of photons striking the film is linear in
> time, but the film response requires a certain number of seconds to
> "set up", hence is SUBlinear.  Thus four 15-second exposures might not
> be the same as a 60-second exposure.  Any astronomers who could comment
> on this?

> And how do you propose to keep the photos in registration?  Especially
> if they're astronomical observations from a rotating earth.  I suppose
> the telescope would take care of most of this...

> --Ron Bruck



> > Hi

> > I've just emailed Ed Hamrick asking whether he'd be able to add the
> > following feature to VueScan:  make it possible for VueScan to perform a
> > multi-pass scan when scanning from files stored on the hard disk.

> > What exactly do I mean?  Well basically I'd like to be able to have
> > several images in the usual scan0001.tif scan0002.tif etc format in the
> > source directory, and to tell VueScan how many images to scan through
> > when I press the "scan" button, and to integrate those separate images
> > into one image, just like when you tell VueScan to do a 4X multi-pass
> > scan to get rid of noise.

> > Why would this be useful?  I've just got myself a digicam, and I'd like
> > to be able to take longer exposure shots than the current limit of 15
> > seconds.  This would be possible if I could take 4 or more 15-second
> > shots from the same position, put them into the VueScan source
> > directory, and set VueScan going.  Apart from reducing CCD noise in the
> > eventual images, it would also make it possible to get brighter images
> > without amplifying noise.

> > Another use could also be to fake multiple exposure shots.  Rather than
> > cranking the shutter whilst retaining the film in its current position
> > and taking a second exposure, and hoping that the two images were
> > aligned, this could also allow (with a couple of extra settings) to
> > integrate the two images into one image seamlessly from our favourite
> > scanning application VueScan  :-)

> > Given that most of the functionality is already there in VueScan, why
> > not enable multi-pass scanning for "scan from disk" operations?

> > Would anybody else find this feature useful?

> > Cheers

> > Nicole

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Bart van der Wol » Tue, 10 Sep 2002 01:39:10



SNIP

Quote:> I'm not talking about holding the camera in my hand, but putting it on a
> tripod and using my infra red remote control each time I want to set off
> another frame.

SNIP

With a digicam that would work, not with film. So I assume you use a
digicam.

Bart

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Ed Hamric » Tue, 10 Sep 2002 02:25:37



> I intend to use successive shots of the same scene to generate a
> composite image.  The scene may be of star trails for example, or of a
> moonlit scene.  In either case, the least I can expect is for the
> electrical noise to be cancelled out as long as the successive shots are
> not out of register (otherwise blurring will occur).  Other astronomical
> applications seem possible but they may be limited by the problem you
> point to.

I can see how this might be useful, but I can also see
that a grand total of only a few people will ever use this
feature if I add it to VueScan.  Given that there's already
another application available for doing this, I can't see
any reason to add it to VueScan.

Regards,
Ed Hamrick



spam bit-bucket (unless it has "Vue" in the subject).

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Martin Underwoo » Tue, 10 Sep 2002 02:30:32



Quote:

> Ron

> I intend to use successive shots of the same scene to generate a
> composite image.  The scene may be of star trails for example, or of a
> moonlit scene.  In either case, the least I can expect is for the
> electrical noise to be cancelled out as long as the successive shots are
> not out of register (otherwise blurring will occur).  Other astronomical
> applications seem possible but they may be limited by the problem you
> point to.

> How do I intend to keep the successive shots in registration?  By
> keeping my camera on a tripod and using an infra-red remote control to
> set of consecutive shots.  It works beautifully, and all I need is an
> automatic way of combining them, and VueScan's multi-pass scanning would
> do precisely what I want.

> As to the issue of certain detail simply not registering unless the
> film/ccd-plain is exposed for a certain period of time, I really can't
> comment on how this would work.  I expect that we should be able to get
> _some_ improvement, but not necessarily as much as we could if we could
> do a single really long exposure.  Indeed it would be good to get some
> feedback from someone who knows what they are talking about.  :-)

> But like I said in a responce to another person's question, it's not
> like this would be the first time this kind of feature would be
> implemented in a computer program for precisely this sort of function.
> Others have done it too and it seems to work well in their programs, but
> since I already have VueScan and I know how to use it as a workhorse,
> I'd rather have keep using it rather than dish out more money on someone
> else's program.

For shooting night scenes even under bright streetlight, I've often used
exposures (on 200 ASA or 400 ASA film) of much longer than 15 seconds - by
the time you've put on a blue filter to correct for the tungsten light and
maybe a polarising filter to remove excess shine of a wet road etc, you're
talking about exposures of maybe 60 seconds. Also, in a crowded street with
people walking past all the time, it's often beneficial to deliberately use
a long shutter speed (stopping down if necessary to limit the light) so that
the people will be recorded as a blur rather than as ghostly images.

For combining images, many apps such as Paint Shop Pro and Photoshop can
do this. Before I got a proper film scanner that could do multi-exposure,
single pass, I regularly used to make multiple scans on my flatbed scanner
and then combine them in Paint Shop Pro by adding two together selecting a
divisor of 2:

1.jpg + 2.jpg -> 1and2.jpg
3.jpg + 4.jpg -> 3and4.jpg

1and2.jpg + 3and4.jpg -> FinalImage.jpg

Similarly for an average of 8, where further intermediate stages are needed.

With my Epson 1200 PHOTO, I got surprisingly good results (registration
didn't seem to be an issue) and an average of 8 passes gave significant
noise reduction. But that was averaging the results of negative scans. I'm
not sure how good it would be with a digital camera where you're probably at
the low end of the CCD's sensitivity to begin with.

Of course, if you can sacrifice a bit of shadow detail, you can make the
noise less apparent by cutting off the very darkest bit of the histogram in
PSP or by increasing the black level of VueScan.

The same technique of adding images in PSP could be used for creative double
exposures, though I've never tried it. Maybe I will now you've mentioned it.

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Tom Harrison Jr » Tue, 10 Sep 2002 06:35:27


I can see the need as you have described -- there are several utilitarian
reasons and some for cool features.  But I don't see why this should be a
feature of VueScan?  I can't imagine this is something more than a few
people would use, and occasionally, at that.  Since it's already doable in
Photoshop, PSP and perhaps other tools, it seems like it best to keep VS
focused on the scanning.  Just my opinion :-)

On the other hand, there are several different cases where registration of
multiple samples seems to be an issue, including:
* multi-pass multi-scanning on some scanners
* IR cleaning (according to threads in this NG) on at least one scanner
* Long exposure pass

It would seem to me that developing fancier algorithms to align samples
would make most of these problems go away, and is the lion's share of the
effort associated with doing this function.

Tom Harrison


> Bart

> I'm not talking about holding the camera in my hand, but putting it on a
> tripod and using my infra red remote control each time I want to set off
> another frame.

> So registration is definitely not going to be a problem.

> Besides, there is another product on the market which does precisely
> this, but given that all the functionality is already there in VueScan,
> why not use it.  And since others have done it without problems, I'm
> sure that Ed would do a fabulous job too.

> Cheers

> Nicole




> > SNIP
> > > Given that most of the functionality is already there in VueScan, why
> > > not enable multi-pass scanning for "scan from disk" operations?

> > > Would anybody else find this feature useful?

> > Despite the fact that there is nothing wrong with the concept, the
results
> > will be haphazard in practice. The problem is with registration. It
could
> > only work when the images are in perfect pixel alignment (needs
morphing).
> > Since I assume most users use VueScan for scanning film and reflection
> > copies, alignment of different images is everything but guaranteed. Even
> > multiple scan passes of the same image often lack perfect alignment,
causing
> > irregular blur.

> > Bart

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Mike Simmon » Tue, 10 Sep 2002 08:17:11


This is OT now that it's clear that astronomical photos are not the intended
use of this feature.  But here's an answer for those that are interested.  Ron
*did* ask.  :-)


> Would this really work?  I assume your images are VERY dim
> (telescope?).  I can easily foresee the existence of a threshold
> phenomenon, where the number of photons striking the film is linear in
> time, but the film response requires a certain number of seconds to
> "set up", hence is SUBlinear.  Thus four 15-second exposures might not
> be the same as a 60-second exposure.  Any astronomers who could comment
> on this?

Reciprocity failure is the only non-linearity I'm aware of in film
astrophotography, doing the opposite of what you suggest, i.e., causing the
film to become effectively slower with exposure time.  Mulitple images are
often taken and stacked but that's done for various other reasons specific to
astrophotography that I won't go in to.  The images are registered using star
images in the frame.  With film, the frames have to be physically stacked or
registered with successive exposures in the darkroom (I'm not that familiar
with that process; very labor intensive and not done that often).  With CCD
imaging, which has replaced film for many astrophotographers (but not all!),
the registration is done in software, sometimes using automatic procedures.
Digicam's are sometimes used to collect many images of small, bright objects
(short exposures) like planets with the best frames chosen and stacked in
order to overcome problems with atmospheric scintillation in capturing fine
detail.

Quote:> And how do you propose to keep the photos in registration?  Especially
> if they're astronomical observations from a rotating earth.  I suppose
> the telescope would take care of most of this...

The telescope does that, tracking the object across the sky.  This is a lot
simpler to do -- on a basic level -- than one might think and has been done
mechanically for hundreds of years.  Small changes in atmospheric refraction
with changes in an object's altitude and other small errors can be corrected
either by eye with a guide scope or with newer technology such as an
autoguiding CCD camera on a guide scope or within the CCD camera.

Mike Simmons

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Nicole Vincen » Tue, 10 Sep 2002 10:34:19


Oh well, pity that it won't get implemented in VueScan.

Bummer.

Nicole


"multi-pass scanning from file" so I guess it went to the bit bucket.
N.



> > I intend to use successive shots of the same scene to generate a
> > composite image.  The scene may be of star trails for example, or of a
> > moonlit scene.  In either case, the least I can expect is for the
> > electrical noise to be cancelled out as long as the successive shots are
> > not out of register (otherwise blurring will occur).  Other astronomical
> > applications seem possible but they may be limited by the problem you
> > point to.

> I can see how this might be useful, but I can also see
> that a grand total of only a few people will ever use this
> feature if I add it to VueScan.  Given that there's already
> another application available for doing this, I can't see
> any reason to add it to VueScan.

> Regards,
> Ed Hamrick



> spam bit-bucket (unless it has "Vue" in the subject).

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Erik Kraus » Wed, 11 Sep 2002 08:09:53


Hi, Nicole Vincent

Quote:you wrote...
> I've just got myself a digicam, and I'd like
> to be able to take longer exposure shots than the current limit of 15
> seconds.  This would be possible if I could take 4 or more 15-second
> shots from the same position, put them into the VueScan source
> directory, and set VueScan going.  Apart from reducing CCD noise in the
> eventual images, it would also make it possible to get brighter images
> without amplifying noise.

Normal multi pass averages the images data. You would need to add it.
There are several ways to add images with 8 bit per channel. This can
be done with photoshop (Apply Image). You can add up to 4 images with
HDRShop, which is free for non commercial use
http://www.debevec.org/HDRShop and several other applications.

For averaging images you can use PTAverage out of the PanoTools
package: http://www.path.unimelb.edu.au/~dersch/

Unfortunately all these tools produce and work only on 8 bit images.
HDRShop produces HDR images of course, but I found no possibility to
save or convert to 16 bit Tiff. And there are photoshop plugin filters
that average several images to 16 bit (OptiPix).

For this reason I would be glad, if Ed implemented this function you
suggested. An average with 16 bit output would be very fine! Resulting
images could be treated using my photoshop actions. For details have a
look at http://www.erik-krause.de/contrast

--
Erik Krause

 
 
 

VueScan feature request: multi-pass scan from disk

Post by Nicole Vincen » Wed, 11 Sep 2002 08:37:36


Erik

Thank you for all the references.  I'll check them out tonight when I
get back home.

I'd like to see Ed implement this function in VueScan as well, but given
that we are the only two who have _expressed_ an interest in this, I
doubt it'll ever get anywhere near the top of his "to do" list.

Pity though, since I like VueScan's image manipulation algorithms.

Like I said in a previous post in this thread - "bummer".  :-(

Nicole


> Hi, Nicole Vincent
> you wrote...

> > I've just got myself a digicam, and I'd like
> > to be able to take longer exposure shots than the current limit of 15
> > seconds.  This would be possible if I could take 4 or more 15-second
> > shots from the same position, put them into the VueScan source
> > directory, and set VueScan going.  Apart from reducing CCD noise in the
> > eventual images, it would also make it possible to get brighter images
> > without amplifying noise.

> Normal multi pass averages the images data. You would need to add it.
> There are several ways to add images with 8 bit per channel. This can
> be done with photoshop (Apply Image). You can add up to 4 images with
> HDRShop, which is free for non commercial use
> http://www.debevec.org/HDRShop and several other applications.

> For averaging images you can use PTAverage out of the PanoTools
> package: http://www.path.unimelb.edu.au/~dersch/

> Unfortunately all these tools produce and work only on 8 bit images.
> HDRShop produces HDR images of course, but I found no possibility to
> save or convert to 16 bit Tiff. And there are photoshop plugin filters
> that average several images to 16 bit (OptiPix).

> For this reason I would be glad, if Ed implemented this function you
> suggested. An average with 16 bit output would be very fine! Resulting
> images could be treated using my photoshop actions. For details have a
> look at http://www.erik-krause.de/contrast

> --
> Erik Krause

 
 
 

1. VueScan multi-pass scanning and IR scans

Ed

Is there any chance of getting VueScan to only perform the IR scan
_once_ when doing a multi-pass scan with defect removal switched on?  Is
there really that much extra information that one gets from performing
the IR scan multiple times rather than once?

The reason I ask this is because when I perform a 16X multi-pass scan,
it would save a lot of time if the scanner performed the 16 RGB scans,
and only came back for just one IR scan to collect the information about
the defects.

On a related issue, I've noticed that the IR scan can take a _very_ long
time in comparison to the RGB scan.  This made me think about your
comments about how you haven't yet figured out how to control the CCD
gain (or exposure) on the Canon FS4000.  When I use FilmGet (the bundled
TWAIN driver), increasing the gain increases the amount of time each
column of pixels is exposed for (almost as if the scanner was taking a
number of samples on each column), and VueScan's behaviour when
performing the IR scan seems to resemble this,... it's as if you have
managed to control the gain, but only on the IR scan.  Does this
observation get you any closer to figuring out how to control gain on
the Canon FS4000?

Lastly, I've also found that when I use VueScan with my scanner, a
number of bright red, green and blue lines appear on my scans.  I spoke
about this problem to the people at Canon, and a day ago a Canon
representative came across and checked out my scanner, and decided that
this was a hardware problem and that my scanner will be replaced.  Never
the less, what's perplexing about this is that although the lines come
up in VueScan, they do not come up in FilmGet.  Why would this be, and
is it possible to get VueScan to compensate for such problems (or do
whatever it is that FilmGet does) so that VueScan scans don't show up
these lines?  Could it be a problem with the way that VueScan calibrates
the CCD?

Thanks for your continued work on VueScan.

Cheers

Nicole

2. NEW IDEDASD.EXE Cant install Warp4 on 4.3GB drive.

3. Vuescan multi-pass multi-scanning

4. custom computers! at a low cost

5. REPOST: Vuescan multi-pass multi-scanning

6. create list of open models

7. Nikon IVED: Multi-Pass scanning w/ Vuescan

8. Multi-pass scanning in VueScan

9. Multi-pass scanning with Vueprint & Scan Dual II

10. Desperate help request re: Vuescan and Minolta Dual Scan Multi Pro

11. Vuescan w/ Nikon IV-ED ==>I assume multi-pass is possible?

12. Multi-pass scanning: watch out!