Minolta 5400 film scanner - Scan times
Bart van der Wolf
"pioe(rmv)" <"pioe(rmv)"@coldsiberia.org> wrote in message
news:CpOYa.21804$KF1.309083@amstwist00...
SNIP
> But what about its speed?
That of course depends on what settings you use, the connection interface
and your computer's processor/memory speed.
> I have heard that it is excruciatingly slow with ICE on. With
> negatives one routinely has to use ICE.
Whether negatives need it routinely, is a personal matter. I at least try
avoiding the need by working relatively dust free, and careful handling.
Some films are better coated than others to withstand abuse, so it all
depends.
An 'quick' timing test I did on an underexposed Velvia slide:
A 48-bit RGB scan of a 7584x4960 pixel crop took 80 seconds to transfer by
FireWire, or 2821248 bytes/second (say 2.7MB/sec). On a 1.8GHz machine it
averaged a 55% processor load. So a theoretical 1GHz machine would probably
be needed as a minimum, otherwise the scan will be slowed down by the
processor.
Switching on the Grain Dissolver (diffuser), the time was slightly more than
doubled (175sec.), because of the increased exposure time to compensate for
light loss, while the processor load sank to 26% as it had to wait longer
for the data to arrive (1.23MB/sec).
The ICE functionality in the Minolta Scan Utility software can only be
activated with the grain dissolver in place, so that's the starting point to
see if IR needs an additional exposure time. We also need to keep in mind
that the IR opacity varies between films.
The scan time with ICE enabled went to 285sec, BUT the processor load went
to 100%. This means the processor was slowing down the process, and it was
not possible for me to check the IR exposure time increase alone. Whether a
different IR cleaning process, such as in VueScan, will speed-up the process
is a test I haven't done yet (because VueScan isn't fully optimized for the
5400 yet). Apparently the Minolta software does the IR processing as the
data comes in, and how efficient that is done is not (pun intended)
transparant to me.
Color negative film took exactly the same time, despite the mask density, in
its basic scan mode. With the diffuser it took 195sec, so 20 seconds (=11%)
where added, and the processor load went to 30% (Hmm, odd). This is based on
the Minolta Scan Utility, which sometimes does odd things (like clipping
some colors), and it was set to Auto Exposure for negatives. With ICE
enabled (379sec), the processor load went to 100% again, so nothing
generally useful can be concluded from that.
So, for the PC involved, a scan time of 7 minutes seems a worst case
scenario. Of course multiscanning will increase exposure time, but if
negatives need it is questionable. Very contrasty slides can benefit from
upto 4x exposures, 8x seems overkill.
Also, perhaps the diffuser alone will already provide enough "damage
control" that it allows to bypass the need for ICE?
Hope that helps in answering the question...
Bart
Alan Browne
Bart van der Wolf wrote:
> "pioe(rmv)" <"pioe(rmv)"@coldsiberia.org> wrote in message
> news:CpOYa.21804$KF1.309083@amstwist00...
> SNIP
>
> Bart
>
>
Bart, thanks very much for the details. I always thought that ICE was a
function executed in the scanner, not the PC...so the PC saturating to
100% was a bit of an odd one... as well as the long scan time. Did you
try it at a more moderate resolutions such as 2700 and 3818...?
Cheers,
Alan
Christoph Breitkopf
> Bart, thanks very much for the details. I always thought that ICE was
> a function executed in the scanner, not the PC...so the PC saturating
> to 100% was a bit of an odd one... as well as the long scan time.
Not at all. The hardware supplies a 4th data channel to detect
dust and scratches, but the image data for these places is
then corrected by the software. That's what takes most of
the time.
Regards,
Chris
--
Bokeh test images: http://www.bokeh.de/en/bokeh_images.html
Bart van der Wolf
"Alan Browne" <alan....@videotron.ca> wrote in message
news:3F370C18...@videotron.ca...
>
>
> Bart van der Wolf wrote:
> > "pioe(rmv)" <"pioe(rmv)"@coldsiberia.org> wrote in message
> > news:CpOYa.21804$KF1.309083@amstwist00...
> > SNIP
> >
> > Bart
> >
> >
>
> Bart, thanks very much for the details. I always thought that ICE was a
> function executed in the scanner, not the PC
It is a combination, the scanner must provide an infrared lightsource image
of the film, the software must "fill in the defects". The software part
apparently is very timeconsuming, either because of its complexity or poor
programming (I hope the first).
...so the PC saturating to
> 100% was a bit of an odd one... as well as the long scan time. Did you
> try it at a more moderate resolutions such as 2700 and 3818...?
No I haven't tested that, but all I can assume that less data is obviously
taking less time in total anyway. It would also seem logical that the
software needs to process several scan lines for processing a single spot,
so shorter scan lines could be cleared from memory faster but the
calculations would remain. Fewer scan lines may cause a significant speed
up, but that is not what makes this scanner stand out... resolution.
Bart
Alan Browne
I was just thinking about the time issues... are these functions at any
given processor speed sensitive to the type of processor, eg: Celeron v.
P4? The P4 is a F.P. intensive machine where the Celeron is not.
Are the Minolta grain disolve/ICE functions dependant on F.P. in the PC,
IOW?
Cheers,
Alan
Bart van der Wolf wrote:
Bart van der Wolf
> Bart,
>
> I was just thinking about the time issues... are these functions at any
> given processor speed sensitive to the type of processor, eg: Celeron v.
> P4? The P4 is a F.P. intensive machine where the Celeron is not.
>
> Are the Minolta grain disolve/ICE functions dependant on F.P. in the PC,
> IOW?
I have no idea how well the licenced ICE library was written. It should use
the processor specific optimizations, but I somehow doubt it is that clever.
P4, Athlon or MAC platform may indeed make a difference. Different CPU cache
sizes may also make a difference, but I have no way of testing that. That is
why I mentioned the Crop size and the Processor speed (a P4, by the way), so
people can compare. The scans I did, still fitted in RAM without swapping to
virtual memory.
Bart
Michael
Have you already done any tests with Vuescan with regard to the
scanning speed improvement using Minolta Elite 5400 and ICE? What is
actually the reason for this enormous increase of scanning time using
ICE on this device? Nikon scanners need more time for an ICE scan as
well but the scan wiht ICE enabled takes perhaps 1,5-2 times longer,
compared with scans with ICE disabled. In case of Elite 5400 the
factor is about 5-10. A software issue?
Regards
Mike
Ed Hamrick
> What is
> actually the reason for this enormous increase of scanning time using
> ICE on this device?
There are three possible factors:
1) CPU-intensive processing in the Minolta software
2) Longer CCD exposure time caused by grain diffuser
3) Doubled CCD exposure time (double-exposure, RGB and RGB+IR)
VueScan does the infrared processing after the scan is completely
in memory, while the Minolta software appears to do more processing
while the scan is in process. However, I haven't done any analysis
of the CPU utilization of the Minolta software vs. VueScan.
The Nikon scanners don't need a grain diffuser like the Minolta,
since the light is collimated. The Minolta appears to use multiple
uncollimaged IR LED's, and need the diffuser for the IR.
This increases the CCD exposure time.
The Nikon scanners turn on red, green and blue LED's sequentially,
so they only need 33% longer to turn on one additional LED color (IR).
The Minolta scanner does one exposure with the IR LED off, and
one with the IR LED on, which increases the scan time.
Regards,
Ed Hamrick
Robert D Feinman
2.
When I scan I only get half of each scan line captured and there still
are density stripes along the length of the film.
I have grain reduction and IR filtering off.
Ralf Hartings wrote:
> Hi,
>
> I downloaded and installed the newest Vuescan version (7.6.56) today
> with improved support for the Minolta Scan Elite 5400.
> I noticed that scanning times have increased enormously.
> With the .54 version one batch scan at 2700 dpi incl the IR channel took
> less than 3 minites per frame. It now takes 7 minutes (!) with the same
> settings, Which is much longer than with the original Minolta software.
>
> Has anybody similar experiences?
>
>
> Ralf Hartings
> Sweden
>
--
Robert D Feinman
robertd...@netscape.net
Landscapes, Cityscapes, Panoramic Photographs:
http://robertdfeinman.com
Bart van der Wolf
"Robert D Feinman" <robertd...@netscape.net> wrote in message
news:165c49e4.03081...@posting.google.com...
> Am I the only one having problems with .56 and the 5400?
No. I have issues with the filmbase color of Color Neg film when I select
anything else than Generic.
> I'm using USB 2.
> When I scan I only get half of each scan line captured and there still
> are density stripes along the length of the film.
> I have grain reduction and IR filtering off.
I haven't noticed that (yet). I did notice that the IR cleaning is fixed,
but at a price (increased scan time). I'll look a bit closer without IR
enabled and confirm if I have the same, using FireWire.
Bart
Jeroen
> > When I scan I only get half of each scan line captured and there still
> > are density stripes along the length of the film.
> > I have grain reduction and IR filtering off.
I also noticed these stripes, but only with IR cleaning on and less
visible than with the previous versions.
Stange thing is, that I only see them with blue (sky) slides. No
negatives tested yet. Everything else goes perfect even with (light)
IR cleaning.
I'm using USB2, but whith Firewire the same problem occurs. Did a
fresh Windows install, so it can't be a driver problem or whatever.
But it's good to have VueScan back... (even twice: 1 version for raw
scanning, 1 for processing)
BTW: increased scanning time is no problem. Just raw scan the whole
film and let VS process it while you are sleeping. :-D
Dierk Haasis
<bvd...@nospam.nl> wrote:
>"Robert D Feinman" <robertd...@netscape.net> wrote in message
>news:165c49e4.03081...@posting.google.com...
>> Am I the only one having problems with .56 and the 5400?
>
>No. I have issues with the filmbase color of Color Neg film when I select
>anything else than Generic.
Nice to read, I am currently evaluating Vuescan with the Minolta 5400
and stumbled upon two strange phenomena. One being a very odd colour
rendition when I chose (correctly!) the Kodak Royal Gold 100-Gen 2
profile - the pictures got a green cast and where too bright.
The second problem showed up when I engaged ICE, and the option to
regain colour: I got a kind of blurred black grid onto the pictures,
looked as if I shot through a fence.
--
Dierk
Ralf Hartings
>
> BTW: increased scanning time is no problem. Just raw scan the whole
> film and let VS process it while you are sleeping. :-D
That's what I do too. However, using the IR cleaning on the 5400 gives
scanning times per frame of roughly 7 minutes (@2700dpi) in batch mode.
I.e. a strip of 4 frames gives you about 30 minutes. Batch scanning of a
role of 36 exposures means about 9 times 30 minutes= 270 minutes or
about 4,5 hour!
Not using the IR cleaning leads to scanning times of 1 minute/frame or
about 40 minutes per role. Thats quite a difference with 4,5 hour! I
could do a lot of other things in those 4 extra hours....
Then you start the processing while sleeping...
Ralf Hartings
Bart van der Wolf
"Jeroen" <in...@jouwnaam.com> wrote in message
news:9644a2b.03081...@posting.google.com...
> > > I'm using USB 2.
> > > When I scan I only get half of each scan line captured and there still
> > > are density stripes along the length of the film.
> > > I have grain reduction and IR filtering off.
>
> I also noticed these stripes, but only with IR cleaning on and less
> visible than with the previous versions.
The stripes you're mentioning, result from a wrong
calibration/initialization by VueScan. This is an issue that still needs to
be resolved, and it can be resolved because the Minolta Scan Utility does
not show these calibration artifacts.
Bart
Ed Hamrick
> calibration/initialization by VueScan.
I don't see any stripes when using VueScan 7.6.57.
Regards,
Ed Hamrick
Ed Hamrick
> calibration/initialization by VueScan. This is an issue that still needs
to
> be resolved
If you're seeing stripes when scanning from raw scan files in
VueScan 7.6.57 but not when scanning from the scanner directly,
try downloading a fresh copy. I found one small thing wrong when
"Input|Scan mode" was set to "Scan Elite 5400", so I re-released
VueScan 7.6.57 with this fix.
You can download it from:
http://www.hamrick.com/vsm.html
Regards,
Ed Hamrick
Jeroen
Thanks Ed!
Dierk Haasis
wrote:
>If you're seeing stripes when scanning from raw scan files in
>VueScan 7.6.57 but not when scanning from the scanner directly,
>try downloading a fresh copy. I found one small thing wrong when
>"Input|Scan mode" was set to "Scan Elite 5400", so I re-released
>VueScan 7.6.57 with this fix.
Any idea what caused those blurred fence-like lines on my test scans
(on the monitor)? I hope my post has been coming through.
BTW, I know, I am still in eval period, but there is one thing bugging
me like hell: Although I do have a dedicated Vuescan folder on D:\
partition to which I download your setup package, the installer puts
it invariably onto C:\ partition. Very annoying. couldn't that be
changed (either to "install into directory you are started from" or
better still with an option to choose a destination)?
--
Dierk
Dave
scanner?
Dave
"Ed Hamrick" <use...@hamrick.com> wrote in message
news:bhonk8$f8a$1...@ngspool-d02.news.aol.com...
Robert Feinman
I no longer get scan lines, I get full size images (rather than 1/2), and with
the proper settings I can eliminate the clipping I showed on my web site examples.
I'll be purchasing a copy as soon as I figure out how to use it properly.
Is there anyway to set the output profile to anything other than sRGB or is that
a function of the advanced version?
--
Bart van der Wolf
> If you're seeing stripes when scanning from raw scan files in
> VueScan 7.6.57 but not when scanning from the scanner directly,
No, I already see it when scanning from film directly. Mis-calibration is
very visible above VS density 1.70 (1.40 true density). The stripes are also
picked up in the IR channel, so at least it is consistent. In theory it
could be my scanner acting up, but I don't have the issue (all the way up to
measured true density 3.01) with the Minolta software at all..., so it must
be VueScan :-(
Bart
Dierk Haasis
<bvd...@nospam.nl> wrote:
>No, I already see it when scanning from film directly. Mis-calibration is
>very visible above VS density 1.70 (1.40 true density). The stripes are also
>picked up in the IR channel, so at least it is consistent. In theory it
>could be my scanner acting up, but I don't have the issue (all the way up to
>measured true density 3.01) with the Minolta software at all..., so it must
>be VueScan :-(
If I could have a look at an example we can sort that out; if yours
look like mine (I just didn't save them) it is most likely *not* the
scanner.
--
Dierk
Bart van der Wolf
"Dierk Haasis" <Dierk....@Write4U.de> wrote in message
news:1nr0kv4p0u5k7dgql...@4ax.com...
http://www.xs4all.nl/~bvdwolf/main/downloads/index.htm
Same part of the film (steptablet), same scanner, only difference the
software. I'm sure it is not the scanner, because the Minolta software knows
how to compensate when I switch between them.
In case you wonder what you're looking at;
it's a piece of a stepwedge scanned as a positive image in
RAW/16-bit/channel linear. The raw TIFF file is saved as JPEG (no 2.2 gamma
adjustment). The darker patches would show up in the bright areas of
negatives, like skies. Darker patches than the ones shown are all stripe,
almost no detail. If the image is converted to HSB space, it becomes obvious
that the main inconsistency is in Hue, and a bit in Brightness, thus a
calibration issue.
Bart
Dierk Haasis
<bvd...@nospam.nl> wrote:
>Same part of the film (steptablet), same scanner, only difference the
>software. I'm sure it is not the scanner, because the Minolta software knows
>how to compensate when I switch between them.
Well, the stripes I see in your examples are much finer than mine,
yours look more like very fine sand came into your camera.
I had hoped we had the same problem, it isn't. Well, hopefully I get
me some time during the end of the week to test again and find out
more.
Thank you!
PS: Some samples of scans done with the SE 5400 and Minolta software
can be found on www.Private-Parts.Write4U.de/Minolta. They have been
scanned with just a bit of tweaking the curves within the software,
the lion was scanned with ICE. I put them up to give people an
impression about the quality of the scanner (especially grain).
Scanned at 2700 as 48bit TIFF, converted to JPEG.
--
Dierk
Bart van der Wolf
"Dierk Haasis" <Dierk....@Write4U.de> wrote in message
SNIP
> Well, the stripes I see in your examples are much finer than mine,
The scan sample wasn't at full resolution, and the gamma was 1 (or there
about).
> I had hoped we had the same problem, it isn't.
If you post an example, maybe someone (like Ed Hamrick) can have a look...
Bart
Dierk Haasis
<bvd...@nospam.nl> wrote:
>If you post an example, maybe someone (like Ed Hamrick) can have a look...
It'll have to wait for a few days as I am up to my ears in work. I
hope I get it done at the end of the week.
--
Dierk
Jeroen
'Transparency'. When I changed it to 'Scan Elite 5400' (as it should
be) the stripes were gone and I got perfect clean images.
Maybe this helps someone...