Canon FS2710 slow scans with Canocraft FS - fast with Vuescan

[larry07]

Have a FS2710 film scanner connected via the provided Adaptec SCSI card
under Windows XP. The unit works fine. I usually scan using Photoshop
and use the Twain interface to drive the Canocraft FS software. Scans
work fine but just to scan one negative at full 2720 dpi resolution and
24 bit color it can take up to 3 minutes before the image becomes
available in Photoshop.

Out of interest tried Vuescan. Using Vuescan at the same resolution and
color depth a single negative scans in about 45 seconds! Even taking
into account saving the scanned file and firing up Photoshop to edit
the picture, the time difference is huge.

Given that both pieces of software are using the ASPI layer to talk to
the scanner can anybody help me out with why one should be so much
faster than the other. The scanner is rated at about 45 seconds for a
scan and that presumably is using Canocraft FS software. If I use the
software directly, without going via Photoshop, there is no noticeable
speed difference - it's still very slow

Thanks

Larry

 

[Don]

Have a FS2710 film scanner connected via the provided Adaptec SCSI card
under Windows XP. The unit works fine. I usually scan using Photoshop
and use the Twain interface to drive the Canocraft FS software. Scans
work fine but just to scan one negative at full 2720 dpi resolution and
24 bit color it can take up to 3 minutes before the image becomes
available in Photoshop.

Out of interest tried Vuescan. Using Vuescan at the same resolution and
color depth a single negative scans in about 45 seconds! Even taking
into account saving the scanned file and firing up Photoshop to edit
the picture, the time difference is huge.

Given that both pieces of software are using the ASPI layer to talk to
the scanner can anybody help me out with why one should be so much
faster than the other. The scanner is rated at about 45 seconds for a
scan and that presumably is using Canocraft FS software. If I use the
software directly, without going via Photoshop, there is no noticeable
speed difference - it's still very slow

It's like jumping out of an airplane. You can do so with a parachute,
or you can do so without a parachute. If you leave the parachute
behind you'll reach the ground considerably faster! ;o) Using Vuescan
is just like that and Vuescan output looks about the same! ;o)

But seriously, there are several reasons for this.

One is that your native Canon software uses the recommended ("proper")
way of accessing the scanner i.e. TWAIN. Due to TWAIN's complexity
this takes longer but it will work on any system. Indeed, that's the
whole point of TWAIN i.e. to abstract the hardware. Vuescan bypasses
TWAIN and accesses the scanner directly which can speed things up.
That's how all scanners worked initially but people quickly realized
that was very messy and required custom programming for each new
hardware and system, which is why TWAIN was invented as a common
interface. That way programs only need to talk to TWAIN and let it
sort out low level hardware access.

Another reason is that native software is usually very careful about
access ("belt and suspenders") while Vuescan is notoriously sloppy.
One example of this was the Vuescan Minolta saga where native Minolta
software worked just fine while Vuescan struggled with it for over two
years with "stripes" appearing in scans. This was finally solved when
a knowledgeable contributor here explained (indirectly) to the Vuescan
author how this should be done.

In general, Vuescan is just too buggy and unreliable to be used for
any serious scanning. It may OK for a casual scan intended as a highly
compressed small JPG but that's about it. On the other hand, if you
don't care for quality it may be acceptable in which case find a
version with least bugs and stick with it. Avoid upgrading
automatically and let others debug the most blatant bugs first.

Don.

 

[larry07]

This all sounds reasonable. I don't have high end scanning needs. Just
wanted something better than my Konica film scanner which only does
1200dpi! But while it has lower resolution it does have something over
the Canon - more automation - when you put an APS cartridge in, it
preview scans the entire roll and when you put a negative strip in, it
shows all the negatives.

I was just surprised that the Canon (which I bought used) is much
slower than what the specs say and that the only way I can get close to
the rated speed is to use non standard software.

What I might do is use Vuescan to do the fast previews and then
Photoshop and Twain to do the final scan - best of both worlds I guess

Larry

 

[Bruce Graham]

This all sounds reasonable. I don't have high end scanning needs. Just
wanted something better than my Konica film scanner which only does
1200dpi! But while it has lower resolution it does have something over
the Canon - more automation - when you put an APS cartridge in, it
preview scans the entire roll and when you put a negative strip in, it
shows all the negatives.

I was just surprised that the Canon (which I bought used) is much
slower than what the specs say and that the only way I can get close to
the rated speed is to use non standard software.

What I might do is use Vuescan to do the fast previews and then
Photoshop and Twain to do the final scan - best of both worlds I guess

Larry

you will probably find that you will get better results using Vuescan
especially on transparencies. I have the Canon FS4000 and I'm guessing
the supplied Canon 2700 software is similar. The main reason my Canon
4000 software is so slow is that always waits for the lamp to stabilise,
then calibrates the scanner before every scan. Vuescan leaves this under
the control of the user (you calibrate when you think you need to). I
usually disagree with Don's conclusions but his analogy of the parachute
might be accurate this time. I personally prefer to fly most times
without my calibration parachute.
Bruce

 

[Roger S.]

"In general, Vuescan is just too buggy and unreliable to be used for
any serious scanning. It may OK for a casual scan intended as a highly
compressed small JPG but that's about it"

This statement isn't based on any first-hand experience with Filmget
and Vuescan, so take it with a grain of salt.

I use the FS4000US with Vuescan. About a year ago VS's calibration
didn't work consistently with the scanner, and I noticed faint
red/green lines. I now calibrate when I start scanning and haven't
noticed any problems. Filmget seems to move the holder all the way in
and then all the way out again before it starts scanning. I don't know
if this improves the alignment but I can't see any difference in the
results compared to Vuescan.

The current version of VS seems fully functional and bug-free except
manual cropping doesn't work when you batch scan slides- leave it on
auto or maximum and don't fix the rotation.

 

[Don]

I was just surprised that the Canon (which I bought used) is much
slower than what the specs say and that the only way I can get close to
the rated speed is to use non standard software.

This may also be because of XP and SCSI. I seem to recall reading
people having problem with SCSI devices under XP.

BTW, I personally run W98 so take that with a boulder of salt...

Anyway, if that's the case, non standard software talks to SCSI
directly which would explain it.

Don.

 

[Don]

"In general, Vuescan is just too buggy and unreliable to be used for
any serious scanning. It may OK for a casual scan intended as a highly
compressed small JPG but that's about it"

This statement isn't based on any first-hand experience with Filmget
and Vuescan

That's true. The above statement is not based on a single isolated
case without any metrics or context.

Instead, it's based on an interrupted flow of bug reports and detailed
messages by many irate Vuescan users over a period of 4 years.

I use the FS4000US with Vuescan. About a year ago VS's calibration
didn't work consistently with the scanner, and I noticed faint
red/green lines.

Did you ever have any *calibration* problems with Filmget?

There you go!

Don.

P.S. Before you rush to change the subject, just a reminder that the
subject you yourself brought up was *calibration*.

 

[Don]

I
usually disagree with Don's conclusions but his analogy of the parachute
might be accurate this time. I personally prefer to fly most times
without my calibration parachute.

In which case, Vuescan is not only perfect for you, but you deserve
it! ;o)

On a serious note, that's exactly what I've been saying all along. If
one doesn't care for quality (e.g. doesn't wait for the lamp to
stabilize and then calibrate) then even Vuescan will do.

The only reason such a neutral, matter-of-fact statement makes some
Vuescan users explode with blind rage is because they have that huge
chip on their shoulder. The ones that don't, simply admit they don't
care for quality, shrug, and continue using it.

Don.

 

[Roger S.]

For the record I'm using SCSI on a WinXP laptop with Vuescan.

Calibration has never been a problem in my experience with Filmget.

Vuescan has had a host of problems with this scanner since I started
using it in September 2004. Ed seems to test VS with Nikons, as I
don't hear many bugs about them. Fortunately most of the problems are
currently resolved.
The only "bug" I've noticed with Filmget are strange digital noise-like
artifacts on color negatives when using FARE cleaning. Vuescan does a
mediocre job of cleaning but doesn't leave artifacts. I've not had
acceptable results with Filmget and darker slides- the exposure just
isn't adequate and it's not clear if there's a way to change it. I
just scanned about 50 slides with Vuescan, many of them night shots on
Provia 100F and am happy with the results.

I think even Don would agree that the best course of action is to try
both programs on the same slides/negatives and see which one gives you
better results for your intended purpose.

 

[Bruce Graham]

For the record I'm using SCSI on a WinXP laptop with Vuescan.

Calibration has never been a problem in my experience with Filmget.

except that Filmget calibrates every time you use it - so it works but
wastes a lot of time.

Vuescan has had a host of problems with this scanner since I started
using it in September 2004. Ed seems to test VS with Nikons, as I
don't hear many bugs about them. Fortunately most of the problems are
currently resolved.
The only "bug" I've noticed with Filmget are strange digital noise-like
artifacts on color negatives when using FARE cleaning.

Yes, Filmget scribbles over some fine detail, especially lines, with
random angled stripes of cloned background.

Vuescan does a
mediocre job of cleaning but doesn't leave artifacts. I've not had
acceptable results with Filmget and darker slides- the exposure just
isn't adequate and it's not clear if there's a way to change it.

and then Filmget clips all the shadows to black so you don't see the
noise. It also clips the highlights - if it has a real whitepoint
adjustment I haven't found it.

just scanned about 50 slides with Vuescan, many of them night shots on
Provia 100F and am happy with the results.

I think even Don would agree that the best course of action is to try
both programs on the same slides/negatives and see which one gives you
better results for your intended purpose.

Don has yet to agree that Vuescan has any redeeming feature.

Bruce G

 

[Bruce Graham]

In which case, Vuescan is not only perfect for you, but you deserve
it! ;o)

On a serious note, that's exactly what I've been saying all along. If
one doesn't care for quality (e.g. doesn't wait for the lamp to
stabilize and then calibrate) then even Vuescan will do.

The only reason such a neutral, matter-of-fact statement makes some
Vuescan users explode with blind rage is because they have that huge
chip on their shoulder. The ones that don't, simply admit they don't
care for quality, shrug, and continue using it.

shrug, Vuescan gives better quality than Filmget for most scans (and the
calibration holds for a long time - I think Vuescan automatically
calibrates every month now).

 

[Roger S.]

"Don has yet to agree that Vuescan has any redeeming feature"

Well it's a good thing we don't listen to him, then ; )

 

[larry07]

I think the FS2710 had some problem with XP but I didn't have them.
This is because I had been using a Konica SCSI scanner for a while on
98 and migrated that to XP so I had an ASPI layer already installed.
Besides which I had 3 SCSI devices on the chain - the Konica scanner, a
hard drive and a HP Scanjet 4P so was pretty familiar with making it
work. Only had to install the Canocraft FS software and I was done.

Non standard software bypassing the TWAIN interface might explain the
difference in speeds - it's just surprising that's all

Larry

 

[Don]

shrug, Vuescan gives better quality than Filmget for most scans

That's my point, exactly! BTW, I wasn't attacking you. Unlike the
"rabid Vuescan sect" you're happy with the results. So what if Vuescan
is not a quality product? Big deal! It does what you want and that's
all that counts.

(and the
calibration holds for a long time - I think Vuescan automatically
calibrates every month now).

That premise is not correct. A calibration does not "hold" because the
conditions inside the scanner change constantly. Therefore - if one
cares for such things - the software (or the user) should calibrate at
regular intervals.

Now, in case of conventional light source scanners, the light bulb
also slowly deteriorate over time. In that case a monthly calibration
should cover this gradual change.

However, during regular use the conditions fluctuate quite a lot, far
more than the gradual deterioration of the conventional light bulb,
and - again, if one cares for such things - that means more frequent
calibration.

Don.

 

[Don]

I think even Don would agree that the best course of action is to try
both programs on the same slides/negatives and see which one gives you
better results for your intended purpose.

Not "even" but *especially* Don would agree!

Again, this is blatantly self-evident to anyone who actually *reads*
Don's messages instead of just emotionally lashing out.

!=> Don often includes the link to the Vuescan home page so people can
make up their own mind. <=!

However, it's also only fair to give a rundown of known Vuescan
"issues" because once they part with their money there are no refunds
no matter how unusable Vuescan turns out to be later, as many
frustrated "users" have repeatedly reported here.

Don.

P.S. Can Don now stop referring to Don in the third person? ;o)

 

[Don]

Don has yet to agree that Vuescan has any redeeming feature.

That's a prime example of baseless mud-slinging.

And that after I just complimented you in the other message about how
reasonable you were! (fx. shakes head) Oh well... :-/

If you bothered to actually *read* my messages before throwing mud you
may have noticed that they are always objective, factual and give
credit where credit is due.

Here are two specific instances where I complimented Vuescan in case
of my old LS-30:

1. It retrieved the full 10-bits (NikonScan only produces 8-bit files)

2. It had no artificial limit on exposure (NikonScan limits exposure
to +/-2 EV). However, the Vuescan exposure bug kicks in as the values
go up resulting in "random exposures" and the readout is out of sync
with the actual exposure value. And, of course, Vuescan employs the
immensely inane "multiplier" instead of the industry standard EV.

I have also repeatedly stated that (as an assembler programmer at
heart) I find direct access to hardware "neat".

So much for your theory...

Don.

 

[Don]

"Don has yet to agree that Vuescan has any redeeming feature"

Well it's a good thing we don't listen to him, then ; )

The facts, i.e. "your" (plural) obsessive and reflex attacks, indicate
exactly the opposite.

If "you" (plural) really did listen, i.e. actually *read* before
lashing out, maybe you'd come up with some cogent and civil responses.

Don.

 

[Bruce Graham]

That's a prime example of baseless mud-slinging.

And that after I just complimented you in the other message about how
reasonable you were! (fx. shakes head) Oh well... :-/

If you bothered to actually *read* my messages before throwing mud you
may have noticed that they are always objective, factual and give
credit where credit is due.

Here are two specific instances where I complimented Vuescan in case
of my old LS-30:

1. It retrieved the full 10-bits (NikonScan only produces 8-bit files)

2. It had no artificial limit on exposure (NikonScan limits exposure
to +/-2 EV). However, the Vuescan exposure bug kicks in as the values
go up resulting in "random exposures" and the readout is out of sync
with the actual exposure value. And, of course, Vuescan employs the
immensely inane "multiplier" instead of the industry standard EV.

I have also repeatedly stated that (as an assembler programmer at
heart) I find direct access to hardware "neat".

So much for your theory...

Don.

well I'm glad to stand corrected!

BTW Canon Filmget also has an arbitrary exposure scale - I think it
relates to the scanner stepper motor multiplier. With all of these odd
animals in the scanner farmyard, it would be difficult for Mr Hamrick to
convert them all to EV (yes, I know about lowest common denominator -
Vuescan is a lot better than that).

Bruce G

 

[Bruce Graham]

That's my point, exactly! BTW, I wasn't attacking you. Unlike the
"rabid Vuescan sect" you're happy with the results. So what if Vuescan
is not a quality product? Big deal! It does what you want and that's
all that counts.


That premise is not correct. A calibration does not "hold" because the
conditions inside the scanner change constantly. Therefore - if one
cares for such things - the software (or the user) should calibrate at
regular intervals.

Now, in case of conventional light source scanners, the light bulb
also slowly deteriorate over time. In that case a monthly calibration
should cover this gradual change.

However, during regular use the conditions fluctuate quite a lot, far
more than the gradual deterioration of the conventional light bulb,
and - again, if one cares for such things - that means more frequent
calibration.

Don.

I'm not sure what you mean by "conventional light bulb" in this context.
I think the Canon has a flouro source which may be what you meant.

Anyway, failure to calibrate for a long period can result in very slight
banding, which I think is drift (and maybe microscopic dust on) in the
CCD/amplifier chain rather than the lamp. Monthly calibration seems to
be enough to eliminate this effect.

I'm sure that some spectral changes occur as well, but given that I
perceptually colour correct each scan, that effect is quite irrelevant in
my workflow. Final correction is done in post processing anyway.
Conceivably, the spectral response could change during the scan if
scanning commences immediately after the light is switched on but I have
not been able to see any change in colour across the scan. Both Vuescan
and Filmget do this in batch scanning when IR cleaning is used because
the main lamp must be turned off while the IR scan is done, then on again
for the next RGB scan.

Bruce G

 

[Don]

I'm not sure what you mean by "conventional light bulb" in this context.
I think the Canon has a flouro source which may be what you meant.

That's right. Virtually any light source other than LEDs suffers from
that and it eventually burns out. LEDs, by contrast, last "forever"
and will outlive both the scanner and the user.

Anyway, failure to calibrate for a long period can result in very slight
banding, which I think is drift (and maybe microscopic dust on) in the
CCD/amplifier chain rather than the lamp. Monthly calibration seems to
be enough to eliminate this effect.

I'm sure that some spectral changes occur as well, but given that I
perceptually colour correct each scan, that effect is quite irrelevant in
my workflow. Final correction is done in post processing anyway.
Conceivably, the spectral response could change during the scan if
scanning commences immediately after the light is switched on but I have
not been able to see any change in colour across the scan. Both Vuescan
and Filmget do this in batch scanning when IR cleaning is used because
the main lamp must be turned off while the IR scan is done, then on again
for the next RGB scan.

Yes, often times we lose track when chasing "perfection" so it helps
to lean back, every now and then, and look at the big picture in full
context. As in: How important is such accuracy, really? I do this all
the time, as well, and then make compromises accordingly.

In case of calibration, in my workflow it doesn't take any time so I
calibrate before each image. Right now, I'm busy with mounted slides
so I take a slide out, click on calibrate, and then use the blow brush
to clean the next slide. By the time I'm done cleaning the scanner is
done calibrating so I don't really have to wait. But in real terms,
it's an overkill potentially putting extra wear and tear on the
scanner as the assembly shuffles back and forth.

Don.