canoscan fs4000 - gradual deterioration in results

[Frank Stacey]

I have been using my canon scanner over a the last few years to scan in 100s
of family slides. The quality of these slides is not always that good and
I have been pleased with the scanner's results. Recently the slides I have
been scanning have been generally rather dark and the results have been
disappointing - no variation in the dark regions at all altho' they are
visible (just) on the slides.

Could my system be showing its age in some way?
I am using Photoshop with canon filmGet FS 1.0 driver. I recently upgraded
to 1.0.4 - which seemed to speed things up a bit but I couldn't detect any
other difference. Two questions occur to me:-

a) What is happening while the scanner is "calibrating"? Could the
particular slide(s) that are loaded at the time effect the calibration?

b) could the light source be aging - is it something that can be or should
be replaced? There's no mention of such a thing in the manuals.

While I am on the subject there is one other thing that has always puzzled
me. FS get offers the opportunity to make various corrections based on a
preview of the frame. Does selecting any of these make on iota of
difference to the actual scanning? I have rarely found the corrections
improve matters beyond what I can achieve in Photoshop itself once the scan,
with no corrections at all, has been completed.

TIA
Frank

 

[Steven]

a) What is happening while the scanner is "calibrating"? Could the
particular slide(s) that are loaded at the time effect the calibration?

No. When calibrating, the scanner views the light through a gap between
two frames in the holder.

b) could the light source be aging - is it something that can be or should
be replaced? There's no mention of such a thing in the manuals.

Possibly. If you scan a blank frame (i.e. no slide in holder) is the
resultant scan very close to white ? Also, have you tried manually
setting the exposure level ? Does this make any difference ?

The light probably will dim with age but there is a detector that
adjusts the voltage to compensate for this.

 

[Don]

b) could the light source be aging - is it something that can be or should
be replaced? There's no mention of such a thing in the manuals.

Conventional light bulbs do that over time and eventually burn out.
That's just one of the reasons why the LEDs are considered superior.

You should be able (in theory) to get a replacement but they tend to
be pricey. However, I've read here once about someone getting an
equivalent bulb for his scanner (a flatbed) at a hardware store for a
fraction of the price but I don't remember the scanner model or the
details.

While I am on the subject there is one other thing that has always puzzled
me. FS get offers the opportunity to make various corrections based on a
preview of the frame. Does selecting any of these make on iota of
difference to the actual scanning?

Nope. All that is just "pretend". The scanner still scans the same and
applies all of those after the scan has been completed. No different
to doing all that later in an external image editor as you have
discovered. Actually, doing all that later has many advantages!

I have rarely found the corrections
improve matters beyond what I can achieve in Photoshop itself once the scan,
with no corrections at all, has been completed.

Exactly! That's also known as "scanning raw" i.e. the only things you
adjust in the scanner software are various hardware settings such as
focus or exposure. Everything else (curves and friends) you either
turn off or set to neutral (flat lines). Also, use maximum resolution
and bit depth.

Some people archive such a scan as their "digital negative" and then
work on a copy reducing color depth and resolution as required. That
way when your prints fade or you get a larger monitor you simply go
back to your digital negative and create a new version of the image.
This is not only faster but it's easier on the scanner. However, most
of all, by creating a digital negative you've frozen further film
deterioration!

Anyway, see a parallel message in the thread "Re: Newbie scanner
questions" where I wrote more about all this yesterday.

Don.

 

[Roger S.]

I use the identical scanner and have scanned hundreds of frames in the
last year. Fimget makes corrections with 8-bit precision, so you are
better off "scanning raw" and correcting in 16 bit mode in Photoshop as
Don says.

If you are scanning slides, Filmget simply isn't adequate as it cuts
off the shadow areas to hide noise. While Vuescan is unreliable with
this scanner, with some versions working better than others, it's
really the only game in town. Version 8.3.33 scans slides well and you
can "lock exposure" and increase the exposure to bring out more detail
in denser slides. The base exposure is 1 and Vuescan accepts whole
number values from 1 to 6.

If you're having trouble with dense Velvia or Provia slides, try to
make an exposure at 1 or 2 and then another at 6 and merge the results
with the average images feature of Photomatix http://www.hdrsoft.com/
Be sure to set cropping to "maximum" in Vuescan or the images might not
align well. You should get images without shadow noise that have
better detail and saturation than a single scan.

 

[bmoag]

I have been using this ancient scanner since it first came out. It may not
be perfect but it is good enough for my purposes that I could not justify
replacing it. Its just kind of s l o w.
I agree with Roger S that Vuescan tends to give better results with
transparencies with the Canon scanner compared to the Canon software. For
negative materials I have never been able to get Vuescan to work properly
and find the Canon software more usable.
The Canon is able to scan even dense old Kodachromes but, as you have
noticed, it can be difficult to get desired detail out of dark areas.
For some images you are best off making scans at different exposure
settings, e.g. one for the base exposure and one for the dark areas and
combining the two in Photoshop. If you understand how to combine these scans
seamlessly as layers in Photoshop you can obtain a final image with wider
latitude and less overall noise. This method obviously also works with
negative materials.
Although Vuescan tends to give better results with transparencies, in my
experience meaning more accurate color, for some images the Canon software
may work better. It is up to you if the particular image is valuable enough
to experiment with to get the best results. It can be slow going with the
4000.

 

[Roger S.]

I've been quite happy with the negative handling and encourage you to
search the archives of this group for workflow advice for using Vuescan
with negatives. The main flaw with Vuescan is inconsistent IR
cleaning. Up to 3.3.32, bumping IR exposure to 3 inproved its
effectiveness.

I find that the Canon Filmget IR cleaning leaves weird digital
noise-like artifacts making that unusable for me. The free software I
recommended above (the simple combine 2 images feature is free) does
what bmoag is talking about in an automated fashion.
I don't see why this would help for color negatives as negs are well
within the scanner's dynamic range (full tonal range without visible
scanner noise).