Vuescan dust removal on FS4000 broken

[Ian Worthington]

I first started this thread last year (http://groups.google.com/group/
comp.periphs.scanners/browse_thread/thread/35335d1555f9fd8c/
5edf75dba9525b67?hl=en#5edf75dba9525b67) but then got very busy at
work and ran out of time.

I've looked at the results with Vuescan 8.3.75 today and dust removal
(colour negs) sill doesn't work (IR3, heavy removal).

Looking more closely it seems the IR channel appears to be displaced
from the RGB image. Is this possible? Could I have a bad unit? The
scanner appeared to do two physical passes: is this expected?)


ian
....

 

[Barry Watzman]

The Nikon scanners should not have any registration problems. Because
they use LED illumination, unlike other forms of illumination (lamps),
all "passes" for the final scan are done at once. As the mechanism
stops at each "scan line", the unit "flashes" the separate LEDs for red,
blue, green and IR in sequence, then the carriage moves to the next
"scan line".

There are (or may be, depending on settings) multiple passes, but they
are for auto-focus, auto-exposure and then the actual scan. However, as
far as I understand the actual scan itself is done in a single pass.

 

[Steven Saunderson]

Looking more closely it seems the IR channel appears to be displaced
from the RGB image. Is this possible? Could I have a bad unit? The
scanner appeared to do two physical passes: is this expected?)

The FS4000 needs two passes. The first has the normal light enabled for
RGB collection. The second has the IR LED enabled for the dust scan.

I think the scans will often be misaligned and the software (Vuescan in
this case) has to resolve this. Have you tried an older version of
Vuescan to see if it solves your problem ?

 

[Jeff Randall]

I've looked at the results with Vuescan 8.3.75 today and dust removal
(colour negs) sill doesn't work (IR3, heavy removal).

Have you tried the latest version of VueScan 8.4.56? I think somewhere
along the way Ed fixed or at least improved the FS4000 cleaning
function problem.

jr

 

[Roger S.]

Have you tried the latest version of VueScan 8.4.56? I think somewhere
along the way Ed fixed or at least improved the FS4000 cleaning
function problem.

As I've happily not been scanning I haven't tried it. Last time I
checked 8.4.37? it worked better than it ever has for single images
but managed to simply not save with IR correction for batch scans of
color negatives. The second pass was done and if you hit the save
button you could get a file with IR cleaning applied (sometimes) but
it didn't work automatically.
I don't know if this was fixed.

Please help Ed troubleshoot, other FS4000US users.

He did fix a huge focus problem around 8.4.3? that didn't exist at the
time of 8.3.75. "Progress"

Roger

 

[Ian Worthington]

As I've happily not been scanning I haven't tried it. Last time I
checked 8.4.37? it worked better than it ever has for single images
but managed to simply not save with IR correction for batch scans of
color negatives. The second pass was done and if you hit the save
button you could get a file with IR cleaning applied (sometimes) but
it didn't work automatically.
I don't know if this was fixed.

Please help Ed troubleshoot, other FS4000US users.

He did fix a huge focus problem around 8.4.3? that didn't exist at the
time of 8.3.75. "Progress"

Roger

Am trying 8.4.56. It seems better, the alignment problem seems to be fixed.

I'm trying some batch scanning of APS film with it and am seing two things:

1. If I save to RAW file format 64 bit RGBI, the output seems
significantly grainier than if I select file format AUTO. Do I
misunerstand that "auto" is the best quality setting available?
Although the file size is the same in both cases, the contents clearly
differ: AUTO will generate Explorer thumbnails and can be read in
acdsee, whereas 64bit rgbi doesn't do either of these things.

2. The IR layer is blank. This could of course be as a result of the
film being sealed in its cassette, but I'm surprised to see no dust at
all on there from lab handling.

What I want as output from the process is:

i. The best quality "digital negative" I can get to be used for storage
and any future work I choose to do.

ii. A set of reasonably high-quality jpegs. These I seem to be able to
get by post-processing the RAW scans in vuescan and saving at 95% (gives
a 3MB file: good enough). I also save a TIFF at this stage for the day
when the vuescan-format RAW file is unreadable. The jpegs from 64b RGBI
though are much granier than those from AUTO. What's the best fix for that?

ian
....

 

[Ian Worthington]

As I've happily not been scanning I haven't tried it. Last time I
checked 8.4.37? it worked better than it ever has for single images
but managed to simply not save with IR correction for batch scans of
color negatives. The second pass was done and if you hit the save
button you could get a file with IR cleaning applied (sometimes) but
it didn't work automatically.
I don't know if this was fixed.

Please help Ed troubleshoot, other FS4000US users.

He did fix a huge focus problem around 8.4.3? that didn't exist at the
time of 8.3.75. "Progress"

Roger

Am trying 8.4.56. It seems better, the alignment problem seems to be
fixed.

I'm trying some batch scanning of APS film with it and am seing two
things:

1. If I save to RAW file format 64 bit RGBI, the output seems
significantly grainier than if I select file format AUTO. Do I
misunerstand that "auto" is the best quality setting available?
Although the file size is the same in both cases, the contents clearly
differ: AUTO will generate Explorer thumbnails and can be read in
acdsee, whereas 64bit rgbi doesn't do either of these things.

2. The IR layer is blank. This could of course be as a result of the
film being sealed in its cassette, but I'm surprised to see no dust at
all on there from lab handling.

What I want as output from the process is:

i. The best quality "digital negative" I can get to be used for
storage and any future work I choose to do.

ii. A set of reasonably high-quality jpegs. These I seem to be able
to get by post-processing the RAW scans in vuescan and saving at 95%
(gives a 3MB file: good enough). I also save a TIFF at this stage for
the day when the vuescan-format RAW file is unreadable. The jpegs
from 64b RGBI though are much granier than those from AUTO. What's
the best fix for that?

ian
....

 

[Roger S.]

64 bit RGBI files are only useful in Vuescan. If you want a TIFF for
output or archiving save as 48 bit and maximum resolution. The extra
channel's probably messing things up in the other programs.

"2. The IR layer is blank. This could of course be as a result of
the
film being sealed in its cassette, but I'm surprised to see no dust
at
all on there from lab handling. "

When you click on pixel colors does anything show up in Red? If not
IR cleaning isn't working and this is the same problem I've been
having. There will be *something* in that channel if IR is doing
anything.

 

[Ian Worthington]

64 bit RGBI files are only useful in Vuescan. If you want a TIFF for
output or archiving save as 48 bit and maximum resolution. The extra
channel's probably messing things up in the other programs.

"2. The IR layer is blank. This could of course be as a result of
the
film being sealed in its cassette, but I'm surprised to see no dust
at
all on there from lab handling. "

When you click on pixel colors does anything show up in Red? If not
IR cleaning isn't working and this is the same problem I've been
having. There will be *something* in that channel if IR is doing
anything.

Thanks Roger, I'll give (2) a try.

Regarding (1), I'm still confused why saving in 64b RGBI format should
give a noisier image than those saved in AUTO format. Any thoughts?

The idea is to save two versions: one RAW for future work by vuescan,
and then one in TIFF for compatability.

Thanks,

Ian

 

[Ian Worthington]

64 bit RGBI files are only useful in Vuescan. If you want a TIFF for
output or archiving save as 48 bit and maximum resolution. The extra
channel's probably messing things up in the other programs.

"2. The IR layer is blank. This could of course be as a result of
the
film being sealed in its cassette, but I'm surprised to see no dust
at
all on there from lab handling. "

When you click on pixel colors does anything show up in Red? If not
IR cleaning isn't working and this is the same problem I've been
having. There will be *something* in that channel if IR is doing
anything.

IR cleaning seems ok for me. This is what I see:

When I scan dusty 35mm color negs and select Color > View color >
Infrared, I see plenty of black dots on a white background. Clicking
Pixel Colors turns a varying number of those dots red, seemingly
dependent on the setting of Filter > Infrared clean. I have to have
Infrared clean on something other than None, else the whole IR layer
appears a very dark gray with darker spots (this is whatI saw earlier
and led me to believe that was an problem), the larger of which appear
to correspond to the defects. Most of the dots do not appear to
correspond to any visible defect (dust) and turning on/off cleaning does
not affect those locations. It does those remove the larger of the
identified defects.

When I scan APS negs, it seems to work the same. Finding some visible
defects took a while, but they too seemed to get correctly fully using
the Heavy setting.

ian
....

 

[Roger S.]

IR cleaning seems ok for me.  This is what I see:

When I scan dusty 35mm color negs and select Color > View color >
Infrared, I see plenty of black dots on a white background. Clicking
Pixel Colors turns a varying number of those dots red, seemingly
dependent on the setting of Filter > Infrared clean.  I have to have
Infrared clean on something other than None, else the whole IR layer
appears a very dark gray with darker spots (this is whatI saw earlier
and led me to believe that was an problem), the larger of which appear
to correspond to the defects.  Most of the dots do not appear to
correspond to any visible defect (dust) and turning on/off cleaning does
not affect those locations.  It does those remove the larger of the
identified defects.

When I scan APS negs, it seems to work the same. Finding some visible
defects took a while, but they too seemed to get correctly fully using
the Heavy setting.

ian
...

This performance seems normal. In Photoshop if you jack up the
contrast of the IR channel using levels you should be able to see the
defects. Changing IR cleaning from low to high changes the threshold
for prominence of dust to be removed.
I usually do cleaning on "light" and encourage you to see if heavy is
reducing image detail.

 

[Ian Worthington]

Hmm.

More confusion on this.

When I scan at crop maximum or auto, I get a totally dark IR layer (as
viewed in vuescan or Photoshop). When I scan a manual crop area I get a
white IR layer with some black dots in it.

Selecting pixel colors in vuescan seems to put red dots on both types
and it does appear to be removing white dots but the difference doesn't
fill me with confidence.

ian

 

[degrub]

Just curious are you viewing the pixels at 100% or just what PS is
automatically giving you ? That could explain the differnce between the
crop scan and the full scan.

Regards,

 

[Ian Worthington]

Just curious are you viewing the pixels at 100% or just what PS is
automatically giving you ? That could explain the differnce between the
crop scan and the full scan.

Regards,

At 100%.

i

 

[DenverDad]

Ian,

In the previous thread you linked, I reported that I was using an
older version of Vuescan - 8.3.07 to be specific - due to precisely
the same kind of IR cleaning problems you were seeing. Since then I
have occasionally tried the latest version hoping that I would finally
be able to take advantage of the newer features and slicker
interface. Unfortunately, the IR cleaning problem has persisted for
me in every new version I have tried. So I am STILL using 8.3.07!
Fortunately, it works very well for me and it only takes the "light"
setting to get nearly perfect results. (I can send you a copy of the
executable for that version if you want to try it)

I don't remember the exact history of all this - you can search the
archives and find at least a couple threads where I was trying to
troubleshoot IR cleaning problems myself. But as I recall IR cleaning
has been broken and fixed at least a couple times. I can say that Ed
Hamrick was very receptive and straightforward in working to fix it
the one time I reported the problem to him. So it may be something
you want to consider if you are still seeing some artifacts (i.e., the
dots you report - this is something I have seen too).

If I get a chance I will try the latest version myself and we can
compare notes.

Jeff

 

[Roger S.]

Cleaning seemed fine in the early 8.4.xx and more effective than the
old version was (no chipping away at edges in slides, etc). However,
focusing didn't work right on these versions. Focusing now works
again but IR seems broken.

For better or worse my local pro lab no longer does C-41 so I'm not
shooting much film anyway or I'd try to troubleshoot.