Vuescan 8.01, Epson 4870, RGBI infrared channel offset

[Anthony Ho]

Just got an epson 4870, and the digital ICE feature is very slow (25minutes
for a 645 scanned at 2400dpi).

I learnt from the group that vuescan is much faster in infrared dust
removal. I downloaded a demo and apparently the whole process is much
faster - just about 8 minutes for the first pass + infrared pass + dust
removing calculations.

However, I found most of the defects were not removed. I looked into the
problem by observing a 64bit RGBI TIFF, and realized that the alpha
(infrared) channel is just misregistered, i.e., the red patches (masked
portions of the alpha channel) are all shifted to the left of the defects,
and to make the matter worse, the degree of shifting is not uniform (it
shifts more at the edge of the scanner). As a result the dust removal is
applied to the wrong spots, I believe.

I know 4870 is supported by vuescan. I suspect if that is my scanner's
hardware problem or is it caused by deflection of the glass platform?
However, the epson driver's ICE works. I am puzzled.

Your comment will be much appreciated... thanks.

ant.

 

[Ed Hamrick]

However, I found most of the defects were not removed. I looked into the
problem by observing a 64bit RGBI TIFF, and realized that the alpha
(infrared) channel is just misregistered

Make sure you're doing single-pass scanning (i.e. one pass for RGB
and one pass for infrared).

It seems quite strange that you've found that the infrared data
doesn't align with the RGB data - I haven't gotten other reports
of this.

Regards,
Ed Hamrick

 

[Anthony Ho]

Thanks, I have been using single RGB pass.

I should send the scanner to Epson for a check then.

Anthony

 

[Jan]

Exactly the same problem here since only a few days and a few dozens
of scans. I first noticed something was wrong in the Epson Scan
software. My first thought was that the infrared channel broke down in
some way.
But then Vuescan allowed me to see that the infrared channel works but
is shifted also to the left. I didn't check whether the distance is
constant across the scanner width.

Should I send my scanner in for repair or is there something else I
can do before that ?

Jan

 

[Robert Feinman]

Exactly the same problem here since only a few days and a few dozens
of scans. I first noticed something was wrong in the Epson Scan
software. My first thought was that the infrared channel broke down in
some way.
But then Vuescan allowed me to see that the infrared channel works but
is shifted also to the left. I didn't check whether the distance is
constant across the scanner width.

Should I send my scanner in for repair or is there something else I
can do before that ?

Jan

I noticed something like this the other day too. I thought it seemed
strange, but was too busy to investigate. The ligher image was shifted
several pixels away from the flaw.
I was scanning 4x5 at 2400 dpi with light cleaning and latest version
of viewscan. Windows XP. Single pass scanning, no oversampling, black
and white points set at 0.
So I would say it's a software problem.

 

[Jan]

Well, not a software problem after all, rather a kind of hardware
problem... Here's what I found out almost by accident after about 8
hours of testing this weird problem under different circumstances (I
was preparing to take the scanner in for repair and wanted to document
the problem as good as possible).

I carefully cleaned both the glassplates, and more specifically the
top 10 mm (about 1/2 inch) where the CCD and the backlight stay during
"warming up" of the scanner.
Cleaning the bottom glassplate is something I do regularly, but until
now I had never cleaned the top glassplate, and I didn't pay any
particular attention to the top 10 mm, because that's not the place
where the film or slides are anyway.
This better-than-usual cleaning cured the nasty problem for me. Dust
is removed again, no offset any more between the infrared scan and the
RGB scan. >>>>>
I love ICE !

This makes me conclude that the scanner does more than just "warming
up". I think it also kind of calibrates itself, using the notch in the
holders for film and slides. I'd be _very_ interested to know how this
exactly works.

So folks : clean those glassplates, including the top area !

Jan

 

[RSD99]

"Jan" posted:
"...
I think it also kind of calibrates itself, using the notch in the
holders for film and slides.
...."

That's exactly what is happening ... try putting a couple of little scraps of paper in
that area, and look at the resulting scans from your negatives (or transparencies) ...
they'll have vertical streaks from where the 'calibration area' was obstructed!

"Jan" also posted:
"...
I'd be _very_ interested to know how this
exactly works.
...."

So would I ... on a really technical basis ... because then I could use it to my
advantage!

 

[Jan]

You must be right. I remember now that a few weeks ago I had a tiny
(but clearly visible) vertical streak throughout the slides in one
column. The film didn't show any damage, so I rescanned the slides and
it had disappeared.
I know now that it must have been a tiny bit of dirt in the
calibration area.

I wonder what else this calibration area influences. ICE we know for
sure now, but also other things like colour etc ? The area is also
used for warming up. How does the scanner know when the right
temperature is reached ? Does it tell from the precise colour of the
lamp ? I once lifted the lid during "warming up", and the warming up
didn't come to an end (I kept it open for about 2 minutes).
Could anybody with technical knowledge explain ?

Thanks,
Jan

 

[Ed Hamrick]

This makes me conclude that the scanner does more than just "warming
up". I think it also kind of calibrates itself, using the notch in the
holders for film and slides. I'd be _very_ interested to know how this
exactly works.

I'm just guessing, but it's possible that this area is used to
help align the RGB and Infrared scan passes.

Regards,
Ed Hamrick