difficulty drum scanning negatives

[Gordon Moat]

Thanks for all who responded. I got the scans from the office and they
look horrible. I don´t think it´s grain, ít´s noise, noise, noise!
Colors look posterised with no gradation observed. The histogram shows
no abnormalities however (?)

That comment makes me think you are viewing it on a monitor, and judging it as noise. A smooth
Histogram should indicate a smooth tonal transition. The reasoning behind drum scanning is to
eventually print the image. I think at this point, you should do a test print, proof print, or
match print, and then make a better judgement.

I don't believe the problem is the film,
it's the scan that it's bad. If anybody is interested I can send
portion of the image for viewing.

J

All monitors are such poor resolution in comparison to printing, especially commercial offset
printing. I have seen many image files that seemed noisy on a monitor, yet printed very
smoothly. You could easily be running into a limitation of viewing on a monitor.

If you can do a test print, then you would have a better item to judge scan quality. If it
still looks bad, then the original scan is at fault.

If you still find that after viewing printed results that things are not working, it is then
down to operator error, or a weird technical problem. I remember working on one workstation
that showed unusual results on most scans. What we finally traced that down to was an extra
monitor interfering with the SCSI cable of the scanner, and causing strange noise issues. The
reason that was so tough to track was that the monitor was a secondary monitor, and not always
used at that workstation. Routing the SCSI cable further from that monitor solved the unusual
noise problem. While I would be surprised if that is the problem you are having, if nothing
else is working, then investigate that direction. Best of luck.

Ciao!

Gordon Moat
A G Studio
<http://www.allgstudio.com>

 

[Jytzel]

Out of curiosity... have you made an optical print of the film yet? And,
how are you observing the scans?

Regards,

Neil

No, not yet. I´m observing the scan on the monitor! I don´t really
understand what you mean.

thanks,
J

 

[Kennedy McEwen]

However, for the purpose of establishing grain aliasing, one should only
have to examine the edge of a high-contrast portion of the image under
magnification and compare that to an equivalent zoom magnification of the
digital file (obviously not to the pixel level). If the on-screen profile
matches the optical profile... no aliasing... if they're overly blocky...
aliasing. Why would this method not yield a "valid comparison"?

Gaaaah!!! Not again! Go back to the start fo the thread and read all my
posts again!

Recorded image resolution is likely to be less than the granular
resolution, so no, you can't just examine a high contrast edge and looks
for jaggies - even if they aren't there, there could still be grain
aliasing. Furthermore, jaggies are not an indication of aliasing,
merely a consequence of a square pixel output filter representation!

You've heard it all before, had the details spelled out in single
syllables and complained that you undertand it all, yet you still cling
to this untruth - so what's the point?

 

[Kennedy McEwen]

Thanks for all who responded. I got the scans from the office and they
look horrible. I don´t think it´s grain, ít´s noise, noise, noise!
Colors look posterised with no gradation observed. The histogram shows
no abnormalities however (?) I don't believe the problem is the film,
it's the scan that it's bad. If anybody is interested I can send
portion of the image for viewing.

Yes, send me a section, but I am going off on business for a week, so
won't be able to comment on it till I get back.

 

[Kennedy McEwen]

That comment makes me think you are viewing it on a monitor, and
judging it as noise. A smooth
Histogram should indicate a smooth tonal transition.

....or a noisy one - as Jytzel has described. Noise will only show up on
a histogram as a distribution around what should be perfectly uniform
areas - but few real world images have such areas so noise, to all
intents and purposes, is not observable from histogram views.

However, I agree with your concern that this is simply a case of viewing
on a monitor. Essentially there are two extreme options:
* viewing at 1:1 on the monitor pixel scale, which will result in the
image appearing soft and grainy, just due to the magnification,
* viewing at actual print scale, in which case the image will look
grainy due to aliasing due to the simple downsampling algorithm for
display.

Jytzel, if you want to view the image effectively on the monitor for a
fairly valid comparison with an actual print then resample the image to
your monitor size, do not just zoom it. For example, a 35mm frame
scanned at 4000ppi is around 5500x3700pixels. My monitor display is
1600x1200pixels, so I would resize the image to 25% of the original, to
get 1375x925pixels.

If you implement the resize in Photoshop or PSP then some filtering is
implemented prior to the resize to reduce aliasing, however it can be
worth additional filtering before the resize to be sure that no aliasing
occurs. In Photoshop, the best way of achieving this is to use gaussian
blur with a radius no more than 2pixels for a resize to 25% - ie. half
the resize factor. After the resize, implement an unsharp mask at
around 150% with a radius of 0.7 pixels to recover the loss in sharpness
caused by downsampling without introducing aliasing. This will show you
roughly what the image will look like when actually printed without
exaggerating the granularity of the image.

 

[Neil Gould]

Hi,

No, not yet. I´m observing the scan on the monitor! I don´t really
understand what you mean.

The best way to know the quality of your scan will be to get a
photographic print made. If artifacts exist, they'll show up in the print,
and you'll have a basis for correcting those problems.

Viewing a scan on a monitor is only useful if you have a good quality
monitor and video card capable of displaying tonal transitions well (I
won't say "accurately", because the color model for transmissive and
reflective viewing are different). These are not commonplace purchases,
and as you've reported visible "posterization" of transitions, I suspect
that this may be part of your problem.

Regards,

 

[Neil Gould]

Gaaaah!!! Not again! Go back to the start fo the thread and read all
my posts again!

Recorded image resolution is likely to be less than the granular
resolution, so no, you can't just examine a high contrast edge and
looks for jaggies - even if they aren't there, there could still be
grain aliasing. Furthermore, jaggies are not an indication of
aliasing, merely a consequence of a square pixel output filter
representation!

I was *not* referring to "jaggies", though that may not make a difference
to your campaign, Kennedy. In an attempt to avoid such
misinterpretation -- an effort that apparently failed on you -- I stated
"obviously not to the pixel level". A salient question might be, "Why look
for artifacts at the edges of high-contrast portions of an image?", as
grain aliasing can have representation in any part of the image. But,
rather than ask such questions, you only jump to conclusions and rant on.

You've heard it all before, had the details spelled out in single
syllables and complained that you undertand it all, yet you still
cling to this untruth - so what's the point?

I'm wondering the same thing... though it's becoming clearer why we're the
only two still hashing about with this.

 

[Jytzel]

That comment makes me think you are viewing it on a monitor, and judging it as noise. A smooth
Histogram should indicate a smooth tonal transition. The reasoning behind drum scanning is to
eventually print the image. I think at this point, you should do a test print, proof print, or
match print, and then make a better judgement.


All monitors are such poor resolution in comparison to printing, especially commercial offset
printing. I have seen many image files that seemed noisy on a monitor, yet printed very
smoothly. You could easily be running into a limitation of viewing on a monitor.

If you can do a test print, then you would have a better item to judge scan quality. If it
still looks bad, then the original scan is at fault.

If you still find that after viewing printed results that things are not working, it is then
down to operator error, or a weird technical problem. I remember working on one workstation
that showed unusual results on most scans. What we finally traced that down to was an extra
monitor interfering with the SCSI cable of the scanner, and causing strange noise issues. The
reason that was so tough to track was that the monitor was a secondary monitor, and not always
used at that workstation. Routing the SCSI cable further from that monitor solved the unusual
noise problem. While I would be surprised if that is the problem you are having, if nothing
else is working, then investigate that direction. Best of luck.

Ciao!

Gordon Moat
A G Studio
<http://www.allgstudio.com>

Thanks Godon,
But why don´t all photos show noise then? The ones scanned from
positives are perfect. If it was just a monitor limitation then it
would affect all scanned images, right?

J

 

[Jytzel]

...or a noisy one - as Jytzel has described. Noise will only show up on
a histogram as a distribution around what should be perfectly uniform
areas - but few real world images have such areas so noise, to all
intents and purposes, is not observable from histogram views.

However, I agree with your concern that this is simply a case of viewing
on a monitor. Essentially there are two extreme options:
* viewing at 1:1 on the monitor pixel scale, which will result in the
image appearing soft and grainy, just due to the magnification,
* viewing at actual print scale, in which case the image will look
grainy due to aliasing due to the simple downsampling algorithm for
display.

Jytzel, if you want to view the image effectively on the monitor for a
fairly valid comparison with an actual print then resample the image to
your monitor size, do not just zoom it. For example, a 35mm frame
scanned at 4000ppi is around 5500x3700pixels. My monitor display is
1600x1200pixels, so I would resize the image to 25% of the original, to
get 1375x925pixels.

If you implement the resize in Photoshop or PSP then some filtering is
implemented prior to the resize to reduce aliasing, however it can be
worth additional filtering before the resize to be sure that no aliasing
occurs. In Photoshop, the best way of achieving this is to use gaussian
blur with a radius no more than 2pixels for a resize to 25% - ie. half
the resize factor. After the resize, implement an unsharp mask at
around 150% with a radius of 0.7 pixels to recover the loss in sharpness
caused by downsampling without introducing aliasing. This will show you
roughly what the image will look like when actually printed without
exaggerating the granularity of the image.

thanks for the tip

 

[Jytzel]

Yes, send me a section, but I am going off on business for a week, so
won't be able to comment on it till I get back.

thanks, send me your email address
(e-mail address removed)

 

[Paul Schmidt]

Thanks Godon,
But why don´t all photos show noise then? The ones scanned from
positives are perfect. If it was just a monitor limitation then it
would affect all scanned images, right?

I think there has been some good information in this thread, some things
that need to be done as part of the process:

Print the scan at 8 x 10
Get a chemical print made, also at 8 x 10

anything that appears in the digital print, that isn't in the chemical
print, is noise introduced by scanning. Often it's easier for a scanner
operator working at a photo shop or service bureau to fault the
process, rather then take the time to adjust and repeat, until they get
it right.

Paul

 

[Gordon Moat]

. . . . .Thanks Godon,
But why don´t all photos show noise then? The ones scanned from
positives are perfect. If it was just a monitor limitation then it
would affect all scanned images, right?

J

Hard to give you a good answer without seeing the differences, but here is an attempt. I have
usually seen more problems with negatives than positives in scanning and viewing. The scanners do a
conversion of the negative issue, and are usually functionally better performing with positives. It
could be in the conversion, noise was introduced into the negative scan. Obviously, other factors
during the scan could still be issues.

Just another guess at this, but I have had better luck scanning B/W negative films as colour
positive, and then inverting in PhotoShop or LivePicture. It seems that the more subtle tonalities
get picked up better, compared to setting the scanner at the colour or monochrome negative
settings.

Ciao!

Gordon Moat
A G Studio
<http://www.allgstudio.com>