Correcting scan data

  • Thread starter Thread starter Steven
  • Start date Start date
S

Steven

Hi,

I am trying to improve the quality of scans from a Canon FS4000 using
homemade software that produces raw 48-bit scans. My problem is the
quality of scans of negatives. Even if I apply curves and other effects
using Photoshop I can't get a result anywhere near the quality produced
by Silverfast or a mini-lab print.

Can anyone supply any details about matrix multiplication to correct the
colours for each pixel ? I'm not currently doing this and I don't know
if it is significant. Does anyone have an example set of matrix values
that I could try ?

Since negatives compress the input density range I presume I have to
expand a portion of the range and lose data above and below it. This
will increase the contrast for the range and force other samples toward
black or white. I am planning to select 1/4 of the captured range and
expand this portion. Does this seem reasonable ?

Thanks,
Steven
 
Steven said:
Hi,

I am trying to improve the quality of scans from a Canon FS4000 using
homemade software that produces raw 48-bit scans. My problem is the
quality of scans of negatives. Even if I apply curves and other effects
using Photoshop I can't get a result anywhere near the quality produced
by Silverfast or a mini-lab print.

Can anyone supply any details about matrix multiplication to correct the
colours for each pixel ? I'm not currently doing this and I don't know
if it is significant. Does anyone have an example set of matrix values
that I could try ?

Since negatives compress the input density range I presume I have to
expand a portion of the range and lose data above and below it. This
will increase the contrast for the range and force other samples toward
black or white. I am planning to select 1/4 of the captured range and
expand this portion. Does this seem reasonable ?

Thanks,
Steven
Seems that you are trying to re-invent the wheel.

Film scanners have the software to convert color negative film and deal
with the orange mask.
How do you think that you could write software better than the manufacturer?
 
Film scanners have the software to convert color negative film and deal
with the orange mask.
How do you think that you could write software better than the
manufacturer?

There is no such thing as best SW for sth
 
Steven said:
Hi,

I am trying to improve the quality of scans from a Canon FS4000 using
homemade software that produces raw 48-bit scans. My problem is the
quality of scans of negatives. Even if I apply curves and other effects
using Photoshop I can't get a result anywhere near the quality produced
by Silverfast or a mini-lab print.

I do not know if I can help you, but would you post a couple of images to
demonstrate what you mean? I assume you're using the manufacturer's driver
and your program works fine with positives right?
 
Seems that you are trying to re-invent the wheel.

Certainly. However, it is a very complicated wheel and I am trying to
understand it.
Film scanners have the software to convert color negative film and deal
with the orange mask.
How do you think that you could write software better than the manufacturer?

I doubt that I can but it is fun learning. This is why I am asking
about the significance of applying a matrix of co-efficients for colour
correction and also about the density compression that occurs with
negatives.

-- Steven
 
I do not know if I can help you, but would you post a couple of images to
demonstrate what you mean? I assume you're using the manufacturer's driver
and your program works fine with positives right?

I can't post anything useful at the moment. Scans with positives are
fine and I am trying to learn about techniques used when processing
negatives. My question about colour correction arose because I just
can't seem to get the colour clarity of a mini-lab print using curves,
etc. in Photoshop. Perhaps processing each pixel (using a matrix)
beforehand will help.

-- Steven
 
Back
Top