Olympus-OM
[Top] [All Lists]

[OM] Re: What should a scanning program (not) do?

Subject: [OM] Re: What should a scanning program (not) do?
From: Moose <olymoose@xxxxxxxxxxxxxxx>
Date: Wed, 28 Dec 2005 16:55:25 -0800
Nils Frohberg wrote:

>I didn't get out shooting film much this year. In order to ease my bad
>conscience, I wrote a very elementary scanning application for my FS4000.
>Up to now, it can't do much more than scan stuff and write it into a TIFF
>file, invert a negative, and remove the orange film mask (& compute the
>orange mask from a clean CN strip). [Fortunately, the FS4000 is sane
>enough to speak real "scanner SCSI2", unlike many SCSI scanners.]
>  
>
Well, I certainly understand the attraction of writing software just for 
fun/learning. I've done it myself.

On the other hand, somebody beat you to this one. Ed Hamrick's VueScan 
started out years ago not far beyound what you are doing. By now, it is 
a sort of Swiss Army knife of scanner software.

>This is not supposed to be a fancy picture editing software. In fact, it
>doesn't have any graphical frontend. You basically tell it which frame to
>scan on your film holder, and the size/offset/dpi, and it goes to work.
>  
>
Although VueScan has a graphical interface, it will do what you 
describe, or scan any selected frames or all in a holder at once.

>The output is not meant to look nice, but rather to keep all information
>in the file. You have to use picture editing software to adjust color
>tint, gamma, etc. (why reinvent the wheel?)
>  
>
VueScan has a RAW scan option which simply saves the un processed data 
from the scanner in a TIFF file. I save these as the digital versions of 
my negs.

>The next thing I want to implement is getting the alpha channel filled
>with IR data, and enable multiple passes of a picture (also with
>different exposure setting in the scanner to fill the higher bits in the
>TIFF file.) 
>
I have found multiple passes, at the same or different exposures to be 
unnecessary with my FS4000.

>I also have to try to get better contrast out of CN, it is very dull after I 
>inverted it.
>  
>
I've posted about this several times recently in various ways. If you 
take the linear output from a scan of the full range of available data 
from CN film and just display or print it that way, it looks very dull. 
We are used to seeing output that has had the ends of the histogram 
compressed and/or clipped and the midtone contrast increased. That's 
what slide film does as a process, what automated prints of cn film do 
and what the software that comes with most scanners does in its default 
setting. When you start rolling your own, you have to make your own 
desicions about how and where in your workflow to make those 
adjustments. The important thing is to realize that all the data is 
there, it's just not optimally arranged yet. :-)

My choice of how to do it is twofold.

For films where I don't have a profile, I accept the flat looking scan 
and do it "my way" in PS. This is what you would have to do with your 
scanner software project. It's not as bad as it may sound after some 
learning curve. I have PS actions to do most of the standard things I 
need to do and I generally do them in new layers, so I can adjust layer 
trasnparency to fine tune the effect. As I mentioned recently, this is 
as effort I did a while ago with the FS2710 to make "something" of all 
shots on a roll (decided after, not before taking it). The down arrow on 
any image will take you to the original scan, and you can see how flat 
they all are <http://galleries.moosemystic.net/Garden04/index.htm>.

For films for which I have made ICC profilles, the profiles do most of 
the work automatically. Again, I've posted this before, but it clearly 
shows the difference between a scans without and with ICC profiles with 
a number of very different subjects. One of the powerful features of 
ViewScan is the ability to profiles films, scanners and printers 
<http://galleries.moosemystic.net/VuesProf/>..

>Does anybody have an idea for any other useful function? I.e., something
>that shouldn't be done in postediting? 
>
Not that I can think of right now, other than ICC profiling, which is 
beyond what you are undertaking now, I think.

>Or am I the only strange guy in
>town that wants be be sure that the raw data from the scanner is on disk
>to be messed around with by programs that were written for that exact
>purpose (Photoshop, PWP, aso..)?
>  
>
Not at all. I don't like to do postprocessing in the scanner app. Again, 
though, one consequence of this philosophy on your part is the dull 
looking raw output you mentioned. Applying contrast correction, curves, 
etc. in your model work flow is the work of the editor app.

My standard practice is to scan a roll to RAW scanner output files, then 
autoscan those files to reversed TIFFs, then "mess around with them" in 
PS. I do consider applying the ICC profile to be a proper part of the 
scanning process, though, rather than later, although it can come later 
in the workflow..

Moose


==============================================
List usage info:     http://www.zuikoholic.com
List nannies:        olympusadmin@xxxxxxxxxx
==============================================

<Prev in Thread] Current Thread [Next in Thread>
Sponsored by Tako
Impressum | Datenschutz