Olympus-OM
[Top] [All Lists]

[OM] Re: Slipping Closer (long)

Subject: [OM] Re: Slipping Closer (long)
From: Moose <olymoose@xxxxxxxxx>
Date: Wed, 19 Jul 2006 01:41:36 -0700
Rob Harrison wrote:
> Apologies--I know many of you have been through this already...but I'm just
> reaching this point... Every day it seems like I¹m slipping closer to moving
> to digital. This was the latest nudge.
>   
Digital does not automatically fix color management issues. Even the 
bodies with the best auto and pre-set WB options and performance will 
not regularly give you accurate color without help. You still have to 
have some kind of a reference between a controlled color and the 
lighting used. At a minimum, you need to either do custom WB with a 
controlled color diffuser over the lens or shoot extra shots with a 
WhiBal or some other reference in the shot.
> I'm trying to print up a 8.5 x 11 portfolio of photos for the local chapter
> of the AIA, for prospective clients to view in the AIA office. With that in
> mind I had a pro 
> photographer friend of mine stop by the other day for a lesson in color
> management, and learned that the literally thousands of jpegs I have from
> ³scanned at development² rolls of film (by a Fuji Frontier system, I think)
> are pretty much worthless in a serious color-managed workflow, since they
> didn¹t come with an embedded profile specific to the scanner that scanned
> them.
>   
As Chuck has pointed out, you can likely get an icc profile for the 
scanner, but that doesn't solve the problem, just one, smaller part of 
it. For full color management, you need icc profiles for the films 
involved. If you have mostly film which still is sold, this isn't a huge 
problem. as you can shoot reference frames.

You need to understand how this works to use it effectively. If, for 
example, you shoot an IT8 target in daylight and use the resulting 
profile, images with that profile applied will reflect the lighting 
under which they were shot. So a sunset shot, or an indoor shot in 
incandescent light will be very warm, and not reflect the true colors of 
the subject. If you also shoot one frame of an IT8 target under whatever 
light you are using for each subject and create profiles from them, your 
sunset, incandescent, fluorescent, etc. light shots will look like they 
were shot in daylight, without the special characteristics of the 
lighting. In fact, using this technique, there is no  need to use 
tungsten film.

The problem, as you can guess, is with any existing shots that were 
taken under lighting circumstances you can't now reproduce. It's still 
pretty easy to get good color balance in an editor if there is at least 
one bit of neutral colored subject in the frame, white, black, gray, 
whatever. In fact, at least in PS, and I assume in most other editors, 
if you have just one of a group of frames shot in the same light that 
has a nice neutral bit in it, you can correct all of them using the one 
reference. Even if the color isn't quite right, at least all of them 
will then match.
> I can _assign_ them or _convert_ them to a profile (Adobe 1998 he
> recommends) in PS7, and now that my monitors and printer are also profiled I
> can see what they're going to print like, but there may or may not be much
> connection to the colors of the original film...which has been my
> frustration: The color seems to be hit or miss.
>   
A little terminology problem here. Adobe 1998 and others are color 
'spaces', not source/device profiles, different animals for different 
porpoises. If an image is not color balanced, assigning a color space 
does nothing to help you. An icc color profile describes the mapping of 
color values in an image to standard colors.
> I could get a film scanner--the 5000ED with roll film adapter for
> example--and scan the film myself. Or I can hire my friend to do the
> scanning on his 900ED at about $15/scan for a 24MB+ scan, 
It's not quite clear what he is offering. A full frame, 16-bit scan on 
my 4000 dpi scanner is about 64mb. 24mb could be 2400 dpi at 8-bit. A 
9000ED scan of color neg film will capture MUCH more dynamic range that 
can be stored in 8 bits, so some kind of compression/clipping is 
necessary. Properly done, that can be fine, but you should be sure to 
have some samples done before committing to the project and up front money.

2400 dpi is probably fine for what you want. Anything over about 
3000-3200 dpi with most film just gives bigger images of the grain, eats 
up storage space and makes PS go slower.
> if I agree to do
> 150 scans over the next couple years. That is, if I pay him $2,000 up front
> he'll do 150 scans for me. In terms of actually getting the work done this
> seems liike a pretty good deal, since the major investment for me is not the
> hardware, but the time it will take to do the scanning--especially if I have
> one of my employees do it. I may want to invest in a film scanner as well,
> for my own personal explorations, but that's kind of a different issue.
>   
Either of these routes will take care of whatever resolution problems 
you have. They may also solve problems of blown highlights and or lost 
shadow detail that are common in cheap scans, as discussed above. By 
themselves, they won't solve your color accuracy issues. Without a 
reference, the only thing any system can do is to assume that the image 
it creates is of a subject with an average color of neutral gray. If 
that is true, or close to true, it can give you good color balance. If 
not, it won't work. Anybody with a professional reason to want accurate 
reproduction of subject colors can't use such a system.

To illustrate the problem with any kind of auto WB, I posted some 
examples before. The large image is the almost all green subject as 
scanned using VueScan with a daylight icc profile for the film used 
created using an IT8 target. Memory isn't perfect, of course, but this 
image looks like I remember the subject, a freshly painted, very green 
barn door in sunlight. The smaller images below are various results of 
scanning with out a profile and using auto WB in the scanning or in PS. 
As you roll the mouse over each, it shows what using auto WB does. As 
you can see with this extreme subject, auto WB is a disaster, and only 
an icc profile comes anywhere close to the original 
<http://www.moosemystic.net/Gallery/tech/WBalance/9-641_25wb.htm>.

I've also posted this several times, showing the difference between 
scans of a wide variety of subjects with and without a film icc profile 
<http://galleries.moosemystic.net/VuesProf/>.

 If I had it with me at the time, I could also have taken a shot with a 
calibrated target in the same light, even something as simple as a grey 
card, or the WhiBal, which gives white, gray and black references, and 
probably come close to the icc profile version.  
> Of course, I still have to _find_ the negatives....scattered over probably 50 
> different rolls....
>   
Can't help with that, I've got my own problems. :-)
> So anyway, Cameta has an E-1 kit with 14-54 lens and battery grip for $949,
> counting the $50 rebate on the lens.... I am sorely tempted..... But does it
> make any sense at all to invest in an E-1 at this point? 
No camera by itself will assure accurate color. You still will at least 
need something like a WhiBal. And remember to use it, my problem all too 
often. :-)
> I'm really liking the look of the Panas*nic L1, which seems to be basically 
> an E-330 with a Le*ca lens.
Yes and no. It is cooler looking and has more film camera like controls. 
But it gives up live view in the non-macro mode and has a fixed LCD, 
both of which seriously detract from it's capabilities from my 
perspective vs. the E-330. One can still get the Leica lens and use it 
on the E-330, if that's the issue.
> And the 5D is down to about $2500 for the body.... But I know I'm going to 
> want AF lenses, so it's looking like a new system could be on the horizon.
>   
The 5D is a really amazing imaging machine. It's not fair to compare it 
to much cheaper bodies, as far as basic image capture is concerned, It 
way outperforms them and costs way more. Of course, for many 
applications, all those super clean pixels are simply overkill. For only 
slightly cropped 8.5x11, you'll never see the difference for almost any 
use except low light, where it is exceptional. It also has it's own 
limitations. WB indoors is not very good. And the specialties of the 
E-330, live-view and the Mode-B macro, can do things the 5D can't.

And yes, most of us with them have also spent on AF lenses, so the total 
money commitment is not insignificant. I have 19-35 and 28-300 mm zooms 
and a 90/2.8 macro as my kit and a 24-85 as backup. By buying 3 of them 
used and staying away from 'L' glass, I've kept the total cost down and 
still get what I consider excellent quality results.

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