Digital BW, The Print

Hi Steve, 

Comments inline...

-----Original Message-----
From: Steve Kale
Sent: 25 November 2004 15:16
To: DigitalBlackandWhiteThePrint@yahoogroups.com
Subject: Re: [Digital BW] Tonal range recording


Can I clarify some nits on this so that I have a clearer understanding?

[M] Sure!

<snip>

> With digital the contrast ratio of the sensor is fixed,

The gamma (contrast) of the sensor is not linear (or another way to put it,
the senor is linear) but we apply a contrast/gamma curve when converting from
RAW to RGB.  This curve need not be fixed.  Right?

[M] I think I see what you mean, but it's unconventional to talk about gamma
being linear when what you really mean is that there is a log curve involved
in the representation, so I'd avoid that. Look at it this way... Any image
file format represents pixel values in the form of numbers of some level of
precision. By saying that a given representation assumes some level of
"gamma", you are saying that the numbers must be interpreted on a log scale
rather than being directly proportional to the intensity. As long as you know
what the relationship is, you can get near-identical output on your monitor
from files that are encoded linear, in 2.2 gamma, or using any other value,
although some gamma values will make better use of the available numeric
precision than others. RAW files are very close to linear simply because of
the physical characteristics of the sensor. It would be possible to generate a
linear RGB TIFF from the RAW file but in practice this is done very rarely,
simply because the most commonly used colour spaces such as AdobeRGB and sRGB
are gamma 2.2 spaces.

> so the best you can do is make sure you set the exposure so that you 
> use the full sensitivity that's available,

I am getting there on this but see my other post.

>and then use curves in
> Photoshop to increase or reduce the contrast later.

Or better yet use a better transformation curve in the RAW processing stage?

[M] That would be better in the sense that it would provide slightly higher
precision. It might not be better in practical terms, because the curve
processing capabilities of the RAW converter are likely to be less flexible
and convenient than Photoshop's (you can't use masking to restrict changes to
part of the image, for example) and if there are any operations that you
wished to do before the curve manipulations, you'd be stuck. If you've chosen
a 2.2 gamma space as the target for your RAW conversion, the benefits of the
slightly improved numerical accuracy would be marginal anyway even in 8-bit,
and probably near-to-nonexistant in 16-bit.

If you really wanted to be obsessive about it, you could convert the RAW file
to linear 16-bit RGB, work on that in PS and then as a final step use Convert
to Profile with AdobeRGB as the destination space. That ought to retain the
highest possible precision, but my own experiments with linear RGB files in PS
have taught me that the display output routines are not really designed for
this and you get some unwelcome effects on the monitor. I have a strong
suspicion that the whole display output piece of PS is 8-bit, so your
beautifully precise 16-bit linear file values get chopped to 8 bits *before*
being fed into the conversion, leading to very ugly shadow rendering on the
display.

> If you don't succeed in
> using the full dynamic range (i.e., the histogram is bunched up down 
> the left-hand side) then if you choose to increase the shadow contrast 
> you'll see more noise than if you'd done a better job of the exposure.

Because noise is fixed and I have less observations to play with hence signal
to noise falls?  This is the bit I am currently chomping on with rusty math!

[M] Yeah, that's it in a nutshell. Think you're definitely getting there.

==============================================================================
This message is for the sole use of the intended recipient. If you received
this message in error please delete it and notify us. If this message was
misdirected, CSFB does not waive any confidentiality or privilege. CSFB
retains and monitors electronic communications sent through its network.
Instructions transmitted over this system are not binding on CSFB until they
are confirmed by us. Message transmission is not guaranteed to be secure.
==============================================================================