So far I've found the Canon XT sensor to be a lot like using slide film.
I'll be using a spot meter to be sure the highlights are in the range. The
range seems to be +3 to -3 or -4.
The shadows really depend on what you want to do with them and how much
noise you can tolerate. If the values are going to be very low in the
print, the -4 is OK. Below that the noise gets too high for me. If I need
a good image below -3, I'll either bracket or find some other tricks to
capture the information.
By the way, the "window" in my old Pentax spot meter shows just -3 to +3
stops -- the very range that seems to be safe for the Canon sensor.
The longer version of the post:
> Here's a post from the Adobe link: http://www.adobeforums.com/cgi-
> bin/webx?14@...2v9Dp0.31@.3bb6a869.3bbb1172/24:
> Bruce Fraser wrote: ...
>
> "... If you underexpose, ... you're only
> capturing half the tones the camera can record, ...
Yes, and what I've found is that the quality of the image is probably better
in the top half than the bottom half of the light range. On the other hand,
the noise in the shadows gets progressively worse as you go to far down.
So, the better image quality appears to be up out of the shadows.
> Fraser continued: "Within limits, digital raw give you a huge degree
>of latitude in where you place the
> midtone, but those limits are imposed by where you set the highlight
> point. ... so the key decision you need to make when
> exposing for digital capture is where you want to set your
> highlight point. That's why I spot-meter on the highlights. ..."
And why I'll be doing the same thing.
This is really a lot like slide film. It's not that complicated, except
that the highlight cut-off point may be more sudden.
In comparing to slides, my experience is that while they may look great when
projected on a screen (I'm showing my age here), when I'd try to pull more
information from the shadows, while there may have been something there, the
grain and image quality was terrible.
I've used ground glass over the lens and going up and down the scale
manually to explore how the sensor captures different light values. I've
also take shots of resolution targets and printed paper in a controlled
setting to see how detail holds up in the highlights. While the nature of
the target makes a difference, in general I don't think highlights over +3
are very useful. They are mostly blown out, which is OK for spectral
highlights, but not if you need any detail out of them.
>
> Bruce continued: "Digital isn't panchromatic either-it's way more
> sensitive to red and IR than it is to blue ...."
Well, that may be true in the abstract, but it's not what I see in with the
Canon XT in my tests. What I see is that, regardless of the native
sensitivity of the cmos sensor, Canon filters over the top of it have simply
made a compromise between daylight and tungsten lighting. In outside shots,
reading from the left, dark side of the histogram, the RGB spikes of a
ground-glass, daylight, outside exposure are in that order -- R, G, and then
B being the brightest. In an indoor, tungsten-light environment, the
histogram order is B, G and the R being the brightest. The extent of
displacement or offset of the lines is about the same with both. That is,
the extent to which they have to be moved by the raw processor to get
"white" (R=G=B) is about equal, but in the opposite direction when comparing
indoor and outdoor lighting. Canon's decision in designing its filters
makes perfect sense to me.
Rick Murai wrote:
> Due to their obvious cutoff I've found that the ND grads don't offer
> seamless control of high values...
I agree. I didn't use them when I shot slides, and I doubt I'll use them
with digital. Many love them, but it's not my cup of tea.
Paul
www.PaulRoark.com