Dynamic range

[stamper]

Interesting article in this week's Amateur photographer uk about increasing dynamic range At the end of it he concludes

If I were photographing a landscape with the D80 and took a meter reading from the sky I could increase the recommended exposure by up to three and a half stops to recover detail in the foreground before the sky would totally "burn out" to white

I have that camera but the weather has been awful so I haven't tested the theory yet It seems to be rather ambitious though Any thoughts?

[Guillermo Luijk]

hehe this theory is as old as 2003: Expose to the right.

And of course it works, it's the basis of exposure in digital cameras. Exposing to the right means less noise, and less noise in the low f-stops (the dark areas of the image) means more DR as long as you don't burn the highlights.

[feppe]

I addition you could use contrast masking (good article on this very site on that) or HDR to extend the DR.

Or you could bother Guillermo above to finish his very promising-looking DR software  I'm sure I'm not the only one dying to see it in action!

[John Sheehy]

If I were photographing a landscape with the D80 and took a meter reading from the sky I could increase the recommended exposure by up to three and a half stops to recover detail in the foreground before the sky would totally "burn out" to white
[a href=\"index.php?act=findpost&pid=180244\"][{POST_SNAPBACK}][/a]

3.5 stops is a bit ambitious.  Most DSLRs clip in the green RAW channel about 3.5 stops above metered grey, so +3.5 can clip the green channel.  I'd stay down around +2.66 to +3, unless I tested the camera previously and knew it could do better.

[Guillermo Luijk]

I agree with John, 3.5 stops is too optimistic for most cameras (Fuji Super CCD will do without problem however, and maybe my 350D which has a strong trend to underexpose too). Check here some samples taken at a gray card from different camera brands and models: Log histograms of different cameras using camera light metering over a gray card. Most of them clip the G channel in some point of the last 3.5 f-stops.

This is the particular case of the D80: there is already info in the second half of the 4th f-stop (signed as -3EV):

 I'm sure I'm not the only one dying to see it in action![/quote]
hehe I admit I am being very lazy about it. Partially this is because another forum member showed a very elegant, simple and 100% effective way to implement image blending in PS conceptually flawless from the point of view of achieving what I intended: Yet another method to reduce noise with two exposures.

But I have come now with new ideas for optimum B&W pictures from RAW which can be combined with multiexposure blending, and probably in my Easter vacation will take that again.

BR

[Ray]

hehe this theory is as old as 2003: Expose to the right.

And of course it works, it's the basis of exposure in digital cameras. Exposing to the right means less noise, and less noise in the low f-stops (the dark areas of the image) means more DR as long as you don't burn the highlights.
[a href=\"index.php?act=findpost&pid=180253\"][{POST_SNAPBACK}][/a]

Guillermo,
The theory of expose-to-the-right is not new, but maybe this Amateur Photography article was trying to outline a procedure for achieving accurate ETTR.

It sounds similar to a method I sometimes used with my 5D. With camera in manual and spot meter mode, take a reading of the brightest part in the scene, whether it's a white cloud, a blue sky or a white shirt someone is wearing.

Use the exposure needle in the viewfinder as a guide for accurate exposure, ie. turn the exposure adjustment dial on top of camera till the needle is central. Then increase exposure by 3 stops and viola!, you've got the correct exposure for a full ETTR.

[Guillermo Luijk]

I know Ray, one thing is ETTR, and another thing is the method to achieve it. Anyway using spot metering over the brightest point of the scene and as a rule overexpose by +2EC, +3EC or whatever, is also an old idea.

The problem is that we always need to leave some error margin since the behaviour of camera's spot meter is according to luminance, but in a digital camera luminance is a combination of RGB channels. Think of a scene with a strong blue cast: B contributes less to luminance (or at least it should be that way) so we could be in danger to clip the B channel when using camera's spot meter since luminance will be much closer to the G and R levels than to B.

That is why there is no way today to achieve a perfect ETTR in just one shot (hope in the near future cameras solve this).

IMO the best thing we can do today to ETTR is:

1. Cancel camera WB (UniWB)
2. Have some stantard exposure compensation value in mind for our  particular camera
3. Apply a try-and-repeat tactique until the blinking areas displayed in the camera are just about to start to blow but not yet blown.