MFlenses - A new life in a digital age
MFLenses 2026

Manual Focus Lenses Forum

Explore the world of vintage lenses through repair notes, real test shots, and detailed archive threads packed with images.

Explore the world of vintage lenses through repair notes, real test shots, and detailed archive threads packed with images.
Spotlight

Featured partner placement for the forum front page.

Zones and exposure, for digital cameras
View previous topic :: View next topic  


Reply

Zones and Digital: Two Methods of Exposing

follow-on article

Peace in Lights

edited to remove the attention-getting but misleading "(ETTR, yes or no)" from the topic, which seems to have resulted in people missing the points made by the site entirely.

White Balance in Digital Cameras: Problems

Further edited to add the above link which better explains the effect of white balance on per-channel over or under exposure.


Last edited by ChrisLilley on Sun Jun 08, 2008 1:11 pm; edited 2 times in total


Reply

Thanks Chris. Fascinating stuff and it will take a while to digest. So does it debunk Reichmann's ETTR theory? I could not come to a conclusion based on the first article.


Embarassed Embarassed Embarassed patrickh


Reply

IMHO it is just the way Reichmann writes.
Some people just don't read him the way he means it.
I can't see that it changes his ETTR theory.
If you read one of his article (On Luminous landscape) he puts over all those points very well.
http://www.luminous-landscape.com/tutorials/expose-right.shtml
Shotting to the right (RAW only) is the correct way to expose for best results.
Of course shooting for the right does not work if you shoot JPEG or only use Default RAW conversion.

IMO you can't take anything Reichmann writes and stand it on its own.
He expects his readers to have a decent knowledge of the relative ideas.
It is easy to take a theory apart with other theories, however most if not all of what Reichmann writes isn't theory. It is practice.


Reply

Having re-read the luminous landscape article, it has some clear problems. It promotes using a single channel histogram and assumes that over or under exposure can be accurately judged from that. Since it can't, the rest of the theory is also suspect.

The problem is that even three channel histograms don't tell the whole story because they show white balanced values and thus fail to indicate the effective use of the raw counts (and thus, whether there will be noise issues).

The libraw folks have clearly a better technical grasp (as one would hope since they are writing raw decoders). Unfortunately the sort of info that they are examining is not available in-camera, only in post processing.


Reply

Rob Leslie wrote:
IMHO it is just the way Reichmann writes.
Some people just don't read him the way he means it.

I agree that many people seem to misunderstand what he says and in fact (by carefully protecting hilights) end up exposing to the left! Which gives them the noise problems mentioned.

Rob Leslie wrote:

Shotting to the right (RAW only) is the correct way to expose for best results.


I'm guessing based on that absolutist statement that you either didn't read or didn't understand the site I pointed out. Trying to judge whether any of the channels are actually blown, based on a histogram constructed from a white balanced image, will lead to substantial inaccuracies.

Whichis why people who understood the ETTR concept and overexposed to 'just avoid' clipping the hilights, may in fact find they are clipped in at least one channel.

Rob Leslie wrote:
Of course shooting for the right does not work if you shoot JPEG or only use Default RAW conversion.

(Neither of which would apply, so not sure why you mention it)


Reply

"Trying to judge whether any of the channels are actually blown, based on a histogram constructed from a white balanced image, will lead to substantial inaccuracies."

I don't understand that?
My camera gives me three channels (Seperate) and so does my RAW converter.
What am I missing? Where can I find it?
If I make allowances and move towards under exposure I am going to end up with far more 'Inaccuracies' and maybe a loss of information.
Is this one of those Chicken or Egg problems?
We have to accept that all exposure is a compromise based on the elements in a scene. I may sometimes move to the left if my subject matter requires it, or I may even move further to the right and let highlights blow if my main subject is in shadow.
Theory and Rules are IMO there for us to understand what is going on. in this respect my statement that exposing to the right should be 'Most of the time' rather than the only correct way.

But generally I will expose to the right with reference to my three camera histogram channels. I find 'Minor' blowing of a channel (The Pentax seems to like to blow Blue) isn’t a problem as it can be controlled in ‘Calibrate’ with a little adjustment of that individual Saturation or the Hue
I don't find the idea of exposing to the right to be just a theory. I have shot with perhaps one of the worse DSLRs for noise the Sony Alpha and I regularly shoot 1600asa with both my Pentax cameras. Even at 400asa one can see a vast noise difference between giving the highlights 'Headroom' and shooting right over. It even works on my Canon G5.

I think my point about Default RAW conversion does apply. I know of plenty of people who only see RAW as good lossless file rather than a 'Digital negative' and they only ever convert with the defaults then do adjustment in PS. Daft I know but there are plenty working that way.


Reply

Rob Leslie wrote:
"Trying to judge whether any of the channels are actually blown, based on a histogram constructed from a white balanced image, will lead to substantial inaccuracies."

I don't understand that?
My camera gives me three channels (Seperate) and so does my RAW converter.
What am I missing? Where can I find it?


In brief, what you are missing is that you believe the three channels to be independent and the red, green and blue values you see in tour raw converter software to relate directly to the values provided by the sensor. That isn't the case.

Your sensor gives a single channel per sensor cell, with a linear count from 0 to 4096 (for a 12 bit sensor; 0 .. 16,383 for a 14 bit sensor). If the count on a given sensor cell is at the maximum and more photons arrive during the exposure, the count will still be at maximum (and there may be a spill-over effect to adjacent sensor cells). This is overflow. These counts are the raw data from the sensor and is what should be contained in the raw image.

In a real world image, its likely that many individual sensor cells will show overflow, because the dynamic range of the sensor is far less than the range of real world lightness levels. This is unlikely to be noticed unless a group of contiguous sensor cells have all overflowed (and even then, perhaps not depending on other factsas explained below).

Digital camera software, and almost all raw converter software, does not give you direct access to this information at all.

To construct an RGB image (eg, for image preview in the camera):

- the two missing channels on each sensor cell are estimated by interpolation from nearby cells with the same color filter
- the values in each channel are scaled such that the chosen white point will be produced when the three channels have the same value (depending on the color space chosen, the values in each channel result will be a weighted sum of the three original, interpolated, channels)
- the values are then put through a non-linear transfer function such that black maps to 0, white maps to 255 and mid grey maps to somewhere in the 90 to 130 range depending on the contrast curve selected.

If your camera (or raw converter) gives you a three colour histogram, it is calculated *from that RGB image*. It is not constructed from the original raw data. If a lot of pixels in the calculated RGB image have overflowed then you may see a spike at 255 in one of the channels. on the other hand, this does not tell you whether the overflowed values are clumped together (giving a blown hilight) or are spread out. It does not tell you if some values had overflowed, but were then brought back into range during the interpolation or white balancing operations.

If you camera or raw converter gives you a single histogram then it is hopefully calculated as a weighted sum of the red green and blue channels in the derived image although sometimes, in fact its just the green channel with no input from the other two at all.

Rob Leslie wrote:
If I make allowances and move towards under exposure I am going to end up with far more 'Inaccuracies' and maybe a loss of information.

Yes, but then I never said that you should under expose. You are arguing against a position that I have not taken. Since ETTR is mentioned in the article, you seem to have taken the entire site as "ETTR is wrong" and then mounted a spirited defence of the position "ETTR is always and exactly correct in all cases". While such a black-and-white characterisation might do you credit in a debating society, it is rather answering a question that was not asked and it also seems to have lead you to not consider any merits of the articles in the site I indicated.

Rob Leslie wrote:

We have to accept that all exposure is a compromise based on the elements in a scene. I may sometimes move to the left if my subject matter requires it, or I may even move further to the right and let highlights blow if my main subject is in shadow.

Clearly.

Rob Leslie wrote:
Theory and Rules are IMO there for us to understand what is going on. in this respect my statement that exposing to the right should be 'Most of the time' rather than the only correct way.

But generally I will expose to the right with reference to my three camera histogram channels. I find 'Minor' blowing of a channel (The Pentax seems to like to blow Blue) isn’t a problem as it can be controlled in ‘Calibrate’ with a little adjustment of that individual Saturation or the Hue


Yes, sometimes you can get the lightness values of a blown highlight back. There will be hue shifts however. And sometimes the histograms will not indicate that overflow has in fact occurred but been masked by later processing (especially under light that is very different to natural daylight). Equally the histograms will sometimes show that the blue or red channels are using a lot of the available code space when in fact the actual raw counts only used a tiny fraction of that space.

Rob Leslie wrote:
I don't find the idea of exposing to the right to be just a theory.

I don't recall saying that it was. Once again, you seem to be arguing for a position that is not much in dispute, against an argument that has not been made, and ignoring the interesting details in the process.

Rob Leslie wrote:
I think my point about Default RAW conversion does apply.

Perhaps you missed that the site is created by people who are programming an improved raw converter. Any comments about the general public shooting JPEG, misunderstanding RAW, or using raw converters on default settings then hardly apply, to that site, because those people are not doing that.