From: Floyd L. Davidson on
Doug McDonald <mcdonald(a)scs.uiuc.edu.remove.invalid> wrote:
>I have performed tests on my Canon 30D camera to see
>what happens if I expose less. I exposed 1/3 and 2/3
>stops less than before.
>
>At 2/3 stop less than before yes, even at contrast setting zero,
>I do see a teensy bit of space between the top of the red histogram
>and the top of the display. This means that if I'm willing to
>lose 2/3 stop of unclipped data at the bright end, Floyd is "effectively"
>right.
>
>But if I want that extra 2/3 stop, I'm still correct.

You have verified that you can't read a histogram, and
that what I've previously said is true. Changing
contrast isn't helping you a bit, but WB adjustment
would make your histogram more accurate, as well as make
your preview JPEG image greenish.

Your description suggests that at 0 EC your histogram
was indicating you should be clipping. The red channel
though, due to WB adjustment, is typcially about 1/2 to
3/4 of an fstop higher in the JPEG than it is in the raw
data. (Look in the Exif data and to find out what the
red channel multiplier is, and you can calculate exactly
what the fstop value differential is.)

To *correct* the inaccuracy of your histogram you can
adjust the WB to be more greenish. That will reduce the
red channel in the JPEG to agree with the raw data.

Changing contrast does *not* change the histogram's
representation of exposure, and therefore *cannot* make
the histogram more accurate. Only WB adjustment will do
that for you.

>I should add that my method works --- I use -4 in the camera,
>use the histogram, and find my raw images are very nicely exposed,
>no clipping (except as desired for specular highlights) and the
>data right up to the clipping level.

All you are doing is misunderstanding the histogram, and
because the WB adjustment gives you 1/2 to 3/4 of an
fstop of leeway, you end up with useful results. It is
very unlikely that you have the technical expertize to
even determine how close your raw data is to clipping,
never mind being able to see any difference in losing
that amount of dynamic range at the low end of the
exposure range as a result of the inaccurate histogram.

--
Floyd L. Davidson <http://www.apaflo.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska) floyd(a)apaflo.com
From: DRS on
"Floyd L. Davidson" <floyd(a)apaflo.com> wrote in message
news:87pr8xpx5u.fld(a)apaflo.com
> "DRS" <drs(a)removethis.ihug.com.au> wrote:

[...]

>> Yes, they do, in terms of *accuracy*.
>
> You simply do not know how to read a histogram, and until you
> learn there is little point in talking to you on any topic
> that requires it.

According to you. Not according to everybody else. You maintain you're the
only one here who understands what's going on but that's not supported by
the evidence so I'm quite happy to let you drift off in your fantasy world.



From: Ray Fischer on
Floyd L. Davidson <floyd(a)apaflo.com> wrote:
>Doug McDonald <mcdonald(a)scs.uiuc.edu.remove.invalid> wrote:
>>Floyd L. Davidson wrote:
>>
>>> Hence if the right edge is set correctly with high
>>> contrast or low contrast, either way it is exactly the
>>> same exposure and neither is more accurate than the
>>> other.
>>> You continue to say that low contrast makes it easier
>>> to
>>> see where the edge is, but that is only true for special
>>> cases, and for an equal number of special cases high
>>> contrast would make it easier!
>>>
>>
>>In NO CASE would high contrast make it easier ...
>>at least on a Canon 30D.
>>
>>Mr. Davidson: Let me ask again: What is the serial
>>number of the Canon 30D you have checked this on.
>>
>>If you have not done so, SHUT UP.
>>
>>We admit that you are right if somebody using
>>a Canon 30D is willing to allow 2/3 stop more "slop"
>>than is really necessary.
>
>It does not appear that you understand how to read a
>histogram.
>
>Hmmm... tell me what you make of these two histograms:

You're trying to change the subject.

--
Ray Fischer
rfischer(a)sonic.net

From: Floyd L. Davidson on
"DRS" <drs(a)removethis.ihug.com.au> wrote:
>"Floyd L. Davidson" <floyd(a)apaflo.com> wrote in message
>news:87pr8xpx5u.fld(a)apaflo.com
>> "DRS" <drs(a)removethis.ihug.com.au> wrote:
>
>[...]
>
>>> Yes, they do, in terms of *accuracy*.
>>
>> You simply do not know how to read a histogram, and until you
>> learn there is little point in talking to you on any topic
>> that requires it.
>
>According to you. Not according to everybody else. You maintain you're the
>only one here who understands what's going on but that's not supported by
>the evidence so I'm quite happy to let you drift off in your fantasy world.

The evidence is rather extensive, and nothing I'm saying
is unique. Everyone who does understand it says
basically the same things...

Here is a tutorial that explains how to get an accurate
histogram.

<http://www.guillermoluijk.com/tutorial/uniwb/index_en.htm>

As for reading histograms, I don't know of any tutorials
that don't have at least some errors, but this one is
pretty good:

<http://www.luminous-landscape.com/tutorials/understanding-series/understanding-histograms.shtml>

This tutorial is a pretty good one for learning how to use ETTR:

<http://www.luminous-landscape.com/tutorials/expose-right.shtml>

If you want a lot of detail, do a google search on Iyla Borg and
"white balance" or "histogram".

--
Floyd L. Davidson <http://www.apaflo.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska) floyd(a)apaflo.com
From: Floyd L. Davidson on
rfischer(a)sonic.net (Ray Fischer) wrote:
>Floyd L. Davidson <floyd(a)apaflo.com> wrote:
>>Doug McDonald <mcdonald(a)scs.uiuc.edu.remove.invalid> wrote:
>>>Floyd L. Davidson wrote:
>>>
>>>> Hence if the right edge is set correctly with high
>>>> contrast or low contrast, either way it is exactly the
>>>> same exposure and neither is more accurate than the
>>>> other.
>>>> You continue to say that low contrast makes it easier
>>>> to
>>>> see where the edge is, but that is only true for special
>>>> cases, and for an equal number of special cases high
>>>> contrast would make it easier!
>>>>
>>>
>>>In NO CASE would high contrast make it easier ...
>>>at least on a Canon 30D.
>>>
>>>Mr. Davidson: Let me ask again: What is the serial
>>>number of the Canon 30D you have checked this on.
>>>
>>>If you have not done so, SHUT UP.
>>>
>>>We admit that you are right if somebody using
>>>a Canon 30D is willing to allow 2/3 stop more "slop"
>>>than is really necessary.
>>
>>It does not appear that you understand how to read a
>>histogram.
>>
>>Hmmm... tell me what you make of these two histograms:
>
>You're trying to change the subject.

*Narrowing* the subject.

It is exceedingly difficult to discuss histograms with
people who do not understand what a histogram shows and
how to read it.

Until we get all participants on the same sheet when it
comes to how to read a histogram there simply isn't much
point in discussing the other aspects of their use, eh?

--
Floyd L. Davidson <http://www.apaflo.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska) floyd(a)apaflo.com