From: Wilba on
Floyd L. Davidson wrote:
> Wilba wrote:
>> Floyd L. Davidson wrote:
>>>> Wilba wrote:
>>>>>
>>>>> In everyday raw shooting, you can be quite relaxed about having a
>>>>> small spike on the right of the histogram - that's actually a good
>>>>> thing.
>>>
>>> It is, assuming you want to clip the highlights.
>>
>> The highlights are only clipped in the JPEG. In my experience (with my
>> gear, my settings, my subjects), I have up to two more steps available
>> in the raw.
>
> Probably correct. If you want an *accurate* histogram,
> do *not* bother with setting in camera contrast
> adjustements though. Set the White Balance.
>
> Then you can actually *know* when the raw data is
> clipping rather than only know it is somewhere close.

I've tried that according to instructions referred to in this thread and
elsewhere, and it doesn't work for me (in both senses of "work for me").

> Incidentally, your "two more steps" is really
> interesting, as most people see something between 1/2
> and 1 fstop. That doesn't mean you are wrong, but it
> does mean that the light or the adjustments you use
> might be "unusual".

How would you test it?

>>> Otherwise, it's a fatal error, because there is no way
>>> to recover the lost data.
>>
>> Just underxpose the raw in PP. It's only when the the raw values
>> are clipped that you can't recover. It sounds like you've never
>> actually tried this.
>
> Since that is exactly what I described, I can't see why
> anyone would make such a statement.

No, you description was in the context of an uncalibrated camera. If you
meant working with an accurate histogram you needed to specify that.

>>> On the other hand, if you
>>> back off on exposure just enough to make sure no
>>> highlights are clipped, the loss of 1/2 to 1 stop of
>>> dynamic range probably will mean absolutely nothing
>>> (given that you won't print out anything that can show
>>> more than 5 or 6 fstops and the camera will almost
>>> certainly be recording 8 or more).
>>
>> Conversely, if you clip 1/2 to 1 step in the JPEG it only takes 5 seconds
>> to
>> fix in PP, so you don't have to get hung up on getting a perfect exposure
>> in
>> the field. And that "apparent" overexposure will give you more DR.
>
> That lost dynamic range usually was of no value at all.

Not in scene with more range than your sensor can record.

> You almost certainly clip it (by setting the black
> point) in post production.

Only if you do.

> If it clips in the JPEG and it turns out that for that
> particular light and camera adjustments the raw data is
> *also* clipped, it is lost forever. Short of
> specifically setting White Balance to make the histogram
> accurate, you just do *not* *know* if the raw data is
> being clipped or not.

Correct.

> That is why I'm suggesting that, even with a known
> inaccurate histogram that will often show red or blue
> clipping that is not in the raw data, your best "rule of
> thumb" generality to to assume the histogram is accurate
> and only allow clipping that is acceptable (specular
> highlights, etc).

Or know what your headroom is, and don't sweat it if you're well within it.


From: Floyd L. Davidson on
"Wilba" <usenet(a)CUTTHISimago.com.au> wrote:
>Floyd L. Davidson wrote:
>> Wilba wrote:
>>> Floyd L. Davidson wrote:
>>>> Wilba wrote:
>>>>> Floyd L. Davidson wrote:
>>>>>> Wilba wrote:
>>>>>>> Porte Rouge wrote:
>>>>>>>>
>>>>>>>> I set my exposure to slide the histogram to the right, without
>>>>>>>> clipping
>>>>>>>> ( when I have time), to capture the most tonal levels . So, now when
>>>>>>>> I
>>>>>>>> am editing the photos they are over exposed(not clipped). A sunrise
>>>>>>>> is
>>>>>>>> a good example. The deep colors are washed out. The obvious fix(to
>>>>>>>> me
>>>>>>>> anyway) in Lightroom or CS4 is to reduce the exposure. Now my
>>>>>>>> question
>>>>>>>> is, by reducing exposure in post, am I just ending up in the same
>>>>>>>> place
>>>>>>>> (histogram to the left) as if I had just ignored the histogram when
>>>>>>>> I
>>>>>>>> was shooting and set the exposure to properly expose the image using
>>>>>>>> my light meter? I guess in short I am asking if Lightroom or CS4
>>>>>>>> loses
>>>>>>>> tonal values when you reduce exposure in editing.
>>>>>>>
>>>>>>> Yes, reducing the exposure in Lightroom just puts your histogram back
>>>>>>> to
>>>>>>> the left, so that's a waste of time.
>>>>>>
>>>>>> Yes for the histogram, but it is not a waste of time.
>>>>>>
>>>>>> Increasing the exposure adds signal, but the noise stays
>>>>>> the same. In other words, if you take a shot of an
>>>>>> object that is all grey, with nothing approaching
>>>>>> "white" at all in the entire scene, you could
>>>>>>
>>>>>> 1) Expose to produce an histogram that matches
>>>>>> the scene, showing all values to be in the
>>>>>> middle of the graph. This will produce an
>>>>>> accurately exposed JPEG too.
>>>>>>
>>>>>> or,
>>>>>>
>>>>>> 2) Expose to produce an histogram with the data
>>>>>> pushed so far to the right that it is almost,
>>>>>> but not quite, clipping. The JPEG produce
>>>>>> will be "overexposed", and instead of grey
>>>>>> the scene will be white.
>>>>>>
>>>>>> But we want an image made from the camera raw data, not
>>>>>> the JPEG. And note that with either of the above
>>>>>> methods the "noise" in the data will be the same. The
>>>>>> Signal to Noise Ratio (SNR) for method #2 will be higher
>>>>>> because the signal is higher.
>>>>>>
>>>>>> The final image is made using the RAW converter, or an
>>>>>> image editor like Photoshop after conversion, to reduce
>>>>>> the whites down to grey. And the same process will
>>>>>> *equally* reduce the noise. Thus the noise in the final
>>>>>> image will be *lower* with method 2 than with method 1.
>>>>>>
>>>>>> Reducing the amount of noise in the final product is
>>>>>> probably *not* a waste of time.
>>>>>
>>>>> You're the only one who has suggested otherwise. This fibre is about
>>>>> whether reducing the exposure in post-processing is a good way to
>>>>> recover saturation. Please pay attention to the conversation.
>>>>
>>>> "Recover saturation" means what? Nobody has said any
>>>> such thing. *You* need to pay attention.
>>>
>>>Porte Rouge said (underlines added to emphasise the seminal phrases), "I
>>>set
>>>my exposure to slide the histogram to the right, without clipping ( when I
>>>have time), to capture the most tonal levels . So, now when I am editing
>>>the
>>>photos they are over exposed(not clipped). A sunrise is a good example.
>>>_The_deep_colors_are_washed_out_. The obvious fix(to me anyway) in
>>>Lightroom
>>>or CS4 is to reduce the exposure. Now my question is,
>>>_by_reducing_exposure_in_post_,_am_I_just_ending_up_in_the_same_place_(_histogram_to_the_left_)
>>>as if I had just ignored the histogram when I was shooting and set the
>>>exposure to properly expose the image using my light meter?"
>>>
>>>> The statement (quoted above) was made that, "Yes,
>>>> reducing the exposure in Lightroom just puts your
>>>> histogram back to the left, so that's a waste of time."
>>>
>>>Right, a precise and appropriate answer to Porte's question.
>>
>> As I noted in my original response, that is not a correct
>> statement. It is *not* a waste of time. It does more than
>> "just puts your histogram back".
>>
>>>> That is not true. It does more than just moving the
>>>> "histogram back to the left", and what it does do is
>>>> reduce noise, which is not usually a waste of time.
>>>
>>>I can see I'm going to have to insult you again very soon. :- )
>>
>> That won't help your problem though...
>>
>> Here is the part you missed the significance of:
>>
>> "_The_deep_colors_are_washed_out_. The obvious fix(to
>> me anyway) in Lightroom or CS4 is to reduce the
>> exposure. Now my question is, _by_reducing_exposure_in_
>> _post_,_am_I_just_ending_up_in_the_same_place_(_histogram_to_the_left_)
>> as if I had just ignored the histogram when I was
>> shooting and set the exposure to properly expose the
>> image using my light meter?"
>>
>> The "obvious fix" clearly states that in terms of
>> restoring the color "saturation" (which is actually
>> *not* what it is doing), it works. That was *not* the
>> question he is asking at all, as you seem to think.
>>
>> The question was does it do anything else? Why not just
>> set the in-camera exposure lower? Does it make any
>> difference?
>
>Ah, I see where you've got it all twisted out of shape.

The fact is, I read it correctly and you didn't.

>> The correct answer is that yes it does make other differences.
>
>No-one has ever disputed that. If only it was somehow relevant. :- )

You sure argue about it though. Why?

--
Floyd L. Davidson <http://www.apaflo.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska) floyd(a)apaflo.com
From: Wilba on
Floyd L. Davidson wrote:
> Wilba wrote:
>> Floyd L. Davidson wrote:
>>> Wilba wrote:
>>>> Floyd L. Davidson wrote:
>>>>> Wilba wrote:
>>>>>> Floyd L. Davidson wrote:
>>>>>>> Wilba wrote:
>>>>>>>> Porte Rouge wrote:
>>>>>>>>>
>>>>>>>>> I set my exposure to slide the histogram to the right, without
>>>>>>>>> clipping
>>>>>>>>> ( when I have time), to capture the most tonal levels . So, now
>>>>>>>>> when
>>>>>>>>> I
>>>>>>>>> am editing the photos they are over exposed(not clipped). A
>>>>>>>>> sunrise
>>>>>>>>> is
>>>>>>>>> a good example. The deep colors are washed out. The obvious fix(to
>>>>>>>>> me
>>>>>>>>> anyway) in Lightroom or CS4 is to reduce the exposure. Now my
>>>>>>>>> question
>>>>>>>>> is, by reducing exposure in post, am I just ending up in the same
>>>>>>>>> place
>>>>>>>>> (histogram to the left) as if I had just ignored the histogram
>>>>>>>>> when
>>>>>>>>> I
>>>>>>>>> was shooting and set the exposure to properly expose the image
>>>>>>>>> using
>>>>>>>>> my light meter? I guess in short I am asking if Lightroom or CS4
>>>>>>>>> loses
>>>>>>>>> tonal values when you reduce exposure in editing.
>>>>>>>>
>>>>>>>> Yes, reducing the exposure in Lightroom just puts your histogram
>>>>>>>> back
>>>>>>>> to
>>>>>>>> the left, so that's a waste of time.
>>>>>>>
>>>>>>> Yes for the histogram, but it is not a waste of time.
>>>>>>>
>>>>>>> Increasing the exposure adds signal, but the noise stays
>>>>>>> the same. In other words, if you take a shot of an
>>>>>>> object that is all grey, with nothing approaching
>>>>>>> "white" at all in the entire scene, you could
>>>>>>>
>>>>>>> 1) Expose to produce an histogram that matches
>>>>>>> the scene, showing all values to be in the
>>>>>>> middle of the graph. This will produce an
>>>>>>> accurately exposed JPEG too.
>>>>>>>
>>>>>>> or,
>>>>>>>
>>>>>>> 2) Expose to produce an histogram with the data
>>>>>>> pushed so far to the right that it is almost,
>>>>>>> but not quite, clipping. The JPEG produce
>>>>>>> will be "overexposed", and instead of grey
>>>>>>> the scene will be white.
>>>>>>>
>>>>>>> But we want an image made from the camera raw data, not
>>>>>>> the JPEG. And note that with either of the above
>>>>>>> methods the "noise" in the data will be the same. The
>>>>>>> Signal to Noise Ratio (SNR) for method #2 will be higher
>>>>>>> because the signal is higher.
>>>>>>>
>>>>>>> The final image is made using the RAW converter, or an
>>>>>>> image editor like Photoshop after conversion, to reduce
>>>>>>> the whites down to grey. And the same process will
>>>>>>> *equally* reduce the noise. Thus the noise in the final
>>>>>>> image will be *lower* with method 2 than with method 1.
>>>>>>>
>>>>>>> Reducing the amount of noise in the final product is
>>>>>>> probably *not* a waste of time.
>>>>>>
>>>>>> You're the only one who has suggested otherwise. This fibre is about
>>>>>> whether reducing the exposure in post-processing is a good way to
>>>>>> recover saturation. Please pay attention to the conversation.
>>>>>
>>>>> "Recover saturation" means what? Nobody has said any
>>>>> such thing. *You* need to pay attention.
>>>>
>>>>Porte Rouge said (underlines added to emphasise the seminal phrases), "I
>>>>set
>>>>my exposure to slide the histogram to the right, without clipping ( when
>>>>I
>>>>have time), to capture the most tonal levels . So, now when I am editing
>>>>the
>>>>photos they are over exposed(not clipped). A sunrise is a good example.
>>>>_The_deep_colors_are_washed_out_. The obvious fix(to me anyway) in
>>>>Lightroom
>>>>or CS4 is to reduce the exposure. Now my question is,
>>>>_by_reducing_exposure_in_post_,_am_I_just_ending_up_in_the_same_place_(_histogram_to_the_left_)
>>>>as if I had just ignored the histogram when I was shooting and set the
>>>>exposure to properly expose the image using my light meter?"
>>>>
>>>>> The statement (quoted above) was made that, "Yes,
>>>>> reducing the exposure in Lightroom just puts your
>>>>> histogram back to the left, so that's a waste of time."
>>>>
>>>>Right, a precise and appropriate answer to Porte's question.
>>>
>>> As I noted in my original response, that is not a correct
>>> statement. It is *not* a waste of time. It does more than
>>> "just puts your histogram back".
>>>
>>>>> That is not true. It does more than just moving the
>>>>> "histogram back to the left", and what it does do is
>>>>> reduce noise, which is not usually a waste of time.
>>>>
>>>>I can see I'm going to have to insult you again very soon. :- )
>>>
>>> That won't help your problem though...
>>>
>>> Here is the part you missed the significance of:
>>>
>>> "_The_deep_colors_are_washed_out_. The obvious fix(to
>>> me anyway) in Lightroom or CS4 is to reduce the
>>> exposure. Now my question is, _by_reducing_exposure_in_
>>>
>>> _post_,_am_I_just_ending_up_in_the_same_place_(_histogram_to_the_left_)
>>> as if I had just ignored the histogram when I was
>>> shooting and set the exposure to properly expose the
>>> image using my light meter?"
>>>
>>> The "obvious fix" clearly states that in terms of
>>> restoring the color "saturation" (which is actually
>>> *not* what it is doing), it works. That was *not* the
>>> question he is asking at all, as you seem to think.
>>>
>>> The question was does it do anything else? Why not just
>>> set the in-camera exposure lower? Does it make any
>>> difference?
>>
>> Ah, I see where you've got it all twisted out of shape.
>
> The fact is, I read it correctly and you didn't.
>
>>> The correct answer is that yes it does make other differences.
>>
>> No-one has ever disputed that. If only it was somehow relevant. :- )
>
> You sure argue about it though. Why?

I haven't argued about it at all, neither has anyone else. You're reading
things that aren't there. We've all been talking about very different
things. Seriously!


From: Floyd L. Davidson on
"Wilba" <usenet(a)CUTTHISimago.com.au> wrote:
>Floyd L. Davidson wrote:
>> Wilba wrote:
>>> Floyd L. Davidson wrote:
>>>>> Wilba wrote:
>>>>>>
>>>>>> In everyday raw shooting, you can be quite relaxed about having a
>>>>>> small spike on the right of the histogram - that's actually a good
>>>>>> thing.
>>>>
>>>> It is, assuming you want to clip the highlights.
>>>
>>> The highlights are only clipped in the JPEG. In my experience (with my
>>> gear, my settings, my subjects), I have up to two more steps available
>>> in the raw.
>>
>> Probably correct. If you want an *accurate* histogram,
>> do *not* bother with setting in camera contrast
>> adjustements though. Set the White Balance.
>>
>> Then you can actually *know* when the raw data is
>> clipping rather than only know it is somewhere close.
>
>I've tried that according to instructions referred to in this thread and
>elsewhere, and it doesn't work for me (in both senses of "work for me").

As noted in another article, I highly suspect your
camera needs factory level calibration.

>> Incidentally, your "two more steps" is really
>> interesting, as most people see something between 1/2
>> and 1 fstop. That doesn't mean you are wrong, but it
>> does mean that the light or the adjustments you use
>> might be "unusual".
>
>How would you test it?

Well, first you'd need to have a camera that is known to
work correctly. Second you'd need to be able to follow
technical discussions. Third you would have to find
someone who does not take offence at gratuitous insults
from incompetant people who cannot deal properly with
differences of opinion.

Clear enough?

>>>> Otherwise, it's a fatal error, because there is no way
>>>> to recover the lost data.
>>>
>>> Just underxpose the raw in PP. It's only when the the raw values
>>> are clipped that you can't recover. It sounds like you've never
>>> actually tried this.
>>
>> Since that is exactly what I described, I can't see why
>> anyone would make such a statement.
>
>No, you description was in the context of an uncalibrated camera. If you
>meant working with an accurate histogram you needed to specify that.

You have quite an extensive list of weasel words to
excuse your lack of comprehension.

>>>> On the other hand, if you
>>>> back off on exposure just enough to make sure no
>>>> highlights are clipped, the loss of 1/2 to 1 stop of
>>>> dynamic range probably will mean absolutely nothing
>>>> (given that you won't print out anything that can show
>>>> more than 5 or 6 fstops and the camera will almost
>>>> certainly be recording 8 or more).
>>>
>>> Conversely, if you clip 1/2 to 1 step in the JPEG it only takes 5 seconds
>>> to
>>> fix in PP, so you don't have to get hung up on getting a perfect exposure
>>> in
>>> the field. And that "apparent" overexposure will give you more DR.
>>
>> That lost dynamic range usually was of no value at all.
>
>Not in scene with more range than your sensor can record.

What the sensor can record isn't the issue. It's a
matter of what you can use with the display mechanism (on
your monitor, on a print, etc.).

>> You almost certainly clip it (by setting the black
>> point) in post production.
>
>Only if you do.

Of course only if you do. And, the fact is that you *do*.

>> If it clips in the JPEG and it turns out that for that
>> particular light and camera adjustments the raw data is
>> *also* clipped, it is lost forever. Short of
>> specifically setting White Balance to make the histogram
>> accurate, you just do *not* *know* if the raw data is
>> being clipped or not.
>
>Correct.
>
>> That is why I'm suggesting that, even with a known
>> inaccurate histogram that will often show red or blue
>> clipping that is not in the raw data, your best "rule of
>> thumb" generality to to assume the histogram is accurate
>> and only allow clipping that is acceptable (specular
>> highlights, etc).
>
>Or know what your headroom is, and don't sweat it if you're well within it.

You need to listen carefully. You *don't* know what
your head room is. It changes with different scenes and
different lighting, and in particular if you use any
type of "auto" for white balance.

--
Floyd L. Davidson <http://www.apaflo.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska) floyd(a)apaflo.com
From: Wilba on
Floyd L. Davidson wrote:
>
> You need to listen carefully. You *don't* know what
> your head room is. It changes with different scenes and
> different lighting, and in particular if you use any
> type of "auto" for white balance.

What's a method that can be used to measure it?