Ad
Ad
Ad
Pages: « 1 2 [3] 4 5 »   Bottom of Page
Print
Author Topic: Underexposure or Overexposure?  (Read 18794 times)
standard_observer
Newbie
*
Offline Offline

Posts: 11


« Reply #40 on: January 07, 2008, 04:08:07 PM »
ReplyReply

Quote
On the other hand, a camera with large pixels and low read noise (e.g. Nikon D3 or Canon 1DMIII) can produce quite acceptable pictures at 4 stops underexposure as defined above. Overexpose by 4 stops and you will lose 3 or more stops of highlight detail and the results will most likely not be acceptable. However, you will get good shadow detail.

As Andrew stated, we should be striving for proper exposure. In an imperfect world, slight overexposure can be tolerated, but gross overexposure will cause major data loss. If you have a good camera, gross underexposure will give more noise and less dynamic range, but you will at least have an image. ...
[{POST_SNAPBACK}][/a]
Excellent summary.
 
 
Quote
Those, who keep only the 2048 levels in their eyes are partially blind.
...
Using only the first 2048 levels means, that the seventh stop will consists of only FOUR levels; going to the right increases this to eight levels. So, what about the 70 levels?
[a href=\"index.php?act=findpost&pid=165692\"][{POST_SNAPBACK}][/a]
Recommended homework: see Human Vision and Tonal Levels:
[a href=\"http://www.normankoren.com/digital_tonality.html]http://www.normankoren.com/digital_tonality.html[/url]

If of further interest: about the limitations of the Weber-Fechner Law:
http://www.jgp.org/cgi/reprint/7/2/235.pdf

DPL

--
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #41 on: January 07, 2008, 05:17:56 PM »
ReplyReply

Quote
This is correct only
1. if it is possible to expose to the right using the base ISO; but what about insufficient light, moving subjects hand-held shooting, the necessity to work with small aperture?
[{POST_SNAPBACK}][/a]

In these situations, you can't shoot fully to the right and your criticism is irrelevant.

Quote
2. if the base ISO offers the best DR, which is not always the case.
[a href=\"index.php?act=findpost&pid=165692\"][{POST_SNAPBACK}][/a]

Read my statement again. I said signal:noise, not dynamic range. The two usually go together, but in some cameras such as the Canon 1D MII, you get the best dynamic range at ISO 100 but the best S:N at ISO 50. See [a href=\"http://www.clarkvision.com/imagedetail/evaluation-1d2/index.html]Roger Clark[/url]. If you can find any examples to support your statement, please supply them.

Quote
Higher ISO is always better than lower ISO with the same exposure, adjusted in the raw processing, at least up to some ISO limit (and even later, it is never worse than increasing the brightness in raw processing).
This is woodoo science from a certain level, for the inaccuracy in the data used in the calculation is approaching 100%.
[{POST_SNAPBACK}][/a]

What's woodoo? Some figment of your imagination? How many peer reviewed scientific articles have you published?  Roger has over 200. I already gave you a reference here, so I won't comment further rather than note your assertions are without logic or supporting data.

Quote
Those, who keep only the 2048 levels in their eyes are partially blind.
[a href=\"index.php?act=findpost&pid=165692\"][{POST_SNAPBACK}][/a]

2048 levels are grossly in excess of what is needed by human vision or any reasonable tonal adjustments. For a discussion of recording efficiency see [a href=\"http://www.anyhere.com/gward/hdrenc/hdr_encodings.html] this article[/url] by Greg Ward.

Quote
However, there is another side of the issue: utilizing some of those 2048 levels means at the same time increasing the number of levels at the low end.

What worth is an additional "low noise" stop, which contains only eight levels?

Using only the first 2048 levels means, that the seventh stop will consists of only FOUR levels; going to the right increases this to eight levels. So, what about the 70 levels?
[{POST_SNAPBACK}][/a]

Again, please reread my original statement, "Posterization in the shadows could occur, but usually noise is the limiting factor there." I had already anticipated your and similar criticisms. If you can't make out the levels because of noise, then the number of indistinct levels is not an overriding consideration.

In addition, 8 levels in the darkest f/stop is usually sufficient as explained by [a href=\"http://www.normankoren.com/digital_tonality.html]Norman Koren[/url] because the eye can distinguish fewer levels in shadows.
« Last Edit: January 07, 2008, 06:14:17 PM by bjanes » Logged
jjj
Sr. Member
****
Offline Offline

Posts: 3197



WWW
« Reply #42 on: January 07, 2008, 05:44:24 PM »
ReplyReply

Quote
So, you set it for something in the sun; then, all of a sudden, you have a small window of opportunity to shoot something in the shade. Your manual setting may result in gross under-exposure, far more than 1 stop.  The gamble Edmund speaks of may under-expose a bit at times, but never as much as a fixed manual can.
Not true as you may have a subject that is backlit and then you are grossly underexposed. Whereas you can set a manual exposure at a compromise position if you think you may be shooting in light + shade. But I'd rather set the correct exposure for wherever I'm shooting. Besides on manual I can very easily adjust exposure up or down very easily in a fraction of a second. More easily than with auto exposure.

Quote
It's all a matter of what you have time to pay attention to in the type of shooting you're doing.  Opportunities don't always sit around and wait for you to meter and turn dials.[a href=\"index.php?act=findpost&pid=165536\"][{POST_SNAPBACK}][/a]
And that is why I use manual.  
Logged

Tradition is the Backbone of the Spineless.   Futt Futt Futt Photography
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #43 on: January 07, 2008, 05:48:05 PM »
ReplyReply

Quote
Exactly! Its about getting as many of the few levels in the last stop (shadows) possible.

And what the eye can theoretically see seems rather moot when we're talking about encoding data for a computer to process.
[a href=\"index.php?act=findpost&pid=165696\"][{POST_SNAPBACK}][/a]

As Milton Friedman once observed, "There is no such thing as a free lunch." Photography requires many compromises and achieving the best compromise for the intended visualization is part of the art of photography. If the dynamic range of the subject is approximately equal to that of the camera an exact exposure may capture the entire scene. You can bias the exposure slightly to protect the highlights or the shadows but you can't protect both. In most cases the highlights are more important. Of course, 14 bit ADCs could give more shadow levels, but thus far there has been little observable improvement because the obscuring effects of noise.

If the dynamic range of the subject exceeds that of the camera, you must make some choices.

If your pictures are to be viewed by a robot rather than a human, then your concluding remark may make some sense. However, if your audience consists of humans, then what can be seen is important. If you can use computer processing to bring out hidden detail so that it can be seen by the eye, fine, but otherwise your approach is naive and most likely counterproductive.
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #44 on: January 07, 2008, 05:58:27 PM »
ReplyReply

Quote
Excellent summary.
 
 

Recommended homework: see Human Vision and Tonal Levels:
http://www.normankoren.com/digital_tonality.html

If of further interest: about the limitations of the Weber-Fechner Law:
http://www.jgp.org/cgi/reprint/7/2/235.pdf

DPL

--
[{POST_SNAPBACK}][/a]

DPL,

Excellent information. Had I seen your post prior to making my own, the effort would have not been necessary. The JPL reprint (from the 1920s) is most informative, but will take some time to digest.

Here is a somewhat more accessible description of the limitations of the  [a href=\"http://www.neuro.uu.se/fysiologi/gu/nbb/lectures/WebFech.html]Weber Fechner Law[/url]. The Stevens law (described on the next page) is interesting because the slope of the curve for light is the flattest.

Indeed, Norman did note that the Weber-Fechner law was a first order approximation and that the eye is less sensitive to the darker levels, where 8 rather than 70 may be sufficient.
Logged
John Sheehy
Sr. Member
****
Offline Offline

Posts: 838


« Reply #45 on: January 07, 2008, 09:48:02 PM »
ReplyReply

Quote
Not true as you may have a subject that is backlit and then you are grossly underexposed. Whereas you can set a manual exposure at a compromise position if you think you may be shooting in light + shade. But I'd rather set the correct exposure for wherever I'm shooting. Besides on manual I can very easily adjust exposure up or down very easily in a fraction of a second. More easily than with auto exposure.

And that is why I use manual. 
[a href=\"index.php?act=findpost&pid=165752\"][{POST_SNAPBACK}][/a]

I'm not arguing that there aren't times when manual works best - it is both simple and optimum when the lighting is consistent, and your subject and background have high contrast that can vary intensity in the FOV with a sweep of recomposition.  I'm arguing that there are times when it doesn't work best; case in point - when you don't have time to deal with optimal exposure and must gamble, and both your subject and background will both have an opportunity to suddenly change by a few stops.  Reading Edmund in context, my immediate impression was one of a situation where anything can pop up instantly, and you don't have time to mess around with all parameters of shooting, so something must be simplified and sacrificed, and a -1 EC in an AE mode is the chosen sacrifice.  When you replied that you shoot manual, in that context, it seemed that you referred to one fixed manual setting, which included enough headroom for whites in the brightest light.

If your scenario is one where you take time for exposure issues, then I don't think it applies in that context, and if you are going to insist that you won't ever miss any opportunities while metering, I think you're probably overrating yourself.  However good you are at manually adjusting exposure to meet a relatively immediate need, there is always a more immediate need that you won't be ready for, and I assume that was the context here.  If you have to completely automate exposure, it may be safer to do so with -1 EC, which on cameras like the D3 and Canon mk3 cameras, is not a particularly big sacrifice at ISO 200s and 100, respectively.  I believe the D3 allows manual AV and Tv with auto-ISO, which is even better than AE, as you can keep the manual setting in a range of lighting.
Logged
John Sheehy
Sr. Member
****
Offline Offline

Posts: 838


« Reply #46 on: January 07, 2008, 10:06:37 PM »
ReplyReply

Quote
In these situations, you can't shoot fully to the right and your criticism is irrelevant.
Read my statement again. I said signal:noise, not dynamic range. The two usually go together, but in some cameras such as the Canon 1D MII, you get the best dynamic range at ISO 100 but the best S:N at ISO 50.
[a href=\"index.php?act=findpost&pid=165738\"][{POST_SNAPBACK}][/a]

It really makes one wonder why a company would ruin an opportunity to have both at one ISO, just to meet a standard of having "ISO 100" in some important hierarchy in the cameras' ISO range.  Why not just start at ISO 70?  Who would refuse to buy a camera because it had ISOs 70, 140, 280, 560, 1120 instead of what we have now?  You could even label it something standard like 64, 125, 250, etc or 80, 160, 320, etc (and get the "standard ones with the 1/3 stop fabricated ones anyway, and explain the real ISO sensitivity in the manual).  Almost every Canon DSLR is mildly crippled like this at the bottom of the ISO range.  The XTi is the only one that seems to have RAW saturation at its lowest ISO below sensor saturation (its real sensitivity is ISO 85 when it says ISO 100, but meters at 120, and under-exposes the RAW data - they had to screw something up!).
Logged
John Sheehy
Sr. Member
****
Offline Offline

Posts: 838


« Reply #47 on: January 07, 2008, 10:40:54 PM »
ReplyReply

Quote
Yes. And there IS such a thing as correct exposure, which is what ETTR is intending to accomplish.

http://www.digitalphotopro.com/tech/exposing-for-raw.html
[a href=\"index.php?act=findpost&pid=165621\"][{POST_SNAPBACK}][/a]

I have funny feelings about the phrase "correct exposure" in the context of ETTR.  The connotations of the word "correct" imply to me a context of something like shooting slide film, where you want the slide projected onto the screen to have "normal" intensity for the scene; a scene that is high key should be bery bright on the screen; a scene that is a black beetle on a black cloth should be fairly dark on the screen.  The same would apply to in-camera JPEGs displayed with a digital projector, or even regular film developed and printed with fixed parameters (no auto-levels).

People who do not understand the value of ETTR talk a lot about "correct exposure", and are usually referring to the act of literally metering a gray card, or some equivalent exposure technique.  And, to tell the truth, I really don't see anything wrong with equating the phrase to this.  This "no processing" paradigm needs to exist, even if we don't use it most of the time, out of personal choice, if only as a frame of reference.  IOW, a gray FOV, exposed correctly, will leave RAW highlights unused.  An exposure that leaves the brightest excursions just below RAW saturation is an "optimal exposure" (assuming, of course, that Av and Tv values are not compromised).  "Optimal exposure" is a phrase I am more comfortable with in the context of ETTR.
Logged
John Sheehy
Sr. Member
****
Offline Offline

Posts: 838


« Reply #48 on: January 07, 2008, 11:09:16 PM »
ReplyReply

Quote
how does one characterize these aspects eg S:N vs ISO ?
[a href=\"index.php?act=findpost&pid=165648\"][{POST_SNAPBACK}][/a]

I see too many ways to interpret your question.  Could you whittle it down a notch?
Logged
The View
Sr. Member
****
Offline Offline

Posts: 989


« Reply #49 on: January 08, 2008, 02:24:16 AM »
ReplyReply

AESTHETIC CONSEQUENCES OF EXPOSING TO THE RIGHT.

I have here a portrait, taken outside, exposed to the right.

Nothing blows, there isn't a single pixel blown.

But through this overexposure that I did exposing to the right, I somewhat changed the appearance of the skin. It looks drier.

Summary: exposing to the right may result in technically perfect images, but this overexposure changes the image characteristics. I found that skin not exposed to the right looked better than exposed to the right.
Logged

Deserts, Cities, Woods, Faces - View of the World.
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #50 on: January 08, 2008, 05:52:24 AM »
ReplyReply

Quote
AESTHETIC CONSEQUENCES OF EXPOSING TO THE RIGHT.

I have here a portrait, taken outside, exposed to the right.

Nothing blows, there isn't a single pixel blown.

But through this overexposure that I did exposing to the right, I somewhat changed the appearance of the skin. It looks drier.

Summary: exposing to the right may result in technically perfect images, but this overexposure changes the image characteristics. I found that skin not exposed to the right looked better than exposed to the right.
[a href=\"index.php?act=findpost&pid=165837\"][{POST_SNAPBACK}][/a]

You might supply images to back up your impressions. Also, after obtaining your ETTR image, did you bring the tones back down to their proper values with the exposure controls? Since the sensor is linear, the properly processed ETTR and non ETTR images should be quite similar, except the ETTR image may have better tonality and less noise.
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 8576



WWW
« Reply #51 on: January 08, 2008, 08:38:26 AM »
ReplyReply

Quote
People who do not understand the value of ETTR talk a lot about "correct exposure", and are usually referring to the act of literally metering a gray card, or some equivalent exposure technique.

I can't speak for most people. I can say I use the term "correct exposure" with ETTR because I find people think of ETTR as "over exposure" which it isn't. When they see the suggested exposure based on their meters, based on a JPEG capture, they tend to then think ETTR is over exposure in this context, but its not.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
eronald
Sr. Member
****
Offline Offline

Posts: 3855



WWW
« Reply #52 on: January 08, 2008, 08:56:22 AM »
ReplyReply

John,

Assuming ambient illuminantion is adequately exposing the gray card, this card or an incident meter will allow you to determine the unity/area reflected illumination off a diffusing surface. Which will theoretically yield optimal exposure ETTR for such diffusing surfaces.

However, shiny surfaces (metal, glass plastic, wood, paint, water) or textures like skin have specular and radiosity effects associated with them which will render such an exposure inappropriate whenever these entities are imaged.

I don't think it's realistic to assume that we photograph scenes that don't contain skin, metal, glass, plastic, wood, paint or water. Such are present even in studio or architecture work, and worse even in every landscape, street, indoors or nightclub scene.  Such scenes also frequently integrate radiant light sources.

So the gray card or incident light approach needs to be tempered with some headroom assumptions, and a decision of how much of the image one is willing to burn or color out. It  would be interesting to debate how we could determine the appropriate exposure absent an HDR device.

Edmund




Quote
People who do not understand the value of ETTR talk a lot about "correct exposure", and are usually referring to the act of literally metering a gray card, or some equivalent exposure technique.  And, to tell the truth, I really don't see anything wrong with equating the phrase to this.  This "no processing" paradigm needs to exist, even if we don't use it most of the time, out of personal choice, if only as a frame of reference.  IOW, a gray FOV, exposed correctly, will leave RAW highlights unused.  An exposure that leaves the brightest excursions just below RAW saturation is an "optimal exposure" (assuming, of course, that Av and Tv values are not compromised).  "Optimal exposure" is a phrase I am more comfortable with in the context of ETTR.
[a href=\"index.php?act=findpost&pid=165808\"][{POST_SNAPBACK}][/a]
« Last Edit: January 08, 2008, 08:57:14 AM by eronald » Logged

Edmund Ronald, Ph.D. 
Schewe
Sr. Member
****
Offline Offline

Posts: 5415


WWW
« Reply #53 on: January 08, 2008, 11:05:18 AM »
ReplyReply

Quote
It  would be interesting to debate how we could determine the appropriate exposure absent an HDR device.
[a href=\"index.php?act=findpost&pid=165879\"][{POST_SNAPBACK}][/a]


Pretty easy...and some people have been doing this for years with film. It's called a spot meter...

Meter the lightest tone you wish to maintain texture. Measure the darkest tone you want to maintain texture. What is the range between them. If less than 7-8 stops, then you can ETTR. If over 9-10 stops then you need to evaluate the scene to determine if the shadows are more important than the highlights. If so, then let the highlights blow and expose for the shadows. If the highlights are more important, then expose for them.

If all detail ranges are important, either do an HDR bracket, add fill light or find a better scene to shoot.
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #54 on: January 08, 2008, 05:08:37 PM »
ReplyReply

Quote
However, there is another side of the issue: utilizing some of those 2048 levels means at the same time increasing the number of levels at the low end.

What worth is an additional "low noise" stop, which contains only eight levels?

Using only the first 2048 levels means, that the seventh stop will consists of only FOUR levels; going to the right increases this to eight levels. So, what about the 70 levels?
[{POST_SNAPBACK}][/a]

There is an easy solution to solving the problem of excessive levels in the highlights and  the paucity of tones in the shadows and that is to change the encoding of the data. That the brightest f/stop contains 2048 of the 4096 levels of a 12 bit capture is not due to the nature of light or peculiarities of digital capture, but to the use of linear integer encoding, which works well for current cameras, but would not be satisfactory when we get true HDR output from our cameras.

For example, in discussing the scRBG standard, [a href=\"http://www.anyhere.com/gward/hdrenc/hdr_encodings.html]Greg Ward[/url] has noted, "a linear ramp is employed to simplify graphics hardware and image-processing operations. However, a linear encoding spends most of its precision at the high end, where the eye can detect little difference in adjacent code values. Meanwhile, the low end is impoverished in such a way that the effective dynamic range of this format is only about 3.5 orders of magnitude, not really adequate from human perception standpoint...". You seem to regard plethora of highlight tones to be   an advantage, but the experts seem to think otherwise.

By going to floating point encoding (there are special formats than represent the required precision with relatively few bits), the encoding precision is nearly constant over the entire range. Log encoding would make the precision totally uniform, but then you would no longer have raw data. These considerations are discussed by Ward and I would encourage thoughtful readers to look at his paper.

Since noise, not posterization, is most limiting in the shadows, no major manufacturer seems to have done this for still cameras. Another solution to improve shadow tones would be to add a few bits to the ADC, but thus far the results have not been impressive, again for the reasons cited above.
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #55 on: January 08, 2008, 05:22:01 PM »
ReplyReply

Quote
Pretty easy...and some people have been doing this for years with film. It's called a spot meter...

Meter the lightest tone you wish to maintain texture. Measure the darkest tone you want to maintain texture. What is the range between them. If less than 7-8 stops, then you can ETTR. If over 9-10 stops then you need to evaluate the scene to determine if the shadows are more important than the highlights. If so, then let the highlights blow and expose for the shadows. If the highlights are more important, then expose for them.

If all detail ranges are important, either do an HDR bracket, add fill light or find a better scene to shoot.
[a href=\"index.php?act=findpost&pid=165904\"][{POST_SNAPBACK}][/a]

Jeff,

An eminently concise post. I don't see much room for argument, but others might.

Details of implementation are of interest. I know you and Bruce have discussed this before. Bruce mentioned that he had borrowed a Minolta Spot meter from you and his preferred method was to take a reading from a highlight for which he wished to maintain some texture, and then placed the exposure so many f/stops over indicated according to previous tests. I think he used a factor of 3 EV or so. I have experimented with this method, but find the camera histogram sufficiently reliable for my purposes, especially after running some tests to correlate the appearance of the histogram to the contents of the raw file?

I presume a similar procedure would apply to cases where you wish to preserve shadow tones, but you would use a different offset.

What do you use most in your day to day work?

Bill
Logged
The View
Sr. Member
****
Offline Offline

Posts: 989


« Reply #56 on: January 08, 2008, 06:08:21 PM »
ReplyReply

Quote
You might supply images to back up your impressions. Also, after obtaining your ETTR image, did you bring the tones back down to their proper values with the exposure controls? Since the sensor is linear, the properly processed ETTR and non ETTR images should be quite similar, except the ETTR image may have better tonality and less noise.
[a href=\"index.php?act=findpost&pid=165857\"][{POST_SNAPBACK}][/a]

Yes, I use the exposure controls.

I posted my comments more in a sense of me exploring the issues.

I remember, in black-an-white film, when you exposed too much (and the negative got too "dense", you lost some skin details.

Of course, that's film, not digital. But it's just a connection that sprang up.

I can see that there is still a lot of room for exploration in this issue. To express yourself in your own photographic language.

All you say makes perfect sense. I have to explore this, if there is a "right" amount of exposure, even in RAW.
Logged

Deserts, Cities, Woods, Faces - View of the World.
The View
Sr. Member
****
Offline Offline

Posts: 989


« Reply #57 on: January 08, 2008, 06:11:31 PM »
ReplyReply

... Ansel Adam's Zone System...

Could there be an equivalent in RAW shooting?

Could it be possible that ETTR preserves more detail in mid-tones, but also changes their character?
« Last Edit: January 08, 2008, 06:12:43 PM by The View » Logged

Deserts, Cities, Woods, Faces - View of the World.
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #58 on: January 08, 2008, 06:32:04 PM »
ReplyReply

Quote
Could it be possible that ETTR preserves more detail in mid-tones, but also changes their character?
"Character" is not in the metadata of raw images :-)

You need to define in technical terms, what kind of changes you mean, or provide comparative examples to point out what you mean.
Logged

Gabor
djgarcia
Sr. Member
****
Offline Offline

Posts: 343



WWW
« Reply #59 on: January 08, 2008, 06:46:26 PM »
ReplyReply

Quote
"Character" is not in the metadata of raw images :-)

You need to define in technical terms, what kind of changes you mean, or provide comparative examples to point out what you mean.
[a href=\"index.php?act=findpost&pid=166000\"][{POST_SNAPBACK}][/a]
I think a non-linear interpretation across the zones introduced because of the zone-shift through ETTR might be a quantitative interpretation of "character change" ...

For me spot metering and zone system visualization to which ETTR philosophy (?!)augmented by two or three brackets has been applied has worked nicely, using the built-in spot meter linked to the AF point in my 1Ds 2/3.
Logged

Over-Equipped Snapshooter - EOS 1dsII & 1DsIII, Zeiss & Leica lenses
http://improbablystructuredlayers.net
Pages: « 1 2 [3] 4 5 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad