Pages: 1 2 [3] 4   Go Down

Author Topic: Using Spot Metering and Exposure  (Read 10337 times)

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1645
    • http://www.guillermoluijk.com
Re: Using Spot Metering and Exposure
« Reply #40 on: May 28, 2017, 04:19:52 AM »

The remarkable aspect of "headroom" is that, even among cameras of the same make and model, that headroom can vary.  Three cameras of the same brand and model were tested: one had an extra  2/3 stop, the next one and 1/3 stops, and the third  had two and 2/3 stops of extra DR ....  indicating that each and every camera ought be tested for its actual raw-accessible DR.!

If any of the responders above have done such testing of their own camera to determine how much extra raw-accessible DR it has beyond the point that the JPEG clipping warning /blinkies is set off, I'm hoping they would report their results at base ISO.

I doubt such a variance exists. Do you have RAW files proving this for the same exact scene?.

RAW highlight headroom isn't such a simple concept as referred to in this thread since RAW files contain three colour channels which can clip individually (generally G is the first to fall). How they clip and how this clipping correlates to camera light metering, heavily depends on the kind of subject/light present in the scene.

In such a multichannel/multisubject/multisprectrum environment it's nonsense stating "this camera has 2,5 stops of RAW highlight headroom". Just spot meter over any other scene and the figure will vary because of the changing variables.

If the RAW highlight headroom is evaluated against camera JPEG instead of camera metering, variance in determining it increases exponentially because of white balance, tone curve, saturation,... and any other process involved in building the JPEG file than can clip some channel.

I did some spot metering tests on lab (uniform white surface under tungsten light) and real world scenes for the Canon 350D and 5D, and got these very different highlight RAW headrooms:

http://www.guillermoluijk.com/article/spot/index.htm

Lab 350D:


Lab 5D:


Real world scenes 5D:









While the lab test over a uniform surface and tungsten light provided 3,5 stops of headroom, this real world scene has 1 whole stop less highlight headroom, surely because the tiny dark areas on the wall fooled camera spot metering. More examples in the provided link.

So forget to accurately characterize your camera's highlight headroom and learn by practice how your camera's JPEG correlate to RAW clipping under different situations. UniWB may help.

Asking camera manufacturers for RAW histograms has demonstrated to be wasted time. Those interested in such things must be considered niche nerds by those splendid marketing departments.

Regards
« Last Edit: May 28, 2017, 05:52:32 AM by Guillermo Luijk »
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3266
Re: Using Spot Metering and Exposure
« Reply #41 on: May 28, 2017, 09:27:16 AM »

The loss of DR with increasing ISO is extremely variable. I have one camera with extra DR of one and 1/3 stops at ISO 100 that loses its first 1/3 stop at ISO 400, and another with one and 1/3 stop at ISO 100 that waits until ISO 1600 before it loses its first 1/3 stop of DR.

y'gotta test your camera!

Generally speaking, with an ISO less camera (such as the Nikon D810 and other cameras using recent Sony sensors), one loses 1 stop of DR for each doubling of ISO. This is simple physics. In doubling of ISO, the amplifier gain is doubled and the analog to digital converter (ADC) overflows one stop earlier, halving the DR. The relationship between ISO and DR is shown graphically for many cameras on Bill Claff's Photographic DR page. Examples are shown below in the attachment. The interval between x-axis markers is 2.5 f/stops. For the D810 increasing the ISO by 25 stops from ISO 100 loses 2 stops of DR and the curve is nearly linear.

Many cameras not using Sony sensors have the DR at low ISO limited by high read noise, but the RN drops as ISO is increased and reaches a minimum at higher ISO. For the Canon EOS 1DX, DR hardly changes as the ISO is increased by 2.5 stops from ISO 100, but at ISOs above about 800, RN reaches a minimum and the curve becomes linear and 1 stop of DR is lost for each doubling of ISO.

The Sony 7RM2 uses Aptina variable gain technology, and the DR plot shows discontinuities when the gain is changed, but the 1 stop loss of DR for each doubling of ISO generally holds.

I find it hard to believe that any camera can go from ISO 100 to 1600 with only 1/3 stop loss of DR. What camera is that?

Regards,

Bill
« Last Edit: May 28, 2017, 09:39:14 AM by bjanes »
Logged

uuglypher

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
Re: Using Spot Metering and Exposure
« Reply #42 on: May 28, 2017, 03:25:21 PM »

The camera that looses 1/3 stop of DR BETWEEN ISO 800 and ISO 1600 is a sony a99.  Of course, one ought use the lowesst ISO possible, but should also depend on specific exposure tests at different ZISOs to determine DR loss, NOT theoretical predictions of a linear rate of loss. And this inconsistent rate of DR LOSS has been demonstrated not only in some Sony's , but also some Nikons , some Canons, a Pentax, and a Fuji,

My best recommendation, if one is seriously interested,  is to test each individual camera (sensor) yourself. It would be a shame if one consistently were to leave one...or two...or possibly three stops (i think that would be a record) of DR "sitting on the table unused...and thereby forfeiting significant image data quality (how much dependent on the bit-depth of the data!). The greatest unused DR I've found was Two and 2/3 stops in a Pentax (model?) and in a Nikon d800. I also tested a d800 that had only 2/3 stop of unused DR.

Which, incidentally, explains my surprise when I was told quite explicitly by dxO, that they test one ( count 'em , 1) camera of each brand and model for their vaunted test reports.

Curious, eh?

Best regards,
Dave
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3266
Re: Using Spot Metering and Exposure
« Reply #43 on: May 28, 2017, 03:55:40 PM »

The camera that looses 1/3 stop of DR BETWEEN ISO 800 and ISO 1600 is a sony a99.  Of course, one ought use the lowesst ISO possible, but should also depend on specific exposure tests at different ZISOs to determine DR loss, NOT theoretical predictions of a linear rate of loss. And this inconsistent rate of DR LOSS has been demonstrated not only in some Sony's , but also some Nikons , some Canons, a Pentax, and a Fuji,

Dave,
According to Bill Claff's Chart for the A99V, the DR at ISO 400 is 9.19 stops and the DR at ISO 1600 is 7.39 stops. Two doublings of ISO results in slightly less than 2 stops of DR. Where do you get your figures?

Regards,

Bill
Logged

uuglypher

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
Re: Using Spot Metering and Exposure
« Reply #44 on: May 28, 2017, 06:54:23 PM »

I have seen the various charts and "official ratings" , but have found they vary in minor to significant degrees from what one will find by exposure tests on one's own individual camera or cameras. I'll admit that  It seemed puzzling until I was made aware of the phenomena of "process variance /process excursions" and subsequent "performance variance" that have plagued the semiconductor production iindustry from the time of Shockley's first publication on the topic in, as I recall, 1960 or '61 up to the present day. A study of the history of those phenomena disabused me of the idea that our cameras' image sensors were highly refined devices that deliver Dynamic Range within strictly defined limits.

Anyone willing to personally test the raw-accessible dynamic range of a number of cameras, including multiple cameras of the same brand and model will discover, as a number of us have, that there is considerable variation in the upper limit of DR delivered by ostensibly identical sensors. The camera manufacturers assure that their sensors will deliver at least the DR necessary to at least blow out highlights of JPEG files, but seem uninterested in how much more so-called "headroom" their sensors will deliver. That, of course, makes providing an in-camera raw histogram a pricy option evidently not-to-be-considered, what with how it would boost the per-unit price and all!

What amazes me is that so few have actually done the due diligence of conscientiously testing their cameras for the actual raw-accessible DR they can deliver. It's not hard. certainly not beyond the capabilities of those likely to be participating in these fora. They just need to do it.  Testing my own cameras and those of friends, and participants in classes, courses,  and workshops, I've personally seen a range of of extra raw-accessible DR of from 2/3 stop to two and 2/3 stops, with the great majority of cameras having at least one full stop of unused DR. There are rumors of some cameras with three stop or more of unused DR.

But even one full stop of extra DR is nothing to sneeze at. Failure to use it in raw image data capture results in forfeiture of up to 50% of tonal and chromatic spectra..or more, depending on scene DR. You all should be able to figure out what that amounts to depending upon 12-bit-depth vs. 14 bit-depth.

Best regards,
Dave
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3266
Re: Using Spot Metering and Exposure
« Reply #45 on: May 29, 2017, 09:18:57 AM »

Anyone willing to personally test the raw-accessible dynamic range of a number of cameras, including multiple cameras of the same brand and model will discover, as a number of us have, that there is considerable variation in the upper limit of DR delivered by ostensibly identical sensors.

Bill Claff has addressed the possible variation in sensor performance of cameras of the same make and model through a collaboration where users around the world take a series of paired pictures according to his protocol and send them in to him for analysis using his custom automated protocol. When I first got my Nikon D3, I performed a series of tests as outlined by Roger Clark to determine the DR of the sensor and also sent files to Bill Claff. My results were quite close to his and those published for the D3 by Peter Facey and others. No significant variation here. I have also performed a similar analysis of my D800e and my results agree with Bill's and the DXO results.
 

What amazes me is that so few have actually done the due diligence of conscientiously testing their cameras for the actual raw-accessible DR they can deliver. It's not hard. certainly not beyond the capabilities of those likely to be participating in these fora. They just need to do it.  Testing my own cameras and those of friends, and participants in classes, courses,  and workshops, I've personally seen a range of of extra raw-accessible DR of from 2/3 stop to two and 2/3 stops, with the great majority of cameras having at least one full stop of unused DR. There are rumors of some cameras with three stop or more of unused DR.

If you look at the linked sites by Roger and Peter, you will see that the methods are rather laborious and not at all as simple as you describe. You are new to this forum and do not have a track record by which we can judge your qualifications. Why don't you tell us your testing protocol and publish your raw data. Until then, I will give little credence to your posts, which are out of the mainstream here.

Regards,

Bill
Logged

uuglypher

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
Re: Using Spot Metering and Exposure
« Reply #46 on: May 29, 2017, 02:47:48 PM »

Logged

bclaff

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 86
    • Photons to Photos
Re: Using Spot Metering and Exposure
« Reply #47 on: May 29, 2017, 03:16:18 PM »

Bill Claff has addressed the possible variation in sensor performance of cameras of the same make and model through a collaboration where users around the world take a series of paired pictures according to his protocol and send them in to him for analysis using his custom automated protocol. When I first got my Nikon D3, I performed a series of tests as outlined by Roger Clark to determine the DR of the sensor and also sent files to Bill Claff. My results were quite close to his and those published for the D3 by Peter Facey and others. No significant variation here. I have also performed a similar analysis of my D800e and my results agree with Bill's and the DXO results.
 

If you look at the linked sites by Roger and Peter, you will see that the methods are rather laborious and not at all as simple as you describe. You are new to this forum and do not have a track record by which we can judge your qualifications. Why don't you tell us your testing protocol and publish your raw data. Until then, I will give little credence to your posts, which are out of the mainstream here.

Regards,

Bill
Bill, thanks for your help over the years in contributing files.
FWIW, I don't use image pairs. Image pairs can be used to remove Fixed Pattern Noise (FPN) to get more accurate results. There are others ways to accomplish this.
In any case, I'm more focus on what the photographer will actually observe so in most of my tests I consciously don't remove FPN because it will be in the image you take.

Regards,
Bill
Logged
Visit me at Photons to Photos

uuglypher

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
Re: Using Spot Metering and Exposure
« Reply #48 on: May 29, 2017, 04:16:07 PM »

Hi, Bill,
You asked for a description of my test exposure system. Have you tried what I described in the link provided? It certainly is not a complex procedure.

Attached are some examples of use of the extra raw-accessible DR determined by the described procedure.

It is a system that I and others have found to work well.

If you are satisfied that you are using your camera's full complement of extra raw-accessible DR, I'd be the last to suggest you change what you are doing.

If you do give it a try, I'd be interested in your results .

Best regards,
Dave
Logged

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1645
    • http://www.guillermoluijk.com
Re: Using Spot Metering and Exposure
« Reply #49 on: May 29, 2017, 05:14:00 PM »

Your procedure depends on JPEG highlight clipping, which depends on the white balance set and other adjustments (saturation, colour space,...), so it will provide a different highlight headroom everytime is tested.

Moreover ACR is a fantastic RAW developer but a poor RAW analyser, so you can't trust its highlight warning be an accurate RAW clipping warning tool. Use DCRAW or RAW Digger instead.

Regards
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3266
Re: Using Spot Metering and Exposure
« Reply #50 on: May 29, 2017, 05:58:15 PM »

Bill, thanks for your help over the years in contributing files.
FWIW, I don't use image pairs. Image pairs can be used to remove Fixed Pattern Noise (FPN) to get more accurate results. There are others ways to accomplish this.
In any case, I'm more focus on what the photographer will actually observe so in most of my tests I consciously don't remove FPN because it will be in the image you take.

Regards,
Bill

Bill,

Since I haven't updated my camera since getting the D800e, it has been quite some time since I contributed files, so I didn't remember exactly what I contributed, but I was happy to contribute to your valuable project. I plan to update from the D800e to the D820 or whatever and will, of course, contribute to your project.

Since you have entered into the thread, have you noted much variation in DR among cameras of the same make and model?

Regards,

Bill 
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3266
Re: Using Spot Metering and Exposure
« Reply #51 on: May 29, 2017, 06:06:19 PM »

Your procedure depends on JPEG highlight clipping, which depends on the white balance set and other adjustments (saturation, colour space,...), so it will provide a different highlight headroom everytime is tested.

Moreover ACR is a fantastic RAW developer but a poor RAW analyser, so you can't trust its highlight warning be an accurate RAW clipping warning tool. Use DCRAW or RAW Digger instead.

+1.

As Guillermo knows, ACR uses a baseline exposure offset that has to be taken into account and PV2012 applies a tone curve and automatic highlight correction. I have found that if one uses PV2010 and a linear tone curve (sliders in main panel set to zero) I can get an approximate estimate of highlight clipping. Of course, I use Rawdigger to evaluate my raw files.

Regards,

Bill
Logged

uuglypher

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
Re: Using Spot Metering and Exposure
« Reply #52 on: May 29, 2017, 11:03:34 PM »

"Your procedure depends on JPEG highlight clipping, which depends on the white balance set and other adjustments (saturation, colour space,...), so it will provide a different highlight headroom everytime is tested."
"...a different highlight headroom every time is tested" ???

Not so. In fact,  multiple re-testing of the same camera/sensor at the same ISO yields a great majority of identical results, only rarely evidencing 1/3 stop variation (usually the result of misinterpreting a specular highlight spike antecedent to actual clipping. The key is that the continuous test exposure series carries both the point of jpeg clipping, and includes enough exposure to cover the onset of raw image data clipping.
The testing procedure is straightforward and simple to understand.  The results are easily recognized. Above all, it is an easily performed process and the results obtained are reliable and are reliably employed  in routine use.

Theorizing about it is no match for understanding the process, trying it fairly, and putting it into practice.

Best regards,
Dave
Logged

uuglypher

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
Re: Using Spot Metering and Exposure
« Reply #53 on: May 30, 2017, 01:25:20 AM »

".

+1.

As Guillermo knows, ACR uses a baseline exposure offset that has to be taken into account and PV2012 applies a tone curve and automatic highlight correction. I have found that if one uses PV2010 and a linear tone curve (sliders in main panel set to zero) I can get an approximate estimate of highlight clipping. Of course, I use Rawdigger to evaluate my raw files."

Bill,
Thank you for your comments. Please see my response to Guillermo's comments .
Neither I nor others using similar procedures as that which I've described are, not in the slightest, given to repeatetitive use of what Guillermo suggests are practices predicted to yield unpredictable results.
The practices I summarized in the link that I provided are easy to repeat and yield useful results as the examples I provided illustrate.

best regards,
Dave
Logged

BartvanderWolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 7326
Re: Using Spot Metering and Exposure
« Reply #54 on: May 30, 2017, 10:57:17 AM »

Theorizing about it is no match for understanding the process, trying it fairly, and putting it into practice.

Hi,

It seems you are preaching to a choir of people who know what Dynamic Range means.

What you seem to be advocating is overexposure, and reliance on (Lightroom's) highlight reconstruction ability (which usually works best with achromatic highlights). However, Camera histograms/blinkies show the amount of clipping after White/Blackpoint correction, White Balancing,  and gamma adjustment, yet the Raw data channels may clip at different exposure levels. So OOC JPEG histograms or histograms from e.g. Lightroom's Process 2012 offer poor general guidelines for true DR and Raw data clipping. This is usually understood by the choir, that's why many of them use RawDigger or Fast Raw Converter which do allow to see Raw clipping. A Raw converter like RawTherapee offers more control over the Raw channel data scaling before demosiacing as well as tone-mapping after demosaicing.

Having said that, if timing and storage space allows, it never hurts to use an exposure bracketing strategy if one can, to either pick the best exposed Raw data file, or use an over-exposed shot for improved shadow definition (in Exposure Fusion together with a highlight preserving shot, or by attempted highlight reconstruction of the overexposed/clipped Raw data).

Cheers,
Bart
« Last Edit: May 30, 2017, 02:55:42 PM by BartvanderWolf »
Logged
== If you do what you did, you'll get what you got. ==

uuglypher

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
Re: Using Spot Metering and Exposure
« Reply #55 on: May 30, 2017, 11:47:52 AM »

Hi, Bart,
Not preaching, just offering a practical and simpler alternative that usually requires but a single exposure.
And I am most definitely not advocating "overexposure" , rather accepting the premise that proper raw image data exposure is based on making the brightest possible exposure without clipping highlights, but coming as close thereto as possible, which, incidentally, requires using the entire DR...which, incidentally, requires that maximum available DR for each camera be individually determined at the most frequently used ISOs. Using the lowest possible ISO commensurate with intended image features is always recommended.

Also, to similarly characterize the live jpeg-adjusted histograms of all brands of cameras is complicated by the evident variety of the different manufacturers algorithms used to produce what each considers the "ideal" JPEG.

Again, the system I (and a number of others) use is offered for the interested to try, certainly not by way of proselytizing.

Take it or leave it, but a number of us think that not to give it a rigorous and fair try would be, indeed, regrettable.

Best regards,

Dave
Logged

BrownBear

  • Full Member
  • ***
  • Offline Offline
  • Posts: 101
Re: Using Spot Metering and Exposure
« Reply #56 on: May 30, 2017, 01:05:43 PM »

Reeling this back to the original indoor setting involving multiple light sources and a handheld Sekonic meter, there's a much easier way to reach the Valhalla of perfect exposure.

Switch that little Sekonic gem to incident metering. Meter the two light sources. Add a tiny little splash of fill light to spice the image for exactly the exposure ratio you want and press the shutter release.

No handsprings with your camera, and no post processing.  Just shoot the image and free up all that post-processing time your spot meter indentures you to. 

Oh wait....  This isn't the Pro Business Discussion section, is it.
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3266
Re: Using Spot Metering and Exposure
« Reply #57 on: May 30, 2017, 02:16:01 PM »


+1.

As Guillermo knows, ACR uses a baseline exposure offset that has to be taken into account and PV2012 applies a tone curve and automatic highlight correction. I have found that if one uses PV2010 and a linear tone curve (sliders in main panel set to zero) I can get an approximate estimate of highlight clipping. Of course, I use Rawdigger to evaluate my raw files."

Bill,
Thank you for your comments. Please see my response to Guillermo's comments .
Neither I nor others using similar procedures as that which I've described are, not in the slightest, given to repeatetitive use of what Guillermo suggests are practices predicted to yield unpredictable results.
The practices I summarized in the link that I provided are easy to repeat and yield useful results as the examples I provided illustrate.


Dave,

Guillermo and Bart are two of the most knowledgable posters on this forum and their opinions should be disregarded at one's peril. From some of your examples I suspect that there is highlight clipping that has been "recovered" by ACR. As a learning exercise, I suggest that you take some of your ERADR images and look at the raw histograms in Rawdigger or Fast Raw Viewer and report back to us. Evaluation downloads are available, but I would think that any serious experimenter would want to buy one or both (I have both).

The Rawdigger site has 3 valuable posts that are well worth reading:

How to use the full dynamic range of your camera

How to derive the hidden baseline exposure applied by your raw converter

Forcing a raw converter to render tones accurately

Regards,

Bill







Logged

uuglypher

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
Re: Using Spot Metering and Exposure
« Reply #58 on: May 30, 2017, 08:42:49 PM »

Please, Bill,
Don't mistake considering and using alternative approaches with disregarding. Not a bit of it.
I am, however, impressed with the results I and others have achieved in practice of the techniques used and have been repeatedly bemused at being told that this ot that test or app, plug in , or program will explain why the images we achieve do not reveal what they do, in fact, reveal, particularly when told they can't possibly be satisfactory by those who have never actually put the described technique to the practical test.

The recent example was being told that testing for extra raw-accessible dynamic range as described cannot possibly yield uniformly repeatable results. That was proof positive that the source of that statement was experientially deprived as regards the process. If those results were not reliably repeatable
no one would be successfully using the described technique of EBTR.

Having done the due diligence of learning, testing, refining, and consistently using the technique and passing it on to others who have taken up its use, I am disinclined to discontinue its practice merely because one  authority or another proclaims that it doesn't...shouldn't ...won't work. Rest assured, as long as it does, those of us experienced with the technique will continue in its use.

I do thank you for your comments and links you provided.

Regards,
Dave
Logged

uuglypher

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
Re: Using Spot Metering and Exposure
« Reply #59 on: May 31, 2017, 12:13:32 AM »

And Bill,
I meant to add that I do, indeed have and use both Rawdigger And Fast Raw Viewer.
You asked that I provide you with results of both. You seem to have missed my point that I'm not trying to convince you to use the technique I've earlier described and use. I don't feel the need to do your due diligence. If you, or others, wish to speak more knowledgeably about the EBTR technique, rather than simply theorize as to why it cannot perform as I and others have described and illustrated that it does perform, the basic instructions are there to be followed.  Evaluate the resultant images and check them with Rawdigger and Fast Raw Viewer and draw your own conclusions.

Best regards,
Dave Graham
Logged
Pages: 1 2 [3] 4   Go Up