The Optimum Digital Exposure

[Jim Kasson]

my logic, naturally, is that the further away you are from ISO-less camera the more you are going to use that play with gain (ISO), right ? and the closer you are to the ISO-less camera (may be Nikon D7000 is the ideal example of that in the nature - except cameras where ISO is only by tag in raw file) the less you might be inclined to play with gain (ISO) - unless you want for OOC JPG or postshot-review-in-camera purposes... and naturally if you are shooting some action with D4/D4s (or even A7s) - those are further away from ISO less than D810 - the more chances it seems you have that some important part of the image will be the area below your S/N = 10 and you might get that above 10 by pushing gain/ISO... no ? you have D4 as I recall too.

I can't argue with any of that. I appear to be biased towards getting more light on the sensor than most. With the camera on a tripod, I make most of my exposures at base ISO. When I can't use base ISO, my first thought is to let the histogram slide to the left, and push in post. I only do that for two or three stops usually, because of possible twists, but I've done five in tests successfully with the D4, D800, D810, a7R. When you do that with the M240 you get the dreaded green shadows.  If I get to ISO 800 with the a7S, I jump to 2000 if I can fit the scene in to take advantage of the increased conversion gain.

So we're not that far apart, really. I suspect I put away my camera when the light gets dim sooner than you do. I hardly ever go above ISO 3200 on any camera.

Jim

[Jim Kasson]

I think if you are creating a model you need to introduce some tolerances there too... I don't think it makes sense to talk about anything related to camera hoping for any kind of precision precision to be better than 1/6 EV, no ? is there a real need to be more precise than 1/6 EV with clipping ? are you really going to tune your exposure that close ? if I could I 'd then rather dial 1/3-1/2 ev back right away (one Iliah Borg always mention that some cameras at least are non linear near clipping)... I talking not about any clipping of course, but clipping in valuable part of the image... as sometimes you want to clip (sacrifice) some specular reflections to achieve the proper exposure for the important parts of the image.

Yes, but it's a lot tougher to do a model like that and the parameters are more open to debate and error. I was trying to work with a pencil and paper model. I don't think it's worth the effort for me to develop a computer one.

Jim

[bjanes]

Bill, I'm talking about the making the in-camera histogram that's derived from a JPEG preview serve as a raw clipping indicator, so in that case the in-camera histogram does not come directly from the raw data, although the world would be a better place if that were an option.

I've about come down on the side of saying that in theory (as opposed to close enough in practice, which we know is true already) you can make the in-camera histogram that's derived from a JPEG preview serve as a raw clipping indicator with the proper UniWB settings. However, I'm having to make some assumptions about the raw-to-colorimetric conversion process that aren't always met in practice.

Jim

I've conducted some experiments with daylight illumination (actually 3200K + 80a filter) with the D800e using normal white balance and UNIWB and comparing the camera histogram to the raw histogram as shown by RawDigger. The camera was set to AbobeRGB and the standard picture control was chose.

This image shows nominal exposure histograms by the camera and RawDigger (with the green channels averaged) at daylight WB and at UNIWB. The red channel is blown in the camera histogram, but is 2 EV below clipping in RawDigger. The UNIWB preview gives a better preview and shows no clipping, contrary to the Sunlight WB



Here are the ACR histograms for AdobeRGB and ProPhotoRGB for the sunlight WB. Note that the AdobeRGB rendering shows red channel clipping similar to that shown in the camera histogram.




And the ACR Adobe RGB histogram for the UNIWB shot at the metered exposure. Note that the red is no longer blown, similar to what was seen in the camera histogram.


Giving 2EV more exposure moves the RawDigger two stops to the right and near saturation (a good ETTR exposure). The UNWB histogram shows clipping on the camera histogram, since AdobeRGB can not accommodate the camera gamut at this exposure level. However, the UNIWB image can be rendered into ProPhotoRGB by ACR with no clipping. One would likely use negative exposure compensation in ACR for the ETTR image. This could cause hue shifts and it is not clear if the increased SNR would be worth it in terms of SNR and color accuracy.



And the ACR histograms without exposure adjustment. Screen captures from a WideGamut monitor, with image assigned Monitor profile and then converted to sRGB for web viewing. Some clipping is unavoidable with these transformations.

[Ray]

Perhaps what is lacking in the current trend of this thread towards the minutiae of ETTR precision, is advice on the practicalities of achieving the desired ETTR exposure in the field, in a manner which allows one to get the shot, or capture the moment.

If one is shooting a static scene with camera on tripod, there should be no problem at all. One simply brackets exposure with mirror up.

The difficulties occur when the camera is hand-held. Choosing the appropriate aperture first to achieve the desired DoF usually applies, although it doesn't have to apply. One can keep the shutter speed constant and bracket aperture if DoF is not an issue. This is not something I've experimented with, so far. Maybe I should.

What I used to do, when shooting with Canon DSLRs, was frequently bracket exposure at a fixed aperture. This was not entirely satisfactory because sometimes the greatest exposure (the slowest shutter speed) which produced the most accurate ETTR shot, was too slow for a sharp image when the camera was hand-held, and/or when the subject was not perfectly stationary.
I would have preferred to have been able to bracket ISO in such circumstances, but my Canon cameras did not have this facility. Maybe later models now have this feature.

What pleases me with my current Nikon cameras, both D800E and D7100, is that one can easily separate exposure and focusing by pressing the appropriate buttons whilst the camera viewfinder is still held to eye. Thus there is no need for any bracketing, unless the scene is so contrasty that one chooses to take different exposures in order to merge to HDR.

Again, this procedure was not possible with my Canon cameras (the latest was the 50D). Or perhaps it was possible but I never discovered or realised that it was.

Essentially, with my Nikon cameras, using a single focusing square with camera in manual mode,  I can assign focusing to the AF-ON button, and exposure to the half-pressed shutter button. If I want to expose for the sky, I can move the camera, and/or focusing square in the viewfinder, till the focusing square covers the brightest part of the scene that I consider merits an ETTR.

With forefinger half depressing the shutter button, I can easily change the shutter speed by turning the wheel with my thumb, until the exposure gauge at the foot of the viewfinder looks right for an ETTR. After achieving the optimal exposure setting, I can then move the focusing square to the part of the scene that I want to be in precise focus, and whilst still keeping the shutter button half depressed, press the AF-ON button, recompose the scene and take the shot.

I can also change the order of these two processes. I can focus first. Having pressed the AF-ON button, there is no need to keep my thumb on the button. Focus is locked. I can then swing the focusing square to the brightest part of the scene that I want an optimal exposure for, make the appropriate shutter speed adjustments whilst still looking through the viewfinder, and with forefinger half-depressing the shutter button recompose the scene and take the shot.

If I find that the adjusted shutter speed required for an ETTR is too slow, I have two options. I can simply use what I think is an appropriate shutter speed and underexpose. Alternatively, since Nikon cameras are not truly ISO-less, there may be a noticeable advantage in raising ISO if a significant raise in ISO is required. This I can do by pressing the ISO button with the left forefinger, then turning the wheel behind the shutter button, using the right thumb. This procedure can also be done with eye still glued to the viewfinder. Voila! What could be simpler! 

[Torbjörn Tapani]

EVF with histogram, blinkies, zebras, peaking. That would be simpler. And IBIS.

[Ray]

Not for me. It would be impossible because I don't have a camera with an EVF. I tried a Panasonic FZ200 a while back but didn't like the EVF and sold the camera. 

[jjj]

I'll check into that picture control, but could you let us know how one accesses that flat picture control?
PS

On checking the Nikon web site online, I see that the linear option is available only on the D810 and cameras released subsequent to the release date of the D810. How could one achieve the same effect for the D800?

I know you use Nikon, but you can tweak in camera picture styles with Canon and also install 3rd party variations like Marvels LowC which were developed for video shooting where there is no raw option and these flat profiles give you more wiggle room in the grade than the processed jpeg. So it may be worth investigating in Nikon also allow that customisation of the in camera looks.
I use the Marvels LowC as my default for stills too as it give me a preview image that's more useful for exposure than a more cooked standard jpeg file.

[dwswager]

If we are going all 'Holy Grail', then what I want as a photographer is simply a metering mode that would estimate the saturation point of the sensor and provide an exposure just below that point.  Without metering at the sensor plane and over all pixels, that is a problem, especially in real time.  I then want 2 numbers from the analyzed shot +/-X and +/-Y where X is the stops the shadows are inside or outside the noise threshold of the sensor and Y are the stops the highlights are inside or outside the saturation point. (Note the shadows part is impossible from one shot if the shadows are clipped because there is no way to tell when they would come above the sensor threshold.  Energy deposition can be integrated over time and saturation value estimated.)

However, what is possible right now is a DR-AEB mode.  That is Dynamic Range Auto Exposure Bracketing mode. Lets assume it requires Matrix Metering, Raw quality and A Priority and continuous shooting mode.  The Settings would be

Bracket Steps in Stops
Shadow Shift in Stops

We don't need to specify the number of shots because the camera is going to do it for us (That's what computers are for!).

Lets assume we set the Bracket Steps to 1.5 and Shadow Shift to 4.  When I release the shutter it takes the 1st shot at the metered exposure and analyses the data.  If the highlights are clipped, it continues taking shots 1.5 stops less exposure until it gets one where the highlights have not saturated the sensor.  Then it moves to the shadows and continues taking shots with additional exposure 1.5 stops above the metered value until the lowest pixel value is 4 stops above the threshold limit of the sensor.  Additional constraints on this are frame rate, buffer capacity and processor speed.  The camera would report back if successful or unsuccessful based on the available shutter speeds to comply with request at the selected aperture and ISO.

In another topic I agreed with a poster that predicted that processing power was the likely ingredient for key technical advancements.  That and the ability to run 3rd party apps on camera like Helicon Remote or specialty bracketing and other stuff I haven't even thought of!

[Fine_Art]

The entire conversation around exposure revolves around the idea you have to get it all in one shot. This is, for many types of photography, a carry over from film that is no longer real. For pro sports or active wildlife it is a real constraint. For a lot of other styles we can embrace the freedom of a stream of data capture over time, using software to put it all back together.

If you missed ETTR but got a 3 shot bracket, you can do something like a median add that weights all frames to an average then adds them, probably creating high enough levels that you are filling a 16 or 32 bit space. You wipe out hot pixels, any extreme level error in a frame (like noise), while getting lots of data.

The light in the scene gives you a stream of data. Stopping motion limits our individual frame to a set amount of data. We can still add more frames.

[dwswager]

The entire conversation around exposure revolves around the idea you have to get it all in one shot. This is, for many types of photography, a carry over from film that is no longer real. For pro sports or active wildlife it is a real constraint. For a lot of other styles we can embrace the freedom of a stream of data capture over time, using software to put it all back together.

If you missed ETTR but got a 3 shot bracket, you can do something like a median add that weights all frames to an average then adds them, probably creating high enough levels that you are filling a 16 or 32 bit space. You wipe out hot pixels, any extreme level error in a frame (like noise), while getting lots of data.

The light in the scene gives you a stream of data. Stopping motion limits our individual frame to a set amount of data. We can still add more frames.

First, I think think getting it in 1 shot, out of the camera is the camera maker's paradigm.  I think that is why they are locked into giving only tools that follow that model.  Even AEB is designed to give the 1 good shot out of 3 - 9 alternatives.  It also isn't something we should lose as a lot of situations call for just that. 

Also, when the DR of the scene is within the capture ability of the sensor in one shot, there are lots of times when one shot is preferable, even when we want to push it right to help the shadows and intend to post process later.  There are inherent penalties and potential problems involved in composting multiple shots.  I use Focus Stacking, but in no way think that is a total replacement for Tilts as a method to increase DOF.  Both are tools that have their strengths and limitations and should be used appropriately.

This brings me right back to processing power and ability to run 3rd party applications in camera.  This does not have to be the 'locked down' versus 'free for all' battle like iPhone/Android.  I think a camera manufacturer that opened up their system to 3rd party apps would benefit and could even absorb functionality into the base camera like Microsoft did with Windows.  I'm not asking the manufacturers to allow 3rd parties to alter how the 'base' camera worked per se, just allow them to utilize that functionality in new and useful ways.

[bjanes]

I know you use Nikon, but you can tweak in camera picture styles with Canon and also install 3rd party variations like Marvels LowC which were developed for video shooting where there is no raw option and these flat profiles give you more wiggle room in the grade than the processed jpeg. So it may be worth investigating in Nikon also allow that customisation of the in camera looks.
I use the Marvels LowC as my default for stills too as it give me a preview image that's more useful for exposure than a more cooked standard jpeg file.

Nikon does allow the PictureStyles to be adjusted. Brightness, contrast, and saturation can all be adjusted. It is a lot of work, but one can check the contrast curve by shooting a Stouffer wedge and rendering the image in camera or with NikonCaptureNX2 or its successor which duplicate in camera processing and with CaptureNX2 one can perform multiple renderings using just one shot. Imatest makes plotting the contrast curve easy.

For those rendering into ProPhotoRGB with ACR or other software, saturation clipping must be considered as well. In the example with the red flower I posted, one must underexpose the camera light meter value by one stop to avoid clipping the red channel with daylight white balance. An ETTR exposure of the red flower requires 2 stops over the metered value. UNIWB can help or one could decrease the saturation in picture control, but the preview image would be washed out. Unfortunately, there does not seem to be much interest in these considerations.

Bill

[thierrylegros396]

Another possibility is to know really well your gear.

After some trials, you'll know how your gear reacts to such situations (red or yellow flowers, snow, dark background,...).

That's why I prefer to keep my gear as long as possible.

[jjj]

First, I think think getting it in 1 shot, out of the camera is the camera maker's paradigm.  I think that is why they are locked into giving only tools that follow that model. 

Not exactly a camera maker's paradigm, but a photographer's one. The reality is that for most people's photos, one shot is the only option and even if it isn't, most people do not want multiple slight variations to then faff around with either.

[Ray]

Obviously there is a range of situations in photography and no single technique is suitable for all of those situations.

I'd say the least problematic of those situations, with regard to achieving a good ETTR shot, is when the subject is static and one has a tripod. Shutter speed should not then be a problem. One can bracket to one's heart's content. One doesn't even have to rely upon one's camera's limitation of +/- 2 EV, or +/- 3 EV, if that's the case. Ansel Adams' zone system doesn't apply.

At the other end of the spectrum, when trying to capture a precise moment, whether in sports or for a Henri Cartier-Bresson type of shot, bracketing exposure would not be the best approach. The best ETTR exposure from all the bracketed shots might have a shutter speed which was too slow, and/or might be a less-than-ideal moment of capture. Even a difference of a fraction of a second can be critical when capturing the moment.

In such circumstances, bracketing ISO would be preferable. The best exposure might still not be the best moment captured, but at least the shutter speed should be sufficient.

Choosing a shutter speed sufficient to freeze the action and/or camera shake, and choosing the aperture required for the desired DoF, or even an aperture because one knows it is the sharpest, should be under the control of the photographer, as far as is practical. Allowing the camera to make that choice for one, is not ideal for the creative photographer.

One of the great attractions of the current choice of Nikon cameras, for me, is their essentially ISO-less nature. They are not completely ISO-less, but most of the time are close enough to being  ISO-less, for me, because I'm not completely obsessive about noise. 

For example, as I mentioned before in this thread, in relation to a comment from Bill Janes, if one underexposes 6 stops with the Nikon D800E at base ISO, instead of using ISO 6400, one loses about 0.9 EV of DR, which is noticeable. However, at least half of this loss occurs between the base ISO of 100 and ISO 200.

In other words, if one underexposes by one full stop at ISO 100, on the D800E, instead of using the same exposure at ISO 200, which would produce an ETTR shot at ISO 200, one loses one full stop (or EV) of DR. However, if one increases the ISO setting to 200, using the same exposure, one loses only 0.47 EV. One gains an improvement of at least 1/2 a stop of DR. That's noticeable.

For this reason, if one wishes to use the D800E as though it's ISO-less, one should use ISO 200 as a base ISO. For example, let's consider what happens if one underexposes one full stop at ISO 200 instead of increasing ISO to 400. According to DXOMark (ain't DXO wonderful, Jeremy  ), one loses only 0.08 EV in DR. That's totally irrelevant.

What happens if one underexposes 4 stops at ISO 200, instead of using ISO 3200? One loses 1/10th of a stop of DR. Still irrelevant.

How about a 5-stop underexposre at ISO 200, instead of using ISO 6400? We now lose 0.37 EV of DR, about 1/3rd of a stop. Hmmm! Not particularly relevant for me, although pixel-peepers might consider it so. 

Conclusion? If you want to take advantage of the ISO-less nature of the D800E, and always choose your own aperture and shutter speed, then use ISO 200 as base whenever you are confident that it will not result in blown highlights as a result of your chosen aperture and shutter speed. The exposure indicator in the viewfinder can be a useful guide, allowing for the fact that a certain degree of overesposure, as shown in that indicator, might be a 'correct', or ETTR exposure for RAW shooters.

Of course, DR is not everything but it seems to be the one parameter that can be most influenced by changes in ISO settings. SNR at 18% grey is largely unaffected on the D800E by choosing a higher ISO instead of underexposing. For example, a change of 3dB in SNR is equal to a change of one stop of exposure. A 6-stop underexposure at ISO 100, on the D800E, should result in an 18dB reduction in SNR at 18% (skin tones). Using ISO 6400 instead of underexposing 6 stops, results in a lowering of SNR by 17.9 stops. One gains 0.1dB. Irrelevant!