Best piece of advice one can give on the subject. If you have time and space, always bracket. Then you don't have to fret about exposing to the left, exposing to the right, or whatever. Bracket, then you can make these decisions at your leisure with the benefit of a huge calibrated monitor.
Sneakyracer wrote:
Yea. It depends on the camera. Usually it is better to protect the highlights since once they clip they are unrecoverable. You loose all texture, well, everything. With the shadows the dynamic range is much deeper and at worst you need to use some noise reduction to get it clean. You do loose color accuracy but more often than not there is usable information there. Even so a black part of the shadow area is more pleasing and acceptable than a blown out sky or highlight. At leads to me.
With film ETTR was used on color negative film but with slide film one almost always shot to protect the highlights.
Digital cameras mostly allocate their dynamic range in the shadow area while color neg. film had a huge chunk of its DR on the highlight side.
That graph is from a movie-style shootout that has almost nothing to do with still photography. It shows a metering setting in jpg (for the still cameras), not "DR". Most of the DSLR/mirrorless sensors there are only allowed use of about 1/3 of their pixels, and they're also used in high-speed read mode (LV/video). They're also forced to jpg-settings on ISO800 - try comparing the base ISO raw files in stead...
Of course the Prores raw files are going to win here - but try comparing it to a normal camera still raw... Then the contemporary APS cameras actually outperforms many of the the million-dollar systems used in movies.
Dynamic range isn't "allocated" if you shoot raw, only if you shoot jpg. In a jpg, your highlight and shadow latitude is governed by contrast settings and Ev-compensation (metering settings). In a raw there's only two points - "this is blown" and "this is to noisy". Everything between those points are within DR.
theSuede wrote:
Dynamic range isn't "allocated" if you shoot raw, only if you shoot jpg. In a jpg, your highlight and shadow latitude is governed by contrast settings and Ev-compensation (metering settings). In a raw there's only two points - "this is blown" and "this is to noisy". Everything between those points are within DR.
From a less technical approach I prefer ETTL (that is L) as long as no people or similar interesting things will motivate a blown out sky. Sensors allow for reasonable pushing in selected dark areas. Contrast is good for photography and from my point of view areas with blackouts are (usually) better than whiteouts.
In less challenging light environment I think the camera metering works just fine placing the histogram within limits.
Mostly, I try to put my whites around 235-240. Putting them above 245 @ 250-255 was what I thought (for a while) ETTR was all about ... till I came to realize the diff @ white vs. specular highlight.
That's about as far as I'm really comfortable with on the highlight side of things to preserve detail if my whites are important to me. If that doesn't get my shadows close enough to where I want them, then it's either bracket and blend with at least a +2 (anything less is a waste of time, imo), or throw caution to the wind @ blow them away as irrelevant. But, I'm finding that as my PP has improved, my ability to lift shadows more cleanly has improved also.
My +2 minimum bracket for DR purposes is predicated on the midday diff of key vs. shadows side is over 2 stops. While bracketing in 1 stop (or less) increments might be helpful for getting the most correct exposure in meter challenging situations, I find it far too small to be of any help with DR issues. So for me, it's typically white @ 235, +2 and +4 (or maybe just a single +3).
I must say though, that I was surprised by the OP results @ -5 ... which will likely have me skipping +2, in favor of +3 1/2 in the future for my bracket strategy.
That comparison does in no way look at raw values... -It compares different Adobe base S-curve settings
Do a real exposure bracket, look at the real raw values (not the Adobe DCP curve values) and then say that "raw isn't linear".
No camera EVER has resulted in a different response than this (linear down to noise extinction) in my experience. And I've done this with at least some 40-50 models now.
So why are the RG essentially superimposed and the B less than 1/2 of the RG?
I assume this accounts for why the Blue channel is more prone to noise than the other two.
Is that by intentional design relative to the amount of energy carried by the differing wavelengths and the 1:2:1 relationship of the R:G:B bayer array? Or is it something totally different?
They're (R&G) just coincident for the native white balance in a D7000 for the light I did the sequence in (I use HO halogen without filtering in that lightbox to maximize light efficiency). In that warm light, blue is at about -1.33Ev from R&G.
The second series starts exposure at -2.33Ev since I ran out of shutter speed. I do the tests with a F2.8 (T2.95) lens wide open to remove any mechanical differences in aperture actuation. So the second series ends at -13.33Ev when you superimpose the real exposures used.
The channels have a linear response +/- 0.05Ev from 0Ev to -12Ev, where they gradually fade to the noise floor.
theSuede wrote:
That graph is from a movie-style shootout that has almost nothing to do with still photography. It shows a metering setting in jpg (for the still cameras), not "DR". Most of the DSLR/mirrorless sensors there are only allowed use of about 1/3 of their pixels, and they're also used in high-speed read mode (LV/video). They're also forced to jpg-settings on ISO800 - try comparing the base ISO raw files in stead...
Of course the Prores raw files are going to win here - but try comparing it to a normal camera still raw... Then the contemporary APS cameras actually outperforms many of the the million-dollar systems used in movies.
Dynamic range isn't "allocated" if you shoot raw, only if you shoot jpg. In a jpg, your highlight and shadow latitude is governed by contrast settings and Ev-compensation (metering settings). In a raw there's only two points - "this is blown" and "this is to noisy". Everything between those points are within DR....Show more →
I now where the graph came from.
Lets say a scene has a total of 16 stops of EV from the brightest highlights to the darkest shadows. Given an equal point of exposure (EV) the sensor (or film) will record brighter and brighter areas until it clips them same with the shadows. Not all digital cameras (or film) will clip at the same point or EV of the scene. That is what that graph is showing and the numbers can't obviously be interpreted as each of the camera's full potential because they are in camera processed files and also at iso 800. When you shoot raw since like you said it is linear (the data) and during processing, whether its in camera jpg or pro res, the camera or processor applies a contrast curve to make the image look better for viewing not showing the true DR. But still even in RAW each camera will clip at a different value (given equal exposure and iso).
Color negative film, specially the motion picture stock shown in the graph (Kodak Vision 3, 200T (5213) and 500T (5219)) has a much higher highlight headroom than digital, it will clip at a much higher EV value and it will "roll" into the highlight clipping point with a smoother curve than digital cameras. That is a general characteristic of color negative film. But in the shadows it wont be as good as digital since digital cameras generally "clip" softer in the shadows, the image starts getting noisier and noisier and you start loosing color information but there is information very deep into the shadows until there is none obviously.
Hence the practice of exposing for the shadows with color negative film and let the highlights fall where they may and exposing for the highlights with digital (opposite of ETTR). That is the case today maybe the earlier digital cameras had such horrible shadow noise that ETTR was a viable technique.
I would love to see a graph using each camera at it's optimum iso setting and RAW capture. (The new ARRi Alexas can record ArriRaw in camera although you can use an external recorder like a Gemini 444 or a Codex)
