29 Nov Can we finally kill ETTR?
For virtually the entire duration of the DSLR-revolution I’ve adhered to the ETTR principal for exposing my shots. But is this hack recommended, or even necessary, for the Blackmagic Design Pocket Cinema Camera?
What is ETTR?
Ask two people and you’ll get two different answers. Most will agree that the general concept is to intentionally over-expose an image so as to “compress” more of the image into the upper half of the histogram, and out of the noisy shadow range. Then, when back in the safety of the edit bay, lower the exposure proportionately resulting in a cleaner image.
Some people like to over-expose a set amount (generally one or two stops) and some people prefer to push the histogram as far to the right as possible without clipping highlights. Some claim one technique works better for one camera versus another, but regardless, the intent is usually the same: get an image with less noise.
What are the possible downsides to ETTR?
- Monitoring tools don’t show the shooter (or anyone else on set) what the image is supposed to look like.
- The shooter’s intent is not communicated to the colorist.
- It only benefits lower contrast scenes.
- A portion of the limited “bit bucket” will be wasted (256 steps in 8-bit. 1024 steps in 10-bit) increasing chances of banding when expanded.
- It adds a step in post.
- If not over-exposing by a known amount (metering) and instead using the histogram, it ups the margin of error/variability for exposure correction in post.
- It increases the chance the shooter accidentally clips highlights.
- Most codecs are tuned for the mid-range to be their sweet-spot; by shifting the mid-range to the highlights, there is a greater chance for color error.
Should I ETTR with the Pocket 4K?
You can do anything you want, but given the above, and given the Pocket 4K’s strong sensor and quality codec, I wanted to see if the benefits outweigh the downsides. After over a month of shooting, I’ve decided that no, shooting ETTR isn’t necessary, or even worth the hassle. Let me show you why…
ISO 200 Noise Test
Can’t see a difference? Here, let me blow up the darkest patch 400%.
ISO 800 Noise Test
I don’t know about you, but I can’t see a reason to futz with ETTR for ISO 200 on the Pocket 4K. But what about at higher, noisier ISOs?
Let’s compare two ISO 800 images, one “properly” exposed, the other over-exposed 4 stops!
The biggest difference I see between these two images has creative, not technical, implications…the DOF has changed. Going from f/4 to f/16 really changes what is and isn’t in focus. I believe that’s a decision that should be based on a desired look, not to squeeze a little less noise out of a sensor.
That said, at this extreme (and over-exposing four stops is extreme), there is less noise in the ETTR image…
ETTR’s impact on color
If over-exposing at higher ISOs can yield less noise, is there any harm in doing so? It turns-out that the Pocket 4K is remarkably good about holding color accuracy across the entire luminance range of a single exposure.
- The light RGB lines represent the properly exposed chart.
- The Dark RGB lines represent the ETTR chart.
- The pure RGB lines represent the calibrated chart values.
- The grey lines represent a technically perfect response.
note: The fact that both of the Pocket 4K images are darker than the known calibration values is not a fault of the camera, but rather the target values are based on middle grey being exposed at 50%, whereas in video, we typically expose middle grey for 43% IRE.
Color patch deltaGrey patches don’t tell the whole story. How well does each exposure match the expected values for all 24 targets on the chart?
Across the entire spectrum, there’s not more than a 2% difference between either exposure, and as a whole, the Pocket4K is never more than 7% off of of the target. That’s quite impressive.
note: The known calibrated values are our baseline, so they are all at zero. The chart shows the difference from those known values.
Why not ETTR?
If exposing to the right can yield images with slightly less noise, and there’s little impact to the color, why not use ETTR? I think that’s the wrong question. I think the better question is, “Why bother using ETTR?”
Which of these two images would you rather monitor on set? Which would you rather hand over to a colorist? A client?
Are you a one man band who only needs to make yourself happy? Which one do you think will be easier to work with in post?
But the primary reason I think that ETTR should be put to rest when shooting with the Pocket 4K is ETTR’s dirty little secret…
It sucks-up precious light.
Think about it: each additional stop of over-exposure requires twice as much light. If you have the ability to double, or quadruple, the amount of light hitting your sensor then why not simply lower your ISO, get less noise and expose correctly?