What I learned from Gollywop -- and what I wonder

Started Mar 26, 2013 | Discussions thread
Detail Man
Detail Man Forum Pro • Posts: 16,789
Re: if i wanted to test/compare ETTR ...

All that matters is that F-Number as well as Shutter Speed are minimized as much as possible (given DOF and camera stability considerations), and that the user employ some known to be reliable means (utilizing preview/review/playback indicators such as histograms or "blinkies", or a spot-metering technique accompanied by increasing actual sensor-level Exposure by some known amount) of approaching (but not - at least by very much - exceeding) what are maximum (linear) RAW ADU levels.

Because RAW processors (such as all of Adobe's) may re-scale the RAW ADU levels existing in the OOC RAW image-file, use the free, stable, and non-invasive RawDigger application to assess actual recorded RAW channel ADU levels. It is available for download on multiple platforms here:


Paul Auclair wrote:

... what am i looking for as far as better IQ and why ETTR is better?

The ratio of the recorded Signal divided by the Photon Shot Noise varies with the square-root of the Signal level. This is true on the level of the image-sensor photosites themselves, and is determined by the sensor-level Exposure [a function of Scene Luminance, lens-system T-Number (including losses), and Shutter Time, and not a function of the ISO Sensitivity setting].

By maximizing the (linear) recorded RAW channel levels (using the lowest possible ISO Sensitivity factor that results in such a maximization), you are at the same time maximizing that particular (Photon Shot Noise related) Signal/Noise Ratio (SNR) by having maximized sensor-level Exposure.

This SNR improvement is proportional to the increase in Exposure level (in EV) divided by two.

Contrary to many common conceptions, this improvement has little to do with the number of discrete analog/digital converted RAW ADU levels (because the Photon Shot Noise dominates).

less (read) noise?

Think in terms of the maximum Dynamic Range (DR) - the ratio of the maximum (linear) recordable RAW ADU level divided by the Read Noise. (Input-referred back the level of the photosites), Read Noise in most modern image-sensors remains fairly constant as a function of the ISO Sensitivity setting.

The (input-referred) Read Noise may decrease somewhat to a relative minima at some particular ISO Sensitivity. However, in the specific case where one is maximizing the value of the recordable RAW ADU level, what really matters is the usable Dynamic Range for recording [the ratio of the maximum (linear) recordable RAW ADU level divided by the Read Noise, rather than the Read Noise itself] at any particular ISO Sensitivity setting value.

In all cases (that do not involve ISO Sensitivity adjustments via digital multiplication/division operations following analog/digital conversion), the usable Dynamic Range for recording will be highest at the lowest ISO Sensitivity setting that allows for maximum (linear) RAW ADU channel levels to be recorded.

This DR improvement is directly proportional to the increase in Exposure level (in EV).

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow