Testing my GH2's sensor

Started Feb 20, 2011 | Discussions thread
kenw Veteran Member • Posts: 5,372
Re: Testing my GH2's sensor

Al Bond wrote:

Does "shoot to the right" only hold good for ISOs which are not artificial (ie not digital scalings of lower ISOs)? In which case, I wonder what the highest real ISO of the G1 is?

So ETTR really only makes sense at base ISO. For every other ISO it turns out rather than doing ETTR you'd be better off just lowering the ISO setting on the camera (note, both ETTR or lowering ISO result in a longer exposure). This guy has a pretty good write up on when ETTR does you some good (base ISO only) and when it isn't buying you anything or making things worse (every other ISO):


He also outlines some of the problems you might encounter when doing ETTR.

Now I wouldn't worry too much about some of this being out of one's comfort zone because from what is written it is clear ETTR is out of nearly everyone's comfort zone! Very few people are using it correctly or even seem to understand what is happening. It doesn't help that the often cited "more levels in the highlights" argument actually doesn't hold any water - that is completely the wrong reason to do it and just demonstrates that a lot of the louder proponents of the technique don't understand what they are actually doing. Conceptually of course the "levels" argument is easy to make and seems like common sense but it turns out to be wrong because it ignores what are the actual dominant noise sources in an imaging system.

The reason ETTR "works" is that it increases the signal to noise ratio and this is entirely a function of increasing the exposure so more photons hit the sensor. Whether you increase the exposure by dialing in exposure compensation or lowering the ISO doesn't matter, all that matters is the longer exposure. The read noise and photon shot noise is always significantly larger than the quantization error (the number of "levels"), hence the "levels" argument is missing the whole point. By the time you applied a strong enough tone curve to see the "levels" you'd be mired in so much read and shot noise that you wouldn't be able to identify the levels anyway. You could, in theory, design a camera that was limited by quantization levels but no one has since it is comparatively easy to design sufficient levels (bits) into the system such that the limitation comes from the sensor.

A lot of ETTR proponents think levels have something to do with it, but this would only be the case if the sensor and photon statistics were noiseless - which they aren't. A particularly interesting case would be the NEX5 which appears to be the camera closest to having sufficiently low read noise compared to the number of bits in its ADC that one might imagine the number of quantization levels could be an issue at base ISO in the deep shadows.
Ken W

Rebel XT, XTi, Pany G1, LX3, FZ28, Fuji F30, and a lot of 35mm and 4x5 sitting in the closet...

 kenw's gear list:kenw's gear list
Sony RX1 Panasonic Lumix DMC-G1 Panasonic Lumix DMC-GM1 Olympus E-M5 II Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS +34 more
Post (hide subjects) Posted by
(unknown member)
(unknown member)
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow