Re: Re: [APML] Dynamic Range of CCD

[Aplanatic@aol.com]

Jerry,

<<<At some point on each end, the 
film cannot differentiate detail anymore. For instance, on chrome film, it 
just becomes totally transparent with no tone.  Now, if I record my 
exposures when doing this test, I can tell you that at a certain given 
brightness for the gray card, say for an example off the top of my head, 
with ISO 64 speed film, the correct exposure for the gray card is 1/60th of 
a second at f/16 and the film will not record any tone in the gray card if 
I overexpose it by, say, three stops, or underexpose it by four stops.

This is my definition of exposure latitude and dynamic range. In this 
example, the film would have 7 stops of dynamic range.>>>

Your definition and example are clear and reasonable.  I've spent a fair amount of time thinking about the best way to compare film and CCD dynamic range based on this.  Here's my assessment:

First, let's do the problem assuming short exposures, i.e., plenty of light, as in daylight photography.  

It is well known that one needs a signal-to-noise ratio of about 3 to distiguish an object or detail from the background noise.  For instance, you can't claim with better than 99.5% confidence to have detected the presence of a star unless the S/N is greater than 3 for that star.  This is a simple statistical statement based on probability.  It holds for any detection media, including film and CCDs.  If you underexpose Techpan, for instance, to the point where the object has an S/N of 3, then you're at the lower end of its useful dynamic range from the point of view of claiming that the object has actually been recorded.  Negative film turns clear at this limit, and base fog grain is the "noise".

Using this definition, the lower end of the dynamic range of a CCD camera occurs when the signal is three times the readout noise plus the quantization noise.  For the ST-10E, the readout noise is 11 electrons RMS and the quantization noise is a fraction of an electron.  (The quantization levels for the 16-bit ADC in this camera are 1.5 electrons, and the RMS quantization noise is 1/12 of the quantization level.)  To have a detectable signal from an object, we need about 33 electrons from that object for this camera.  I consider this the lower end of the CCD's practical dynamic range for short exposures.

On the upper end, negative film turns black and the CCD's well fills up.  These are roughly equivalent effects.  The ST-10E has a well depth of 77,000 electons.  I take, as the dynamic range of the CCD, the ratio of the well depth to the minimum detectable signal, i.e., 77,000/33 = 2300.  This is about 11 stops.  In practice, one should stay about a 1/2 stop away from the full well capacity to prevent blooming, so it's fair to say that CCD's have at least 10.5 stops of usable dynamic range in the short exposure limit.

For long astro exposures, things get much more complicated.  To some degree, film and CCD's are both limited by sky background. In this limit, it is the effective speed and linearity of the detector that is most important, and not its dynamic range.  But that is another topic for another day, perhaps?

Dave Rowe



--  APML Archives at <http://astro.umsystem.edu/apml/>  ---
             Unsubscribe at <majordomo@seds.org>