According to Tony Northrop, ISO in digital cameras are fake?

[Sal Santamaura]

I have heard of Northrop - An American company - they used to make quite good aircraft...

One of my former employers. After I left the company it seems they had to merge with Grumman.

 

[warden]

function of the Spectral Quantum Efficiency, Fill Factor, Well Capacity and the exposure time (more correctly the integration time), dark current, Bayerian filter recombination schema for color sensors, analog gain setting, Bit depth of the A to D converter, and finally the digital gain setting.

You could have snuck in dark matter, the Flux Capacitor and a Langstrom 7" Gangley wrench and I would have been none the wiser.

Thanks for sharing - I appreciate the learnings. The video is rather like a high school project, and then some college profs walk in and put some science to it.

 

[markjwyatt]

Digital imagers are a different media than analog silver halide, so a different ISO standard applies. As a matter of fact, the number of ISO standards that apply to digital imaging is absolutely massive, covering the full breadth and scope from design to use to storage of the digital media.

It's a fallacy to say that a digital sensor has a single sensitivity: There is no functional equivalent to traditional analog sensitivity in the digital world. Rather, the output of a digital camera is a function of the Spectral Quantum Efficiency, Fill Factor, Well Capacity and the exposure time (more correctly the integration time), dark current, Bayerian filter recombination schema for color sensors, analog gain setting, Bit depth of the A to D converter, and finally the digital gain setting. Of these, the only real analogous equivalent to film photography is exposure time. The other parameters are dependent either on the physical properties of the silicon detector or on the design of the FPA and ROIC itself (not to mention the post-processing which occurs downstream of the digital gain stage on the output of the A2D). That said, when you really want to geek out about the physics of the processes, Quantum Efficiency was often compared in the early days....but QE has to be redefined for analog as the efficiency of the conversion of photons to developing centers on the grain. Put another way, that process wasn't widely known as "QE" before digital imaging came about. See http://www.imaging.org/site/PDFS/Papers/1998/PICS-0-43/622.pdf for one of the earliest examples.


Analog and digital imaging are almost entirely two different animals, the functioning of which are about as closely related as film photography and oil painting. All three of these media mentioned produce an image in entirely unrelated ways, but the end result -- an image observable with your eyes -- are equivalent (if the oil painter is a realist).

Rather than fall into the trap of analogies between the two media, you really have to consider the digital Focal Plane Array as an entirely different type of imaging system. Blame the early producers of digital cameras for the common practice of trying to compare the two media directly. Looking back to the days before digital photography hit the scene (i.e. the 1980s and 1990s), users of digital imaging systems evaluated the performance of those sensors in terms that made sense to the capture of an image through the photovoltaic effect rather than silver halide. However, discussing the capabilities of the digital camera in terms which make sense to those users -- who lie outside of the realm of consumer / commercial photography -- would have left film photographers unable to appreciate the benefits of the digital cameras that manufacturers were hoping to sell in the early 2000s.

So either independently or in conjunction, all those parameters that affect the digital output aside from exposure time were rolled up into a single sensitivity value equivalent to film's ISO rating, the high equivalent numbers of which were then (conveniently ignoring the very important SNR values) touted as a significant benefit to using digital imaging. Having been a digital imaging user and designer outside of the realm of consumer photography at the time, I watched this comparison of ISO rating spring up and establish itself in the consumer photographic market. Its use is unique to this particular digital imaging market. Satellite imaging systems, for example, have no ISO rating listed in their detailed engineering / user performance specification nor even in management power point presentation because it is misleading in understanding the true sensitivity and performance of the imager.

So from that perspective, what the guy is saying is sensible. However, that sort of rationalization is only one factor in defining the terms that a specialized community adopts. Simplicity in understanding, rules of thumb, and tradition have as much or more to do with terminology as anything else. Very few photographers will recognize the meaning of the term Double Gauss, but we all know what a 50mm lens is (they are both the same thing, almost without exception).

A much deeper explanation, thanks.

When I was working with thermal imaging with cooled InSb sensors (both scanning and array) the software basically allowed the user to set a temperature range, then modified the electronics to create a window. Integration time was something you could manipulate directly also. These were considered quantum well detectors. Analogously, for digital (visual) imagers, maybe software could be created to allow the user to set a zone range (say II-IX) or even a bit range (say, 8-bit, 5-252) then let the electronics place some percentage of the scene in that range (say 99.5% or 95%, or user settable % of the pixels in the scene).

 

[Nodda Duma]

My pleasure, fellas. Ironically, knowing/learning all that stuff and characterizing all the new FPA’s coming out at the time to see if we could use them for work is what kept me shooting film. I didn’t want to process imager data at my computer all day and then go home and process images at my computer all evening!

 

[tedr1]

Here's a little bit more imaging science. The sensors used in "digital" cameras are analogue devices, bright light produces a bigger electrical charge than dim light. Sensor electrical output (the various values of electrical charge corresponding to darker and lighter parts of the scene) passes into an analogue to digital convertor IC (ADC) the output of which is where the "digital" part of the rest of the camera originates, and which perform the camera image display and memory functions. The sensor ADC has to convert the charge values that arrive simultaneously from millions of sensor pixels and do so very rapidly to a high degree of accuracy. In popular formats of digital camera the convertors operate to 12 and often 14 bit accuracy, which after formatting editing and compression is used to produce the familiar 8 bit JPG image.

 

[nmp]

If Northrop had used "erroneous" instead of "fake," this whole controversy might have been avoided. But then he would have had not as many clicks and a big long discussion in his name. It seems his main problem (that I surmised after a few minutes of video before I stopped) was that different cameras when shot with the same iso gave different results.

 

[mmerig]

If Northrop had used "erroneous" instead of "fake," this whole controversy might have been avoided. But then he would have had not as many clicks and a big long discussion in his name. It seems his main problem (that I surmised after a few minutes of video before I stopped) was that different cameras when shot with the same iso gave different results.

I agree that some other word rather than fake could have been better, as "fake" is a loaded term these days. But it was the graphic image and plain curiosity that motivated me to look into the video. I don't really see much controversy, but maybe there is some if people don't even watch the whole video and then criticize it. I am surprised by the mostly negative reaction to a video that did a reasonable job of explaining something using actual measurements. APUG is a tough crowd, I guess.

 

[alentine]

If Northrop had used "erroneous" instead of "fake," this whole controversy might have been avoided.

I think if digital photography has been recognized in its own context as digital photography or imaging in the first place, there will be no shocking discoveries can be made.
As clearly mentioned above, it’s different media.
Image features(on print) for example, between the rendering of electron charge transformation and the optical photon depiction, may become the next hot topic discovery.
I think it has been made intentionally to distort the vocabularies of the concise scientific dictionary to conform with different but established popular and successful context.