Anybody here using base-ISO exclusively or predominantly?

Started 8 months ago | Discussions thread
wchutt
Senior MemberPosts: 1,314Gear list
Like?
Re: Two Things Count: Analog SNR and How Much Analog Signal Reaches The ADC
In reply to nixda, 8 months ago

"How do these sensors behave at base-ISO and low light levels when it comes to noise? In other words, how isoless are these sensors?"

What you actually want to know is how ISOless is the camera's data stream.

The sensor is completely ISOless. The sensor generates a fixed amount of read noise. While the read noise is temperature dependent, we can ignore this. The signal (light amplitude converted to electrical charge) depends completely on the scene's light along with the shutter time and aperture. As the maximum signal level decreases, the maximum analog dynamic range decreases too.

So, if you must use a fixed DOF, and you select ISO 200 (base ISO), and the shutter speed is too slow to eliminate subject motion blur, all can do is decrease the shutter speed. This decreases exposure which reduces the SNR and DR. The sensor behaves exactly as it would with a slower shutter speed. In other words, the sensor just converts light energy to electrical charge regardless of the ISO, the meter, or any thing else. When the light amplitude exceeds a sensor well's ability to store electrical charge, then no useful information can be recorded and artifacts might be created. Otherwise the sensor's performance is constant.

What matters is how well the ADC can model data with signal levels well below the maximum signal the ADC can represent.

Lets say a 10 bit ADC can digitize signals that do not exceed 2 volts. A 1.9 volt signal would be assigned a photon count estimate of approximately 1024 (2^10). A 0.2 volt signal from another sensor well would have a photon-count estimate about ten times lower.

But if the maximum signal level was only 0.2 volts (because the sensor was underexposed with a shorter shutter time at base IS0), then the maximum photon count estimate would also be about 100 and most of the 10 bits would be used to digitize noise. The 0.2 volt signal from the other sensor would now be 0.02 volts. Can the ADC estimate a photon count for this signal level, or is the ADC's read noise equal to or greater than the 0.02 volt signal level?

The second and third figures in this article may help.

http://www.linkwitzlab.com/dpp/A-D-conversion.htm

Increasing ISO such that the 0.2 V maximum level is amplified to become 1.9 volts matches the maximum signal level to the ADC's maximum. But the ADC is still using most of it's ten bits to digitize noise. The amplification increases the signal and noise equally. But he amplification may increase the 0.02 volt signal to a level where the ADC's read noise has less effect on the photon count uncertainty.

 wchutt's gear list:wchutt's gear list
Fujifilm X-Pro1 Fujifilm X-T1 Fujifilm XF 18mm F2 R Fujifilm XF 35mm F1.4 R Fujifilm XF 14mm F2.8 R +20 more
Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow