Doug Kerr
Well-known member
One of the reasons we may, if we can "afford it", prefer to use cameras with "larger" sensor size is to gain a certain "noise performance" at higher ISO sensitivities.
Before I proceed, let me remind you of a contention I use here for simplistically describing the general size of a sensor, the convention I call the "ruler of thirds".
It is based on the entirely repugnant convention of describing sensor sizes by reference to a historically-important Vidicon TV pickup tube. This tube had a target ("sensor") of nominal dimensions 6.6 mm × 8.8 mm, and was referred to (by virtue of the approximate size of its "bottle") as the "2/3 inch" size.
And in fact this convention is one of the premises of the name of the today-important "four thirds" families of digital cameras, whose standardized sensor size is about twice that of the "2/3 inch" Vidicon.
Much as I hate this naming approach, it is widely used, and so I have decided to embrace it as a "clean" way to describe the nominal sensor sizes of several families of digital cameras. My structure goes like this, working always in integral "thirds" (not, by the way. "thirds of an inch" - just "thirds":
"One inch": Three thirds (3/3)
(Micro) Four Thirds system: Four thirds (4/3). (Duh!)
"APS-C" [1.6x]: Five thirds (5/3)
"APS-H"[1.3x]: Six thirds (6.3)
"Full-frame 35-mm": Eight thirds (8/3)
Hasselblad X1D: Ten thirds (10/3)
Now, with this system of sensor size description in hand, let me speak of the matter of "performance at differing ISO sensitivity as influenced by sensor size."
There are two important realities here:
• The "noise performance" of cameras depends on more than sensor size - but it is the "elephant in the room".
• Comparing "noise performance" in a rigorous way involves technical details not handy for us to use "around the cracker barrel".
But that all having been said, it is interesting to look at a way of describing the performance of a camera in a way that largely involves noise performance, even though it is subjective. I refer to the reports, by Imaging Resource, of the "largest good print size" attainable from various cameras at different ISO sensitivity settings. I captured this data from cameras that are hopefully representative of four important sensor size classes.
I present this not in "absolute" terms but it "relative terms", based on the difference in ISO sensitivity that yields a comparable size print adjudged by Imaging Resource as being "equally good".
I'll state the size class first (in thirds) and then state the camera I chose as representative. Then I will give the "relative ISO sensitivity" for a comparable sized "good" print. I will use the first size class as the reference. The differentials are in"full stops" because the Imaging Resource reports were in terms of ISO sensitivities in one-stop steps for a comparable set of print sizes. Of course these are not "precise" - this is not that sort of thing.
8/3 (aka "full frame 35-mm") (Canon 5D3) 0 (reference)
5/3 (aka "APS-C) (Canon 80D) -1
4/3 (aka "four thirds") (Panasonic GH4) -2
3/3 (aka "one inch) (Panasonic FZ1000) -3
So, in broad terms, if we move from a 5/3 camera (APS-C) to an 8/3 camera (full-frame 35-mm), we might expect to get broadly comparable noise performance (as manifested in terms of a "comparably-good" print of a certain size) at twice the ISO sensitivity.
Best regards,
Doug
Before I proceed, let me remind you of a contention I use here for simplistically describing the general size of a sensor, the convention I call the "ruler of thirds".
It is based on the entirely repugnant convention of describing sensor sizes by reference to a historically-important Vidicon TV pickup tube. This tube had a target ("sensor") of nominal dimensions 6.6 mm × 8.8 mm, and was referred to (by virtue of the approximate size of its "bottle") as the "2/3 inch" size.
And in fact this convention is one of the premises of the name of the today-important "four thirds" families of digital cameras, whose standardized sensor size is about twice that of the "2/3 inch" Vidicon.
Much as I hate this naming approach, it is widely used, and so I have decided to embrace it as a "clean" way to describe the nominal sensor sizes of several families of digital cameras. My structure goes like this, working always in integral "thirds" (not, by the way. "thirds of an inch" - just "thirds":
"One inch": Three thirds (3/3)
(Micro) Four Thirds system: Four thirds (4/3). (Duh!)
"APS-C" [1.6x]: Five thirds (5/3)
"APS-H"[1.3x]: Six thirds (6.3)
"Full-frame 35-mm": Eight thirds (8/3)
Hasselblad X1D: Ten thirds (10/3)
Now, with this system of sensor size description in hand, let me speak of the matter of "performance at differing ISO sensitivity as influenced by sensor size."
There are two important realities here:
• The "noise performance" of cameras depends on more than sensor size - but it is the "elephant in the room".
• Comparing "noise performance" in a rigorous way involves technical details not handy for us to use "around the cracker barrel".
But that all having been said, it is interesting to look at a way of describing the performance of a camera in a way that largely involves noise performance, even though it is subjective. I refer to the reports, by Imaging Resource, of the "largest good print size" attainable from various cameras at different ISO sensitivity settings. I captured this data from cameras that are hopefully representative of four important sensor size classes.
I present this not in "absolute" terms but it "relative terms", based on the difference in ISO sensitivity that yields a comparable size print adjudged by Imaging Resource as being "equally good".
I'll state the size class first (in thirds) and then state the camera I chose as representative. Then I will give the "relative ISO sensitivity" for a comparable sized "good" print. I will use the first size class as the reference. The differentials are in"full stops" because the Imaging Resource reports were in terms of ISO sensitivities in one-stop steps for a comparable set of print sizes. Of course these are not "precise" - this is not that sort of thing.
8/3 (aka "full frame 35-mm") (Canon 5D3) 0 (reference)
5/3 (aka "APS-C) (Canon 80D) -1
4/3 (aka "four thirds") (Panasonic GH4) -2
3/3 (aka "one inch) (Panasonic FZ1000) -3
So, in broad terms, if we move from a 5/3 camera (APS-C) to an 8/3 camera (full-frame 35-mm), we might expect to get broadly comparable noise performance (as manifested in terms of a "comparably-good" print of a certain size) at twice the ISO sensitivity.
Best regards,
Doug