Original RED One sensor has an active area of 4520x2540. In normal 16:9 shooting of 4096x2304, the rest of the image is used for "look around". We added a 4.5k wide mode of 4480x1920.
The M-X sensor has the same 5.4micron pixel pitch as the original in the RED One, but the active area is larger, out to 5120x2560.
Now onto measured resolution: All sensors generally have more pixels than their measured resolution, as otherwise there would be too much aliasing occurring as you try to get too much detail onto the sensor, and that extra detail "folds back" to appear as lower frequencies in the image, thus corrupting the image with moire.
So with the RED we record, say, 4096 across, but the measured resolution is somewhere between 78% and 85% of that depending on lens, M or M-X, and level of REDCODE compression. This is measured on a totally un-sharpened image, because on any camera you can always add post sharpening, and most video cameras have such sharpening built into their signal processing if you want it or not. Even top end 1080p HD cameras don't measure a true 1080p because they filter to avoid aliasing, or allow through too much aliasing that actually corrupts the fine detail they're trying to record.
Although Arri advertise a 3.5k sensor, the active area for HD is only 2.8k, the rest being "look around". Now, as far as I know, nobody has published resolution test charts for the Arri, so we don't know the exact details of how it performs. Obviously, they should measure > 2k on their RAW recording, otherwise there'd be little point in them doing RAW recording over 1080p HD-SDI output, although if they properly optically filter and depending on the algorithms use, them may find the RAW resolution is closer to 2.25k, and therefor see little benefit from the RAW. This, of course, depends on the quality of the in-camera downsample and demosaic, which may be improved by offline algorithms.
Graeme
-
http://www.nattress.com - Film Effects and Standards Conversion for FCP
Reply
Like
