Craig Parkes
Well-known member
Tom, the RAW recording off the sensor of the sensor data has 16 bits of precision according to what red has said. Whether that equates to the same amount of color depth and variation as say a 16 bit negative scan in RGB is probably questionable, because after debayering you are arriving at 16 bit values which have been derived via other calculations, including compression and decompression, rather than 16 bit values which have been measured directly as they would be from a scan.
From a RAW point of view what is true according to RED is that REDCODE RAW from Epic and Scarlet offers 4 more bits of precision than REDCODE RAW from a Red ONE. Gunliek is the only person I have seen talking about this in particular with his regular assertion that Epic and Scarlet have a much thicker neg than Red One MX.
How that stacks up to other cameras bit depth on a technical level is difficult to tell, because there aren't other compressed RAW cameras out there to do comparisons against, although at 3:1 compression on Epic you are going to be close to mathematically lossless, so comparing that to a camera that did say 16 bit Camera RAW stills with a bayer sensor would give you a good indication of the bit depth advantage of Epic. AFAIK there are no other 16 bit motion cameras out there.
I'd wager you have to dig pretty deep to visually see the difference between say 16bit linear and 10bit log from a visual point of view, especially given the bit depth limitations of current display devices, and you'd be digging into a purely maths arena to do comparisons of say 16bit linear RGB outputs from two cameras, unless you were doing developmental work on a compositing app where that levl of precision may help more complicated transforms hold up (the idea when compositing being numbers with larger precision plugged in to big complicated equations means more accurate results out the other end).
Bit depth at recording in RAW is a measure of precision. How good the color science in the RED SDK is at using that precision to derive RGB images of whatever bit depth is ultimately all that matters when you are comparing to most other comparable codecs and camera recording bit depths I think.
From a RAW point of view what is true according to RED is that REDCODE RAW from Epic and Scarlet offers 4 more bits of precision than REDCODE RAW from a Red ONE. Gunliek is the only person I have seen talking about this in particular with his regular assertion that Epic and Scarlet have a much thicker neg than Red One MX.
How that stacks up to other cameras bit depth on a technical level is difficult to tell, because there aren't other compressed RAW cameras out there to do comparisons against, although at 3:1 compression on Epic you are going to be close to mathematically lossless, so comparing that to a camera that did say 16 bit Camera RAW stills with a bayer sensor would give you a good indication of the bit depth advantage of Epic. AFAIK there are no other 16 bit motion cameras out there.
I'd wager you have to dig pretty deep to visually see the difference between say 16bit linear and 10bit log from a visual point of view, especially given the bit depth limitations of current display devices, and you'd be digging into a purely maths arena to do comparisons of say 16bit linear RGB outputs from two cameras, unless you were doing developmental work on a compositing app where that levl of precision may help more complicated transforms hold up (the idea when compositing being numbers with larger precision plugged in to big complicated equations means more accurate results out the other end).
Bit depth at recording in RAW is a measure of precision. How good the color science in the RED SDK is at using that precision to derive RGB images of whatever bit depth is ultimately all that matters when you are comparing to most other comparable codecs and camera recording bit depths I think.
