Reader Comment: Olympus E-M5 Mark II High-Res Mode
Related: dynamic range, exposure, noise, Olympus, Olympus EM5, Olympus M4/3, Olympus mirrorless, Olympus OM-D E-M5, Olympus OM-D E-M5 Mark II, reader comment
UPDATE: see the noise comparison of hi-res vs standard-res on the E-M5 Mark II.
Gary L writes:
Thank you for the great article on the EM5 Mark II's sensor shift technology. I agree, this technology is very exciting.
It seems, not only does one get a higher resolution image, but also less noise and more accurate colors. I was wondering if one of the benefits of the hi-res image might also be a higher dynamic range. Would love to hear your thoughts on this.
DIGLLOYD: dynamic range and noise are interrelated in that a higher noise level raises the black level off of black. Since dynamic range is the range of dark to bright, higher noise reduces dynamic range.
To quantify noise, a doubling of exposure time cuts the main type of noise by √2. With the multiple exposure hi-res mode of the EM5 Mark II, there are eight (perhaps nine) exposures made that are interwoven in hardware by the camera in a complex way. It seems likely that the actual usable dynamic range would be higher as a result by reducing (averaging out) errors from noise. However, the file format is 12 bits and that places a hard upper limit on the DR to 12 bits. Still, if the usable dynamic range increases from 10.5 or 11 bits (due to noise) to 12 bits, that would be a significant gain. The effect is probably minimal at base ISO, but might accrue at higher ISOs. But the hi-res mode requires a tripod, and up to 8 seconds exposures are possible, so the use of higher ISO would be unusual in hi-res mode.
I have not set out to prove or disprove whether the actual usable dynamic range is improved, and I don’t have a clear read on it in a technical way. However, I would say this: the huge 9216 X 6912 images 64MP / 105MB raw files themselves are all those samples containing differing data values that are merged. Downsampling that file size 2X linearly (to native resolution) improves per-pixel quality, which includes averaging out per-pixel noise. So in this sense there is lower noise, at the least. Downsampling to ~40 megapixels has similar noise reducing benefits while retaining the increased detail.
