Residual Noise Following Exposure Averaging
Percent of maximum residual noise before and following exposure averaging as a function of the number of exposures averaged for various ISO speeds.
Note that ISO 200 failed, perhaps due to the experimental method. As expected, ISO 160 yielded the least change in noise following averaging. Surprisingly, ISO 800 (and not ISO 1600) yielded the greatest reduction following averaging.
Noise was quantified by the auto-profiling function of Noise Ninja, and represents noise in both the Lum and Chroma channels.
All photos were taken in two sequential three-burst fine-quality 10 megapixel JPEG captures using the Sony DSC-R1 (21.5 x 14.4 mm CMOS sensor with 5.49 µm pixel pitch for 10.3 million effective pixels at a 3:2 aspect ratio). RAW captures developed using ACR would have been better, the the burst mode is not possible for RAW captures with the R1.
This semi-quantitative analysis was propted by Sean McHugh's Noise Reduction by Image Averaging article, and Duncan Smith's recent investigations here, here, here, here, and here.
Residual Noise Following Exposure Averaging
Percent of maximum residual noise before and following exposure averaging as a function of the number of exposures averaged for various ISO speeds.
Note that ISO 200 failed, perhaps due to the experimental method. As expected, ISO 160 yielded the least change in noise following averaging. Surprisingly, ISO 800 (and not ISO 1600) yielded the greatest reduction following averaging.
Noise was quantified by the auto-profiling function of Noise Ninja, and represents noise in both the Lum and Chroma channels.
All photos were taken in two sequential three-burst fine-quality 10 megapixel JPEG captures using the Sony DSC-R1 (21.5 x 14.4 mm CMOS sensor with 5.49 µm pixel pitch for 10.3 million effective pixels at a 3:2 aspect ratio). RAW captures developed using ACR would have been better, the the burst mode is not possible for RAW captures with the R1.
This semi-quantitative analysis was propted by Sean McHugh's Noise Reduction by Image Averaging article, and Duncan Smith's recent investigations here, here, here, here, and here.