This was an experiment to explore new methods to obtain extra dynamic range beyond the capabilities of the current photography gear. This scene is a typical case where an ND graduated filter would either cut the buildings into dark or fail to cover the sun to prevent highlight blowout. Fixing the transition zones of an ND graduated filter and bringing the cut buildings back to light in post-processing would introduce noise problems, because anything as low as the sun will depend on the usable dynamic range of the sensor. Exposure bracketing is also less than ideal / not suitable for long exposure photography as the light condition changes rapidly during sunset. No matter whatever order you bracket, it is susceptible to inconsistent color, wrong light and shadow after blending. Removal of the ND filter (for a quick and correct exposure of the foreground) would cause alignment issues for pixel peepers. Inspired by the stacking techniques in Astrophotography, I used an ND 3.0 filter (ND 1000) without any ND graduated filter, and continuously shot 60 frames, each at f/11, ISO 100 and 8 seconds, deliberately underexposing each frame (just to protect the highlight details around the sun). I batch processed the RAW files and aggressively pulled back the shadow details of each frame (+100 shadow, -100 highlight, +1.5 exposure), thanks to the extraordinary base dynamic range of the Sony IMX094 CMOS sensor inside the Nikon D800. The shadow area for each frame was very noisy due to severely underexposure, but when I stacked all 60 frames and calculated the mean value of each pixel using the Statistics function in Photoshop, the noise went away and the signal-to-noise ratio was significantly improved. This method is easy (no-brainer) and clean (high SNR). There is no more harassment of alignment issues or blending problems.It is applicable to any shape of foreground as long as you have a camera that has excellent base dynamic range as the D800. It is highly recommended for the most picky pixel peepers. Actually thanks to the very high frame rates the sensor and image processor can deliver, the Epic Dragon is certainly adopting this multiple sampling techniques to reduce noise levels (also known as temporal noise reduction). Without such processing, such high SNR would only be possible from a sensor with an exceptional Full Well Capacity. Performances like this seem, to us, above the current technical capabilities of CMOS sensors. via 500px http://ift.tt/1BA7mf9
