Armed with the peak-to-peak (PP) amplitude and frequency of the mirror-slap initial vibration, we can calculate the shutter speed necessary to hold the vibration-induced blur to under 1 pixel.
Consider the image from the preceding post, for a Nikon D800E and a Nikon 400mm f/2.8 lens, as an example. The PP amplitude was 18 pixels and the frequency 10 Hz. Let’s ignore the damping of he first cycle, and assume the shutter opens as soon as the mirror hits the top of its housing, both conservative approximations (they’ll make things look a little worse than they really are).
With a shutter speed of 1/10 second, we’ll get a complete cycle of the 10 Hz oscillation, and a PP error of 18 pixels. With a shutter speed of 1/20 second, we’ll get a half-cycle, and a PP error of 9 pixels. With a shutter speed of 1/40 second, we’ll get a quarter cycle, and a PP error of the same 9 pixels. With a shutter speed of less than 1/40 second, the PP error will be 9 pixels times the sine of the quantity 90 degrees times the ratio of the shutter speed to 1/40 second.
At 1/80 second, the error is thus 0.707 * 9 pixels , or 6 pixels.
At 1/160 second the error is 0.383 * 9 pixels, or 3.5 pixels
At 1/320 second the error is 0.195 * 9 pixels, or a little under 2 pixels
Finally, at 1/640 second, we get the error to under a pixel.
I’m a little surprised how fast a shutter speed it takes to get mirror slap down to a negligible level with a good solid tripod and head, and a lens that, while pretty long, certainly isn’t extreme for sports or wildlife work. If your subject isn’t going anywhere, you can flip the mirror up, wait nine seconds, and trip the shutter, but there are many cases where that strategy will cause you to miss the shot entirely.