Unintentional camera movement - I just invented the UCM acronym as click-bait for the blog title...
UCM (as opposed to ICM which I wrote about in a previous post, here:- ICM impressions) I'm sure you're aware, is a right pain in the workflow. As sensors continue their esurient march onward and upward to ever more resolution, an annoying corollary (particularly for landscapers at the mercy of random meteorological conditions) is an increasing sensitivity to the slightest unwanted movement influence. If you are not careful with your technique, especially if you are hand-holding, it can only mean one thing - more camera-shake blurriness in your images.
Notwithstanding the recent developments in vibration reduction technology in lots of new lenses, camera bodies and telephones, there are still many photos ruined by the old enemy - camera shake. Even with my VR 300mm on a tripod, when conditions are windy and affecting the lens enough to need the VR switched on, I can't get all the gear stable enough to eliminate shake completely. I frequently switch off VR and use a bungee strap to secure my equipment to a solid anchor such as a railing, post or pillar (is that schoolboy tittering I hear over there?!)...
Anyway, encouraged by a discussion with my assiduous brother, I recently conducted investigations into the maths and physics of the conundrum. The mathematicians and physicists among you may very well take issue with my methodology but, after double-checking to the best of my photographer's abilities, I think it's pretty sound...
Obviously sensors come in a variety of dimensions and pixel densities, but my calculations are based specifically on the camera that I use, the Nikon D810 - and they go like this:
Sensor dimensions - in pixels
= 7360 x 4912
Sensor dimensions - in mm
= 35.9 x 24
Sensor dimensions - in microns
= 35,900 x 24,000
Pixels per mm
Pixels per micron
Pixel dimensions - in microns
= 5.125 x 5.125
For the purposes of the demonstration images below I have used a pixel size rounded to 5 x 5 microns, which equates to 200 pixels per linear mm.
If we assume a hand-holding ability that only produces a movement of 2mm per second, then an exposure time of 1/4 second, such as the tripod-mounted London skyline image below, would have 0.5mm movement in it.
0.5mm is 500 microns - or 100 pixels.
The difference between image A and image B is 100 pixels of motion blur.
London lights up at dusk, windy evening, rush-hour traffic, bridge location
Nikon D810, Nikkor 70-300 VR, 1/4 second at f8, ISO 64, tripod
2nd March 2016 at 18.03
It is unlikely you would ever attempt a 1/4 second exposure hand-held, but what about image C below? Better by far than image B, but still unacceptably blurry. What does this image demonstrate? It is actually just 20 pixels of blur, equivalent to raising the shutter speed to 1/128 second and a camera shake movement of 100 microns - or merely a tenth of a millimetre. Even tripod-mounted shots can suffer from that tiny degree of camera shake in the wrong conditions.
It's not until you get down to a single-figure micronic level, say, below a lateral blur of 10 microns (2 pixels) - equating to just a hundredth of a millimetre of camera shake - or a 1/1200 second exposure time (!) that you could either ignore or, depending on end-use, perhaps rescue the lack of sharpness in the image.
Scary stuff for all of us, not just the pixel-peeping brigade.
Of course there are other variables that can affect the outcomes of hand-held or windswept tripod-mounted shots. Some togs have steadier hands than others and may be able to improve on the 2mm per second assumption. Some VR systems are better than others. WA lenses are easier to stabilise than long teles. Pixel densities vary. And so on.
Nevertheless, seeing what effect a movement of just a few microns across the sensor does to an image, it's clear that with the way things are heading in the resolution revolution, if you crave genuinely, pixel-peepingly sharp shots, you're going to have to work hard for them in future.