In my attempts to learn about shooting and editing video I encountered a guideline that stated that for video the shutter speed should be the inverse of the frame rate. So for 25fps the shutter speed should be 1/25 s and so on. The reasoning for this is that it gives moving objects in each frame some motion blur so that motion the change from frame to frame isn’t quite as abrupt. What I read sounded and felt like good advice, but it did lead me to wonder if the same holds when the motion is very fast.
As is readily apparent I shoot a lot of pictures of fast moving four-legged objects that are often black – aka flatcoated retrievers. In my photography I’ve noticed that often even 1/250 s isn’t enough to completely freeze movement when the dogs are in full motion. So I was a bit skeptical whether the advice would apply even when shooting fast moving dogs. Or any other fast moving object for that matter. In defense of my skepticism I must admit that I didn’t doubt the advice when considering objects moving at normal speeds: human walking, normal gesturing, human running etc.
The idea for this shoot is something I’ve had for quite a while, but since I don’t have two cameras capable of shooting manual video, I had to wait for my friend Victor Jason to come over. We’ve reduced most of the variables we could to get similar enough image quality that it wouldn’t be a factor. The following videos were shot with two Canon 7Ds mounted on a single tripod. Both cameras are using a Canon EF 70-200L f/4 USM, although the baseline had the IS version. ISO speed, image styles etc. were also identical. Because we didn’t have ND filters there is a difference in aperture in addition to the shutter speed. Both cameras were also prefocused to the to the line that Topi used when running right to left. My wife Anna served as Topi’s handler and for each take he ran a “mark” – he saw the dummy being thrown and retrieved it once he was given permission. I threw each mark so that they landed in roughly the same spot every time.
We shot two sets of baselines with varying shutter speeds. The first video is shot at 720p with 50fps and the second at 1080p with 25fps.
I added the 50% slow motions just to help in seeing what the differences are with the various shutter speeds since Topi is mostly moving too fast. But the idea was to get him moving across a fairly limited field of view to maximize the amount of movement between each frame and we didn’t have any reliable gear or enough experience to seriously consider panning.
Before I discuss my own observations, here are three screen captures from the 1080p video with the shutter speeds of 1/50 s, 1/200 s and 1/1000 s. Of course in all cases the baseline 1/25 s is in the screen capture as well. BTW, note that the leading image was shot at 1/1000 s although not in the same session.
When originally editing the shots my initial impression was that there will be noticeable differences between say 1/50 s and 1/400 s. Of course this was based on looking at single frames and doing slow scrubs back and forth so the impression wasn’t exactly based on watching the actual footage. When I finally watched the rendered clips, I was surprise at how good the baselines look in both cases. Even when watching them on my 24″ monitor at too close a distance the amount of motion blur in the baseline doesn’t make the footage unacceptable to my eye.
But once I watched the edited clips over and over again I began to notice a few things. Most notably:
- higher shutter speeds show more detail and reflection in the coat (good)
- 1/25 s is practically a blurry blob in frame grabs (bad)
- 1/50 s is blurry but shows more reflection in the coat (better)
- I can’t really see significant differences between 1/200 s and 1/400 s (good)
- In 1/1000 s the motion starts to get a bit jerky after repeated viewings (bad)
My own final conclusion is that almost anything between 1/50 s and 1/400 s goes and produces acceptable enough footage. I’ll probably tend to pick 1/200 s in in most similar situations but I’ll know that lower shutter speeds will probably be better when the motion is slower or smaller on screen. 1/25 s is much too blurry for my taste, at least with a black dog. Hmm… Maybe we should have shot a similar set with Luka after all while we were at it ;)
But I’ll be happy to say that the advice I read (sorry, I don’t remember the source) isn’t wrong but is solid advice for most scenarios. I’d love to hear what other viewers think based on these comparisons or their own experiences.