Advertisements

Has the Olympus OM-D E-M5 conquered the notorious rolling shutter video jello effect?

Olympus claims to have fixed the rolling shutter video wobble with the E-M5 - here is the evidence

The great news is that we now have a review sample OM-D E-M5 and we are starting to test it. One of the easiest tests to do is to verify Olympus' claims that the infamous video wobble that is characteristic of many MOS/CMOS sensor type cameras has been fixed.

This problem is noticeable when the camera is unsteady and is amplified by longer focal length lenses because details on the frame are moving around the frame much faster. What happens is that when recording in video mode the camera'selectro-mechanical focal plane shutter can't be used. The sensor has to perform as an electronic shutter. With most MOS-type sensors the pixels are processed from one side of the frame to the other, a line at a time, from top to bottom, for example. This is called a rolling shutter. Unfortunately, if a subject is moving fast in relation to the read-out, it may have moved in the frame by the time it has been recorded from top to bottom. This creates a skewed effect. Verticals lean and when you view this at 25 or 30 frames per second the leaning becomes a swaying effect. If the subject is moving in two dimensions the rolling shutter effect can be even more absurd, as if you're watching a TV screen made out of wobbling jelly.


Here is the evidence - video wobble has been tamed admirably on the E-M5 compared to the Pen E-P3 - only in high speed panning movements is there a very slight line noticeable with the E-M5

All Micro Four Thirds cameras to date have exhibited this effect to varying degrees. In fact practically all DSLRs with video modes have suffered from the rolling shutter effect, until recently. In the last year I have noticed that some newer cameras don't seem to be as badly affected as others. One possibility is that they are using a 'global' electronic shutter sensor. With such sensors, instead of reading off the pixels line by line pixels are read from all over the sensor. Indeed, this is how old-fashioned CCD sensors work and is why you don't see the rolling shutter effect in many compact cameras that use CCS sensors.

The other possibility is that by driving the sensor read-out much faster you can minimise the effect of a rolling shutter. But this would not eliminate the effect. Without any data from Olympus about the new non-Panasonic sensor used in the E-M5, it's hard to be precise, but in our tests we did see a reduction in rolling shutter effect rather than it being eliminated. Olympus certainly uses very high read-out rates - up to several hundred frames per second - to improve focusing speed and reduce flicker in the viewfinder, so my money is on a very high read frequency being used to minimise the problem.

If there is a negative, I have noticed some occasional frame jitter - I would be grateful if other E-M5 users could let me know if this is their experience too. It could simply be an issue with the video encoding method I use. Whatever, on the whole it looks effective and should be a welcome improvement for E-M5 video shooters.

Reader feedback:

Discuss this story: