Here's a still photo at 1/125:

http://www.palsomedia.com/canon5d2/s..._125_speed.jpg

Movie mode? The shutter speed range in movie mode goes from 1/30th up to 1/125th.

Or exposure simulation? The shutter button information is not reliable.

Take a look: Canon 5D Mark 2 Video Aperture Manipulation Demo on Vimeo

All my tests were under the "Movie Display" mode of the "Stills+Movie" Live View setting, not "Exposure Simulation", and I was looking at the LCD information while recording.

They seem the same shutter exposition. The movie mode has a limited range. The camera doesn't let to be cheated by the EF lens in this case, that's what I think.

Mathieu, it may not have been obvious, but what you just said is exactly the same thing that was being said in posts #39, #41, #42.

A constant speed fan under 60hz light is not a good test.

My son and I did a test last night and confirmed accurate, fast shutter speeds. We're still collecting data...

I've just realized this too. I've been testing using stills + video. This does not give the same results as Movie only. The exposure algorithms for video seems to be something of a mess.

I think Exposure simulation mode is really best for video - it allows you to set the exposure visually and lock it* before recording. It does change the settings once you start recording, but it does so in such a way as to maintain the visual exposure level you've set.

In this mode I've had one situation where I deliberately underexposed in bright sunlight via small aperture, and once I hit record it switched to wide open (f4 on the kit lens, you can feel the lens click when it does this) and shutter speed jumped to 1/500 in order to preserve the underexposed look.

* to do this you have to be in aperture or shutter priority modes

The tests (video and stills) were done using DC (battery) power.

Except for the 3rd (non-EF lens) test where I needed another, more powerful, light to test the higher shutter speed with the low ISO (100) automatically selected by the camera.