Canon 5D Mark II Movie / Video Capability

First published on: Wednesday, 17 September 2008

Last update (Feb 24, 2009): How the 5D2 sets exposure in movie mode

This page is a collection of links to Canon EOS 5D Mark II movie / video capture reviews, samples and other resources.

High quality movies don’t just happen, so maybe it’s time for digital photographers to start thinking about video production techniques.

Read this discussion thread to gain an appreciation of the rather high-spec’d video encoding and format being used in the 5D Mark II. Quote:

The new 5DII uses h264 formatted video at a full raster 1920x1080 size!

h264 is very similar to the emerging AVCHD format, but the data rate Canon is using is absolutely incredible. Here it is from the specs:
1920 x 1080 (16:9) up to 12 mins (Quicktime 1080p H.264; 38.6 Mbits/sec)

This is a very high quality implementation of the h264 CODEC and in a progressive 30P format, should out-perform most professional HD formats. This really is unprecedented to have such high data rates with h264 compression technologies. It will likely be very close an uncompressed look.

Read also about how impressed Vincent Laforet was with the 5D Mark II’s video capabilities when he first took a prototype model on a spin.

How the 5D2 sets exposure in movie mode

“HCR21” explains:

There is no manual control of aperture, ISO or shutter in Movie mode. All is auto

The camera first adjusts ISO than shutter and then aperture in that oder. Exposure control is through ISO change only. There are a number of work-arounds like using Nikon lenses with aperture ring, twisting the lens until it looses contact, and pointing at bright or dark areas and looking the exposure.

Key specification and features

Resolution: • HD -> 1080p, 1920 x 1080 • SD -> 640 x 480

Aspect ratio: 16:9 in HD (high definition), or 4:3 in SD (standard definition).

Frame capture rate: 30 fps (frames per second).

Video and file format: Quicktime .MOV files (H.264 video), MPEG-4 video compression.

Movie clip length: 4GB per clip or a maximum continuous movie capture time of 29 minutes and 59 seconds, whichever comes first. 4GB equates to about 12 minutes of Full HD video (Quicktime 1080p H.264; 38.6 Mbits/sec) or 24 minutes in VGA / SD mode (Quicktime 480p H.264; 17.3 Mbits/sec).

Audio: CD-grade, 16-bit linear PCM at 44.1kHz without compression. The sound is recorded in stereo if you plug in an external stereo microphone into the 3.5mm stereo mic jack. The built-in microphone is a convenience that allows for monaural / mono audio recording. Mic levels and wind noise suppression are adjusted automatically with no provision for manual override.

Autofocus: Contrast-detect AF is available during recording, but will not be quick enough for practical subject tracking. It’s still better than not having any AF at all, and you still have the option of manual AF.

Related discussions

A heated thread on, debating the issues of price and video capabilities of the Nikon D90 vs the Canon 5D Mark II.

Ron Parr explains how a rolling shutter works:

Just to make things simple, let’s assume it takes the same amount of time to reset a pixel as it does read it. Very roughly speaking, to implement a rolling shutter you: 1. Have the electronics cycle through the sensor, one column at a time, resetting the pixels, thereby starting the exposure 2. Wait for the integration (exposure) time 3. Have the electronics cycle through the sensor, one column at a time, reading out the values. If steps 1 and 3 are relatively slow, then you might start step 3 before step 1 is completely finished. You just need to time things so that all pixels in the sensor have the same exposure time. You can think of there being a window of active pixels that slides across the sensor to accomplish the exposure. With this approach, not all pixels are sensing at the same time, so if the image is changing during the duration of the exposure, you will get artifacts. Now imagine that steps 1 and 3 are quite fast. The faster they get, the closer we are to having a sensor where every pixel is active at the same time. In the limit where steps 1 and 3 are instantaneous, we effectively have a global shutter. So, in the limit of infinitely fast reset and readout times, there’s effectively no difference between a rolling and global shutter. In practice readout times are not infinitely fast, so cameras with a global shutter typically have some sort of charge storage mechanism associated with each pixel that allows the sensor to store charge and (in varying degrees) prevent further charge accumulation during readout.

blog comments powered by Disqus