Points in Focus Photography

Canon EOS 5D mark III Review

Videography Performance

Video quality is a hugely complex undertaking to make any kind of serious discussion on, and likely if I went into serious detail including all of the necessary testing, it’s likely that I would more than double the size of this already sizeable review.

Shooting Video

Shooting video with a 5D mark III is a decidedly cinemaesque shooting experience. Simply put, the camera does little to assist you in shooting and where it does provide automation the automation is often hamstrung in frustrating ways.

I don’t fault Canon for focusing the design on manual control. The 5D mark III, like the 5D mark II before it, is clearly targeted at serious users, not mom’s and pop’s looking to shoot video of their kids at soccer practice. When it comes to serious users, we’re already talking about manual everything, shutter speed, aperture, ISO, focus, the works.

That said, I’d almost rather Canon have simply disabled autofocus and auto exposure modes instead of giving users the implementations that they did.

Auto Exposure

The auto exposure modes are available while shooting video. However, I can’t really recommend using any of them in practice. All auto exposure modes force the camera to operate in auto ISO, since the ISO is smoothly adjusted to allow for smooth transitions in the video.

The auto exposure algorithm appears to favor adjusting the ISO over altering the shutter speed or aperture. The latter being adjusted only when the exposure change is much larger than auto ISO range will allow.

Transitions, at least over reasonably small (4-5 stops) ranges are smooth. Though keep in mind since this transition is achieved by way of changing the ISO, the noise characteristics will change as well.

Auto exposure tarnsitions over a 2 stop change in scene brightness.

Auto ISO

Auto ISO is available in all video exposure modes, and it’s the only ISO mode available in the auto exposure modes. Potentially this could be the most useful method for automatically addressing exposure changes, as it won’t affect the depth of field or motion blur. Moreover, since the 5D mark III has quite a bit of ISO latitude before noise starts becoming a consideration, there’s a significant amount of range to work with before the video will be strongly negatively impacted.

However, the biggest problem with Auto ISO is when you pair it with manual exposure mode—where it would simultaneously be most useful for doing the auto exposure work. This would be the obvious pairing given the limitations imposed by the normal auto exposure modes. However, Canon didn’t see fit to provide a mechanism to adjust the exposure compensation for auto ISO while in manual exposure mode, making adjusting for darker or lighter scenes impossible, and therefore rendering the feature largely useless.

Manual Exposure

Ultimately, at least in my experience, the best quality video is achieved in manual exposure mode with a manual ISO set. In some ways, this really limits what you can do with the video features on the 5D mark III.

Moreover, in manual exposure mode, changes in ISO, shutter, or aperture all take effect instantly; there’s no smooth transition. This means you can’t effectively change the ISO manually to compensate for changing conditions.

In my experience, the best option for smoothly dealing with changing conditions is to use a variable ND filter in conjunction with manually set exposure that won’t need to be changed when moving from one environment to the next.

Manual exposure transitions in 1/3rd stop increments for 2 stops.

Focusing

The 5D mark III offers one-shot contrast based autofocus while in video mode, and there is no video servo AF option. However, the utility of the AF system is hampered in various ways.

To start with we’re talking about contrast based AF with no directional queues. The results are the characteristic follies seen in contrast based focusing. This starts with the 50% probably that the camera will initially move focus in the wrong direction and ends with the characteristic overshoot and return when it gets close to focusing. All told, in my opinion, there’s nothing more armature looking in video than the characteristic SLR contrast AF routine.

Activating the autofocus system while recording video. Note this scene is deliberately underexposed. 

The situation with autofocus is compounded by the camera’s apparent insistence that focusing be done at the maximum aperture instead of the currently selected one. Yes, focusing at the maximum aperture insures that the focus is most accurately placed, however, when shooting video the aperture adjustment and the resulting change in exposure is even more distracting than the contrast AF hunting.

All told, I would argue that it’s best not to use the AF system on the 5D mark III when shooting video at all. This is especially true with ring-USM and STM lenses, where the AF system doesn’t cause the focus ring to rotate, which in turn means that any marks or pulling aids you might use will no longer be correct.

Video Performance

How does the 5D mark III stack up when it comes to video?

Improved video capabilities were inarguably one of Canon’s primary design objectives with the 5D mark III. The design objective is most obviously seen in the architecture of Camera’s new sensor. The sensor in the 5D mark III is unique in that it’s supposedly been designed specifically to aid the capture of 1080p video. Canon has achieved this by making the sensor exactly 3x 1920 pixels wide. Thus, the camera can read out 3×3 blocks of pixels to get the full RGB data for a single pixel in the video.

What’s interesting about this technique is that in some respects it seems to be working quite well. Moiré is a non-issue. Moreover, the sensor seems to obey diffraction based softening as you would expect for a camera with 18.75µm pixels instead of 6.25µm ones.

Video Diffraction Test

I did this test back in April of 2013, and the image doesn’t lose any appreciable resolution as the aperture tracks up around f/28.

However, when it comes to resolution, the video resolution is a not nearly as satisfactory as the entire design leads one to suspect it should be, and I’m not sure why. In most tests, including my limited testing, the 5D mark III records about 800 lines of resolution in 1080p mode. Moreover, looking at the uncompressed HDMI output at 100% resolution this doesn’t appear to be limited by the compression codec, but something else in the system.

Some have speculated that resolution loss is due to the AA filter. I don’t believe this is the case at all. The AA filter on the 5D mark III is tuned for the still camera operation, which is nearly 3 times (linearly) higher resolution than the video resolution. Moreover, loss of resolution isn’t seen when looking at a 1920×1280 JPEG using the camera’s JPEG preset, nor should it be given that that’s the still resolution.

Video resolution test – The left half of the frame is 1080p video recorded by the 5D mark III, the right half of the frame is the in camera produced 1920×1280 JPEG. (Note my resolution chart is of poor quality and the exact numbers should not be used for measurements.)

My suspicion is that Canon is either doing something more complicated with their video readout than their comments have implied or they’re simply not recording a 1080p frame when shooting video. The latter suspicion may be at least partially backed up by what the developers of the Magic Lantern supplemental firmware have documented.

That said the overall output of the sensor when recording video is still quite good, especially at high ISOs.

The 5D mark III continues to use H.264/AVC compression for the files saved in the camera, though new to the 5D mark III is an all inter-frame mode (ALL-I). Variable bitrates are used, with the IPB mode averaging 30-35 megabits/second (Mb/s) and the All-I mode averaging 90-100Mb/s.

Canon recommends compact flash cards write at a minimum of 10MB/s (megabytes per second) for IPB, and 30MB/s for All-I. If using SD cards, Canon suggests that write speeds can be about 30% slower. I’m not sure, if this discrepancy is due to the differences in interfaces, or if the camera uses a less aggressive target bitrate when paired with SD cards. However, in either case, faster cards allow the camera to hit the maximum bit rates when needed which is desirable.

When the best quality is needed, the 5D mark III supports clean uncompressed 8-bit 4:2:2 output over HDMI. Though you’ll need an external recorder to take advantage of this.

The biggest consideration when shooting video is that it’s necessary to think about it more like shooting JPEGs than shooting RAW. The highly compressed files just don’t have a lot of latitude for adjustment after the fact. You need to nail white balance when shooting, as doing any kind of significant color, grading in post to correct white balance issues shows the limitations of the compressed files.

The same can be said about establishing the look. Though there are a lot more schools of thought on this and a significant number of options on how to go about this. One argument is to get the look as right in camera as possible. Many people who support thing recommend using a toned down version of the portrait, neutral, or faithful picture styles. Their argument is that since there’s so much less leeway in the files for post processing, getting the look right in camera where it’s baked into the compressed video file gives you a better image to work with.

The other train of thought is to use a flatter picture style, similar to the way Canon Log works, and then add the contrast back into the picture in post by applying a color LUT or curve adjustment to the files. The supposed benefit of using a flat picture style is the ability to extract more detail in the shadow areas that would normally be crushed by using the default Canon pictures styles. The con, of course is that since the contrast is being flattened more processing must be done in post, which can bring out the noise inherently baked into the images.

For the most part, I shoot video using Technicolor’s CineStyle picture style and then apply a curve to it in Premier Pro to recover the contrast. I’ve looked at using less flat picture styles that would let me drop some of the post production corrections, but I’ve yet to find one that I’m really happy with.

The biggest advantages I’ve seen in the 5D mark III’s video is the ability to really crank the ISOs up without getting a tremendous amount of noise. Almost all of the video I’ve shot for Points in Focus of various products, I’ve shot at ISO 2500 or 5000.

That said, one of the biggest complaints I have with the 5D mark III is the lack of any kind of metadata being saved when shooting video. If you want to know what ISO, shutter speed, aperture, lens, or focal length something was shot at, you have to manually slate the video. Ideally, Canon would have stored that information either in the video clip or in some kind of sidecar that went along with the clip. But as it stands, since I don’t remember or sometimes can’t slate, it’s impossible for me to identify any settings after the fact form video I’ve recorded.

Comments

There are no comments on this article yet. Why don't you start the discussion?

Leave a Reply

Basic Rules:
  • All comments are moderated.
  • Abusive, inflamatory, and/or "troll" posts will not be published.
  • Links to online retailrs (eg., Amazon, Ali Express, EBay, etc.) either directly, or indirectly through 3rd party URL shorternrs, will be removed form your post.
  • Extremely long comments (>1000 words) may be blocked by the spam filters automatically.
  • If your comment doesn't show up, it may have been eaten by the spam filters; sorry about that.
  • See the Terms of Use/Privacy Policy for more details.

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.


Follow me on twitter for updates on when new comments and articles are posted.

Email Notice Details: By checking the above checkbox, you are agreeing to recieve one email at the email address provided with this comment, for the sole purpose of notifing you that the article author has reseponded to your comment.

Our cookie and privacy policy. Dismiss