Canon EOS 5D Mark III Digital Camera Review$3,499.00
We found that the Canon 5D Mark III had no problems rendering motion effectively in either of its compression modes (IPB or ALL-I) with minimal noise and little artifacting both ways. The 24p video in our testing did not greatly differ from the 30p to our eyes, though the 24p recording did retain a little more of a filmic "flicker" outside of our controlled lab settings. We noticed some ghosting in our motion example, especially around the monochrome pinwheel, but it's a common effect that we are used to seeing on DSLRs. One aspect that is not improved is the rolling shutter effect, preventing the effective use of quick pans or capture of truly fast moving subjects. That's less a problem with software than it is the simple nature of using a rolling shutter on a CMOS sensor. Until global shutter technology (as used on most CCD sensors typically found in camcorders and high-end video cameras) catches up to the point it can be used on these sensors, we don't anticipate much improvement in this area for DSLRs. More on how CamcorderInfo tests motion.
Compared to the Mark II, motion is rendered quite similarly. One big advantage is the ability to select between IPB and ALL-I compression, as opposed to just the one compression type on the Mark II. You also have the option of selecting a 720/60p mode if you are planning on slowing down action in post-production. Out of camera the two motion examples look quite similar, but the Mark III will set you up better for editing after the fact.
The Canon 5D Mark III is not substantially sharper than its predecessor, the Mark II. Shooting a standard sharpness chart, we found that the Mark III was able to render, at best, around 700 line pairs per picture height (LPPH) of sharpness horizontally, and 750 vertically with its IPB compression (utilizing a 50mm f/1.8 lens at f/7.1, the kit lens was less sharp, getting around 600 LPPH horizontally and 750 vertically). ALL-I compression actually looked a little bit softer here, though both types lost most of their sharpness when the camera was moving.
One big area of improvement over the 5D Mark II is the appearance of a moire effect caused by sampling errors in downsampling from a large, 22-megapixel image sensor to an HD video signal. This was a major issue in the Mark II, with the camera employing line-skipping processing in order to achieve the required speed. The Mark III is faster, and as a result is able to keep moire to an absolute minimum. The improvement is substantial, with the resulting footage able to stand up to subsequent editing much better than the video from the Mark II. If there's one reason to upgrade from a Mark II to a Mark III, this is it.
While this isn't a 100% crop, we took 1080/30p footage from both the Mark II and Mark III with a Canon 50mm f/1.8 lens. Running the footage through Adobe Premiere Pro CS4 with the standard sharpening effect applied, it's clear just how destructive moire is on fine patterns with the Mark II. While shooting a high frequency test chart is a worst-case basis, the limitations on the Mark II are pretty severe, especially if you're planning on using the footage alongside high-end cameras or for professional work. More on how CamcorderInfo tests video sharpness.
Under low light conditions with the standard 24-105mm kit lens we saw no difference in sharpness under low light (60lux conditions). This is despite the camera having to increase ISO to compensate for the drop in ambient light. Under these conditions we saw the same 600 LPPH horizontally and 750 vertically with IPB compression as we did under bright light conditions.
Low Light Sensitivity
With ISO set to automatic, the 5D Mark III was able to render a visible image (defined as a white section in the center of the frame hitting 50 IRE on a waveform monitor) with just 6 lux of light. At this point, the camera was forced into using its maximum ISO for video recording, which is 12800 (6400 is the standard max ISO, but that can be pushed by activating ISO extension for stills). Even at this light level, noise was apparent, but didn't overpower the image. We wouldn't recommend using this setting in any feature films, but when you need video and light isn't there, the Mark III can do the job.