Artificial light White Balance
The EOS 5D's automatic white balance in artificial light has certainly improved since the EOS 20D, most noticeably in incandescent light. That said it's still not perfect and I personally wouldn't rely on using automatic white balance indoors, it's almost always safer to use a preset (or ideally manual preset).
|Incandescent - Auto WB
Red: 6.0%, Blue: -9.4%
|Incandescent - Incandescent preset WB
Red: 2.1%, Blue: -4.1%
|Fluorescent - Auto WB
Red: 0.0%, Blue: -1.8%
|Fluorescent - Fluorescent preset WB
Red: -1.0%, Blue: -3.1%
Long Exposure noise reduction / Night shots
The EOS 5D has typical 'dark frame subtraction' noise reduction for exposures of one second or longer, this can be enabled or disabled via the camera menu. Dark frame subtraction NR works by taking a second equal exposure immediately after the first but with the shutter closed, any hot pixel noise in this second exposure can then be subtracted from the first to produce a cleaner image. With noise reduction off we were only able to identify two or three 'hot pixels' in our 30 second exposure, switched on these are gone with no black pit artifacts.
|Noise reduction Off||Noise reduction On|
|ISO 100, 30 sec, F9||ISO 100, 30 sec, F9|
The EOS 5D features the latest version of Canon's flash metering system, E-TTL II. This utilizes additional 'distance information' provided to the camera by the lens (although not all lenses provide this). The lens used on the shots below as the EF 24-70 mm F2.8L, exposure mode P. As you can see the results were fairly good, although obviously could have been improved with a little tuning. Flash power was well metered and of course the bounced shot looks the best. No color cast in any of these shots, WB communicated well between the flash and camera.
|Canon 430 EX direct||Canon 430 EX bounced|
|Canon 430 EX direct|
Vignetting & Light Falloff
Vignetting and light falloff aren't something we would normally test in our digital SLR reviews, primarily because cropped sensor digital SLR's and don't exhibit much falloff. However the game moves on and the EOS 5D will put full frame sensors into the hands of amateurs who may well be surprised to see it, especially if they are upgrading from a cropped sensor camera (such as the EOS 20D).
Technically vignetting refers to a darkening of the corners of the frame due to a physical obstruction such as the rim of the lens barrel or a filter, light falloff refers to a reduction in the amount of light reaching the far corners of the frame due to the angle of incidence of the light reaching there. Light falloff is sometimes referred to as cos4 vignetting. In this section of the review we will refer to this effect as falloff for simplicity (and more likely accuracy) however it could well be either or both vignetting / light falloff.
It was my experience of shooting with the EOS 5D that it is certainly possible to see falloff in every day shots, although mostly only at large apertures. In an attempt to measure and compare we improved on a test first designed for our prosumer camera reviews.
These images are then processed by our own analysis software which derives an average luminance (Lum) for the four corners of the frame (5% each) as well as the center (10%), the corners are averaged and the difference between this and the center of the frame is recorded. This value can then be plotted (see graphs below) as a representation of the approximate amount of falloff.Hence falloff of -30% would mean that if the luminance center of the frame was at exactly 100% (pure white) the average luminance of the corners would be 70%. Anything more than -20% may well be visible in everyday shots, although this depends on the framing of the shot and the exposure.
Range of falloff
The chart below demonstrates the difference that these figures above can make, we took the blank wall luminance value of 75 (about 190,190,190 RGB) as our normal level. Remember that these patches are solid and the actual effect of shading is a softer gradual roll-off which would not necessarily be so obvious.
As you can see the EOS 5D fully reveals the 24-70's falloff characteristics, naturally worse at wide angle and maximum aperture. At wide angle (24 mm) you would be able to avoid noticeable falloff from about F5.0 onwards, at telephoto from F3.2 onwards. Put the 24-70 on the EOS 20D and you would never have to worry about falloff, it is below noticeable levels even at maximum aperture. The interesting result here is from the EF-S lens on the EOS 20D which does exhibit what could be noticeable falloff at wide angle from F4.0 to F5.6.
Lenses: EF 24-70 mm F2.8L, EF-S 17-85 mm F4.0-F5.6 IS
Canon EF 24-70 mm F2.8L thumbnail examples
Canon EOS 5D
Canon EOS 20D
Again the EOS 5D fully reveals the falloff characteristics of these lenses, although to be fair this probably wouldn't be noticeable on the 50 mm lens beyond F2.2 or the 100 mm lens beyond F3.5. However the comparison is pretty stark, you would get the same level of falloff on the EOS 20D with the 50 mm lens at F1.4 as you would on the EOS 5D at F2.5.
Lenses: EF 50 mm F1.4, EF 100 mm F2.8 Macro
Canon EF 50 mm F1.4 thumbnail examples
Canon EOS 5D
Canon EOS 20D
Conclusion / Is it really a problem?
So we've been able to observe the falloff characteristics of the EOS 5D combined with a few lenses. The big question is will the affect my everyday shots. The short answer is probably not if you understand that it exists and avoid using wide apertures on certain (mostly wide angle) lenses. It also depends on the framing of the scene, we can clearly make out falloff in the example below because of the consistent swathe of blue created by the sky.
|EF 50 mm F1.4 @ F7.1||EF 50 mm F1.4 @ F2.2|
* Note that if you are shooting in RAW it is possible to (mostly) remove the vignetting effect using Adobe Camera RAW, see our samples gallery for an example