5 February 2009

Canon E-TTL II flash: why you should use Canon lenses

Could your choice of lens can cause instability in the flash metering of your camera?
If you use Canon lenses they talk to your camera body providing all kinds of information to help the camera determine the correct shutter speed to reduce camera shake, the maximum and minimum aperture of the lens and also in many cases the distance to the subject. This communication is Canon's own protocol, and whilst third parties have re-engineered it with some success did they get all the information right?

So how does the lens affect the flash?
Canon's E-TTL II flash metering added the use of distance information to the calculation of the required flash output power where the lens provides distance info. Some older Canon lenses don't provide distance information so the algorithm has to work without this valuable information to retain compatibility. Consider what happens if some third party lenses appear to provide distance information, but actually don't provide it in the way Canon expects it. Could be as simple as Canon lenses tell the distance in meters, Sigma maybe in feet. So Canon lens says subject is 4 meters away, approximately 13 feet. Sigma may say 4 feet away or 1.2 meters! The camera just gets 4. You can guess how this info may make the flash exposure wrong. It could be worse though; what if the Sigma always said the lens was focussed at 4 meters. Sometimes it would be right, sometimes wrong giving inconsistent results. With under exposure, over exposure and correct exposure at different subject distances.

We've heard several times of the two Canon EOS users with the same flash, same camera and different lenses having random flash exposures on the non-Canon brand lenses. What's more swapping lenses moves the problem too.

With exclusively Canon lenses on hand this is a tough one to prove, but we welcome your feedback in the comments.


1 comment:

Anonymous said...

I'd actually mark this article down as scaremongering. Though I do appreciate that Sigma et al are reduced to reverse engineering the Canon technology, I think they get much further than you give them credit for. Bear in mind that, apart from distance information, the lens also has to understand focussing instructions, report aperture, report model information (for the EXIF data) and report focal length in use (again for EXIF). My Sigma lenses seem to get all this information correct; whether they report distance I don't know, but they seem to get everything else right so I have a high degree of confidence they could get that right too.

Further to this, I find the argument about subject range and flash exposure slightly spurious because:

a) Even with a distance-reporting Canon lens, a 580EX mounted on the camera still has a large tendancy to underexposure.

b) I've read elsewhere, and maybe you can correct me, that the lens distance information is not always very accurate. Depending on the lens, distance reported may be plus or minus several metres - easily enough to break a flash exposure.