Canon 1D Mark III Incandescent Lighting

Canon EOS-1D Mark III (Prototype) Imaging Characteristics
High ISO, incandescent light source

NOTE: All Canon EOS-1D Mark III images shown here are from a prototype camera: production models may show improved image quality, please don't view the results shown here as final!

By Dave Etchells
Several readers expressed interest in what the Canon EOS-1D Mark III's high-ISO performance looked like under a more challenging light source than our studio HMIs. - They very rightly pointed out that daylight-balanced lighting represents something of a best-case in terms of high-ISO image noise. By contrast, household incandescent lighting (about 2800K color temperature) is a very tough light source to deal with, because its very warm-toned color balance requires a significant boost of the blue-channel gain, the channel most susceptible to noise.

While I want to avoid shooting all our test shots with the prototype sample, and then having to go back and re-shoot all of them again once we get a production model to work with, I felt that the issue of image noise under difficult lighting was an important one to address, important enough to justify the added effort involved.

I shot my long-suffering wife Marti in our "Indoor Portrait" set, which is primarily lit by 60-watt household incandescent bulbs, with fill lighting provided by a pair of 120-watt floods shining through cloth diffusers. This is indeed a very warm-hued light source, measuring about 2800K on my Minolta color meter, a value that's supported by the good color balance produced by the Mark III when I set the white balance to 2800K, as seen below.

I shot this series with the Canon 50mm f/1.8 II lens, stopped down to f/5.0, where it's very sharp. The crops below show the results from the Mark III next to those from the 1D Mark II N, shot with Canon's excellent EF 24-70mm f/2.8L lens, stopped down to f/3.5.

I've inserted comments below the crops themselves, but the overall conclusion is pretty clear: While the 1D Mark II N turned in a really excellent performance, the Mark III clearly betters it by a noticeable margin. Trying to pin it down to a specific differential based on these shots is a judgement call, but to my eye, it looks like a good f-stop's worth of improvement.

Hmm - what about that f/5 aperture I was using with the Mark III though? After I'd shot the above images with the Mark III, it occurred to me that some readers might be concerned that the f/5 aperture would be penalizing the Mark III somewhat, because there after all would be less light hitting the sensor. I really didn't think that this would be an issue, since these exposure times were all in a range where dark-current leakage shouldn't be a significant factor. (In other words, only at longer exposure times would the length of the exposure itself make an appreciable difference in the amount of noise present.)

Rather than leave the results in question (not to mention having to deal with the flood of reader emails), I decided to re-shoot the series, this time using our Canon 100mm f/2.8 macro lens, shot wide open. (At the moment, this is arguably the best lens f/2.8 prime that we had laying around.) The crops below compare the results with this lens at f/2.8 with the results from the 50mm at f/5.0.

Interestingly and somewhat vexingly, while I adjusted the exposure time the same number of stops (five-thirds stops) as the aperture was opened, the 100mm f/2.8 shots came out slightly darker. - This raises a question I've long wondered about, but have never explicitly tested: How well calibrated are different lenses with respect to each other, in terms of how much light they let through at a given f-stop? It appears in this case that the 50mm lets in somewhat more light than its f-stop would indicate, relative to the 100mm.

Despite the difference in brightness levels, looking at similarly-toned parts of the image, it seems pretty plain that the differences in exposure time really make little or no difference in the noise levels seen.