I've been testing out the exciting, new Sigma 85mm 1.4, and as unworthy and unqualified as I am to be a professional lens tester, I feel compelled to informally share my findings...
1.) When it arrived, I thought I had a dud: AF would not stop hunting, nor would it lock on to anything, even high-contrast subjects. It simply jerked back and forth as though it were suffering from a seizure, and didn't stop unless I took my finger off the shutter. This lasted for about 20-25 minutes. Has anyone experienced this?
2.) After 20-25 minutes of AF madness, it literally seemed to "snap out of it," and the AF has been nothing short of stellar since. Simply no more hunting or convulsing. Nearly as fast as the EF 85 1.8, but tremendously faster than the 85L II. It now locks onto everything and does it quickly, even in near darkness. This is simply astounding. (This is using center point on a 5D2; the outer points are useless, but this is true for pretty much any other Canon-brand lenses, too.) This Sigma's AF locked onto things in mere candlelight that the 85L II just gave up on.
3.) The AF on this beast seems to be as accurate, or perhaps even a little bit more so, than the 85L II. Incredible work, Sigma!
4.) Biggest concern for me at the moment: The Sigma seems to consistently underexpose by at least 1/3 to 2/3rds of a stop compared to the 85L II. In shooting both at f/1.4 on aperture priority, the Sigma seems to tell the 5D2 to use a far higher shutter speed. Shooting on full manual eradicates this problem...but sometimes I like to use aperture priority in certain situations. But seriously, I wonder if the Sigma is communicating differently to the camera body than the 85L. I would prefer not to have to punch up the exposure in post for every shot, or to have to use exposure compensation every time I mount this lens (because I probably won't remember to).
5.) The Sigma is super contrasty...almost too contrasty. This is subjective, however. I feel as though I would prefer to add contrast in post, rather than have the lens do it for me. But that's just me.
6.) This is a pretty remarkable lens. It truly blows the EF 85 1.8 out of the water, IMHO. It handles better and feels great in the hands. Robust build, solid hood, zippy AF. The bokeh is astounding and is on par with the 85L (again, subjective). Sharpness is definitely on par with the 85L at 1.6 and above, and trails almost imperceptibly to the 85L wide-open at 1.4.
7.) I still can't help but wonder if that AF problem I experienced upon opening the box will come back to haunt me later.
UPDATE 11/26/10:
Here is an update after testing it extensively for a week...
1.) The convulsive AF that I experienced only on the first day has, unfortunately, returned on three separate occasions. However, fortunately, it lasted only brief seconds each time, and the lens resumed normal function. Even these brief AF "seizures" erode my confidence in this len's future performance and reliability.
2.) Completely unscientifically, I've decided that the Canon 85L II's AF isn't that much slower than the Siggy. When going from MFD to infinity, yes, it takes about twice as long as the Siggy; when within reasonable relative distance, however, it's not bad at all -- far better than people on net forums make it out to be. The Canon's AF locks on firmly and confidently, albeit slightly more deliberately; the Siggy's AF zips to the near AF-lock point, adjusts back-and-forth extremely quickly for a hair of a second, then locks. They're like two entirely different personalities: one older but more experienced (Canon), and the other an adolescent rushing in for the kill but not quite aiming right (Siggy)...if that makes any sense.
3.) At identical exposures on full manual, the Siggy still consistently underexposes by about 2/3rds of a stop. Combing through online forum comments have revealed that the Sigma quite possibly absorbs more light than the 85L due to either its increased number of elements (more elements = more light loss) and/or its coatings. I suspect that most buyers seek a fast 85mm for its brightness and here, for some reason, the Sigma fails.
4.) Lastly, my biggest concern, and my reason to not recommend the Siggy: Its metering communication with Canon's bodies is NOT RELIABLE. My exposures vary quite wildly, with the majority of them being on the darker side. Wildly = 3+ stops in some instances! There's no way to boost the exposure in post without significantly degrading the image. After laboring through hundreds of forum comments on DPreview and P.O.T.N., I've come to the conclusion that there are possibly some errors in Sigma's reverse engineering especially in regards to electronic communication between the lens and camera body. There are also some very incredulous conspiracy theories that Canon actually boosts the exposure in-body BEFORE the file is recorded into RAW for certain (only) Canon lenses that the company knows are prone to light loss. (?!?!) I don't get it, but search the P.O.T.N. for the Sigma 85, and you can read all the juicy comments yourself. At any rate, my Sigma 85 is causing my 5D2 body to meter unreliably, which isn't a good thing, IMHO.
Oh, and yes, the Siggy did arrive needing microfocus adjustment, which on my 5D2 was +13. Just FYI.