You know I've had issues with the DXO lens reviews for a while. I didn't pay attention to them at all until they posted low-ish scores for some lenses I personally knew to be stellar optics. These optics I would always choose to use over alternatives they ranked higher.
Since then my thoughts have been that they cannot properly assign value to the contrast delivered with the resolution. Because that is what I value most, in terms of optics, then their reviews aren't super useful to me. They also ignore physical aspects of lenses pretty thoroughly but that is stated up-front so I don't begrudge them that.
Zoom forward to now, and with their latest Sigma 85mm EX posting they said what I expected them to. They found it to be one of the best lenses they've ever seen. Indeed, I mentioned extremely high sharpness in my own review before basically failing the lens on contrast and usefulness in difficult lighting situations. So we have another DXO review I find of dubious value.
The Sigma 85mm f/1.4 EX is only the second example of a new breed of fast prime lenses from Sigma, recently joining the 50mm f/1.4 EX to form what now looks like a promising new lineup. We typically expect Sigma lenses to underperform compared to Nikon or Canon lenses with an offer of performance proportional to asking price. On most metrics this new lens is a delightfully solid match however – in fact there is little to fault the optical cell of this 85mm lens – and for most photographers the verdict will come down to secondary qualities of off-brand lenses.
Summary: very very sharp, middling contrast, the first copy I received had a permanent focus jam.
DXOMark data is out for the Sigma 85mm, and it both augments and confirms the SLRGear results. It has extremely even performance across the image field, no distortions, low CA.
It's the highest-rated lens on the 5D Mk-II on the DXO site, narrowly edging the Canon 85mm f/1.8. Resolution looks to be between 11MP and 17MP on the 5D Mk-II if one calculates from lppm numbers. This may be near what that camera's AA filter allows anyway, as no other lens has delivered substantially better than that yet (70+ lenses tested).
Reading list is here.
I’m researching this lens as a possible indoor-sport lens that is compact and inexpensive, and can play at least one other role (long normal, portrait). I haven’t been looking for long and I can already say that it has garnered quite a bit of respect. For $900 it looks like there is a lot of value here.
MTF chart and links after the jump...
Several years ago I spent some time shooting with a Sigma 70-200mm f/2.8 lens. It had poor contrast a full aperture, quite soft at 200mm, and obviously the absence of VR limited its usefulness at indoors venues. I disliked it enough that it convinced me only the Nikkor VR lens (back then, version 1) would do for me. That Sigma lens is probably 10 years old now, but to this day it still sets how I perceive Sigma’s medium telephoto lenses.
Zoom to today, so to speak, and enters the new lens.
Looking at them by themselves and ignoring the competition, these charts look pretty good. There is more weakness at 70mm than I would like to see, with a pretty sharp drop to minimal contrast at the corner, but the performance at 200mm looks simply excellent. Possibly this means that there is significant light fall-off at 70mm and that drives down contrast.
If I compare with the previous Sigma version (not shown here) there has been a exchange between lowering performance at 70mm and improving performance at 200mm. This is probably the right choice, not only because this lens is most likely to be used at longer focal lengths, but also because it was done in such a way that DX shooters (or teleconverter users) will never notice. This new lens looks very strong for those photographers.
Compared to the latest Nikkor lens
The obvious, somewhat unfair contender for the Sigma is the $2399 lens from Nikon, also a new design from the last few months.
This lens is the reigning champ, but it also shows some weakness towards the border and corner. What is impressive with this lens however is how even the performance is across the zoom range.
Let’s see how the Sigma compares:
At the wider end of the zoom range the Sigma is beat by an absolute 10% edge in favor of the Nikkor, widening to a 20% edge and relatively twice the contrast at the corner. There’s no doubt the Nikkor is the better lens at shorter focal lengths. Now zoom towards 200mm and the picture changes, the Sigma roughly matches the Nikon but improves upon it significantly towards the corner.
It’s clear that the Sigma is a serious contender. If money is no object to you then you should just get the Nikkor and never look back, but everyone else should give the Sigma some thought. Price is going to be crucial to the success of the Sigma lens.
Compared to the previous Nikkor lens
The old version of the Nikkor is still around and is a spectacular performer on DX and teleconverters. This lens is still a contender for many, even if it has severe issues with FX sensors.
You’ll notice immediately that this lens also has very weak corners at 70mm so this should be a close comparison.
Across the range the two lenses are very close, with one important improvement to the Sigma. At 70mm the Sigma is only very slightly behind – this is a focal length where I thought the Nikkor excelled and seeing the Sigma meet that is exciting. At 200mm the Sigma lags the Nikkor in the center but completely outclasses it at the far border and corner, where the Nikkor is very weak.
This new lens is – in my opinion – most easily described as an improved version of Nikon’s 70-200mm f/2.8 VR version 1. This summarizes the reason for this lens to me. Sigma is caught up to just one step behind Nikon now.
This morning I compared on paper the new Nikon 50mm f/1.4G and Sigma's 50mm f/1.4 EX HSM. That was the plan anyway. I started by overlaying the MTF charts, but then got greedy and overlaid a whole bunch of charts together until I landed with this mess...
Yes, I interpolated the Leica and Zeiss charts in MS paint too. ;^)
The word is out that Nikon's not so proud of their new Fifty at the moment. At least that's what Nikon's body language is saying: it's supposed to go on sale in a little over a month's time but they don't want to let the press touch it in America. Or even take pictures of it from behind a glass box. That smells.
My take from this is that they hadn't predicted Sigma's move and played it too safe. They must have known this for months, I don't know why they went ahead and announced a loser into the competitive marketplace. The least they should do now is delay by a few months and announce that production copies of the lens will be nano-coated. From a brand image perspective that would deflect some of the pressure by tying-in a popular new feature, and deliver more customer confidence with every box. "Who knows what coatings Sigma uses?" would be message between the lines then.
Now that Dpreview and Photozone have both had a chance to take a look, here’s a mix of results from “Sigma Fifty” and its Canon/Nikon competition. These are the FX results from Dpreview. I’ll briefly mention DX (incl. Photozone's) results below.
Wide open the Sigma Fifty pretty clearly dominates. This is what it was designed to do and why it’s such a big bucket of a lens. The poor performance in the corners is almost certainly due to a 2-stop (!!) light fall-off (colloquial vignette). Don’t be too bullish about the Canon or Nikon either, they have 2.7 stops of fall-offs.
The Nikon lags throughout the test but here at f/1.4 the gap is just staggering, especially when you consider that the Sigma/Canon are paired to a 21MP 1Ds-III and the Nikon is on a 12MP D3.
Both the Canon and Nikon improve tremendously around f/1.8 but the Sigma continues to rule the pack through f/2~f/2.8. The Sigma’s center resolution here is just enough to quench the DX format over most of that frame. Light fall-off is now ~1 stop for the Sigma and ~1.5 stops for the Canon/Nikon.
Stopped down all three are solid performers, but the Sigma is at its limits on the 1Ds Mk-III. Somewhere around f/4 the Canon comes out way ahead. It looks like the Canon 50mm f/1.4 has resolution to spare up to ~30MP beyond this aperture. The Nikon has just a bit more resolution still in it, but recall that it’s sitting on a 12MP sensor only here (but colors are normalized so it’s still apples-to-apples there). Light fall-off is nothing to worry about from f/4 onwards for any of them.
What’s the DX performance like?
It’s a mixed bag. Used stopped-down the Sigma Fifty is just good enough to quench DX sensors, and at any aperture has even behavior across the whole frame without any light fall-off. Wide-open it’s not quite as sharp as the Sigma-30 however.
Comparing the results of the Sigma .vs. the Zeiss in the Photozone tests (note: crossing shooting platforms hurts comparison) I believe they are roughly equivalent for most purposes. I would select the Sigma for center-axis sharpness at f/1.4 but probably pick the Zeiss for anything else.
A word about the Bokeh
Back a few months when this lens was announced I was a bit worried about the bokeh, because it's typically not Sigma's strong suit. You’ll want to check out this small collection of images then: http://www.motleypixel.com/reviews/index.htm Certainly that’s not bad. Based on these and other images (on Flickr) I conclude that the bokeh is marginal at f/1.4 but improves suddenly at f/2. From f/2 onwards the bokeh is soft and “relaxed”, I can’t find anything bad to say about it.
A little creative ms-painting to overlay the 30/1.4 and 50/1.4 MTF charts:
Despite the lower MTF in the center, I think the 50/1.4 is a lot more lens. To be honest I'd like to try the 50/1.4. I might try it later this year or early next year.
More likely though, I might sit tight until Nikon ships a new 50 and decide. Maybe we’ll see a Nano-coated f/1.2 AF-S come out. I’m betting we will - that'll be an interesting day.