This will be quite a long post, as I try to aggregate and summarize many pieces of information.
Not only addressing the immediate AF issue of the SIGMA ART 18-35mm/1.8 lens on the D500, but also further observations and thoughts this experience provoked.Where it startedMost of the lenses in my cupboard are Nikkor lenses. Nikkors from 1965 to more recent ones. I have a tiny fraction of non-Nikkors, i.e. 4 Sigma lenses, one Tokina and one Tamron. All of them about 10 years or older. The experience sometimes wasn't as good as it could be, so I did not look too often into these brands. To name an example: The Sigma 50-150mm/2.8 for DX cameras would be a nice lens - and sometimes is -, but its extensive color fringing reduced quite often the joy to use it more often.
The second reason is, that my postprocessing workflow is super simple and I basically use only NX2 or D-NX. Software which has good support for Nikon DSLRs and Nikkors, but can't deal well with non-Nikon lenses (distortion correction, color reproduction, etc...). I never wanted to spend time with creating camera and lens profiles, or loading into my software workflow.
The idea: Get the Sigma Art 18-35mm/1.8 and the Sigma Art 50-100mm/1.8 for my D500In the last years, most of my photography centered around FX, but I still liked to use some DX cameras and lenses due to weight, focal range and size benefits. With the arrival of the D500, my interest in the 2 Sigma Art lenses grew, as the 18-35mm had been hailed by reviewers and internet folklore as super sharp, replacing multiple primes, etc, etc .... and the 50-100mm/1.8 looked like a great telezoom for the D500. Even with its limited zoom ranges. There were some comments, that Sigma lenses need more care for AF tuning. So I got both of them and the Sigma USB dock for firmware updates and customization/AF tuning. So indeed, the lenses seem to perform extremely well when all shooting conditions are the right range. The sharpness at open aperture is really impressive - both with the ww lens and the tele zoom. While the telezoom is quite heavy for a carry-around lens, the size and weight of the 18-35mm fits nicely to the D500.
The first quick checksEvery time I get a new lens (or camera), I do a quick check to get a better feeling for its performance. These are not reviewer like kind of tests, rigid setups and the like, just a few shots which help me to make an inital assessment and find some areas I can rely upon, or need to look deeper. Kind of initial assessment.
For the Sigma Art 50-100mm/1.8 the experience was rather positive. Great sharpness wide open, good clarity in normal light conditions, good (but not excellent) bokeh. AF rather at the slow end, some AF offset (which I intended to fix later with the USB dock). The Sigma Art 18-35mm/1.8 was a different experience. Really great in some shooting conditions, but strong decline in other conditions (crosslight, ghots/flare). I thought I had a bad example, went back to my dealer and checked the other copy he had. Same performance. So I kept mine.
Here are 4 of the first issues I saw with the lens:
1) A 100% crop of the upper edge of the frame, approx in the center. Color abberations in strong cross light. At these intensity levels, the abberations are also visible in smaller output sizes.
2) A 100% crop of the lower right corner of a "typical" landscape photo with focus close to infinity. (Similar performance to the AFS 12-24mm which had been critizised for this, but I never heard it for the Sigma. A bit of surprise when I ran into it)
3) This is the center third of the frame, reduced to 1200 pixels vertical. Taken with the sun shade, the sun is on the edger of the sun shade. Strong visibility of ghost/flare in this part of the image, visible contrast reduction as well.
4) 100% crop in the far end of the unsharp area of the fence. Starting at 1500x1000, the bokeh looks rather at the unrest side of the spectrum
5) When I took images indoor and exposure time approached 1/100s, it was relatively harder to get a reasonable sharp image vs. a lens with VR (but less "absolute" resolution). Not suprising, but the gap was quite large. Could be considered as: How many % of the theoretical resolution is available in your average photo? These number seems to be much higher for VR lenses than for the non-OS Sigma Art lens in these shooting conditions. Don't get me wrong, the Sigma Art is a stellar resolution performer, if on a tripod and all factors under perfect control. But this is not always the case in "real world" photography - at least not in mine.
Now I got curious and started to look deeper into the first findings:
With the kit lens (AFS 16-80mm), ghost/flare seems to be much better under control
@ 80mm
@16mm
or a quick check with the AFS 20mm/1.8G
Tuning the AF of the Sigma Art 18-35mm/1.8With this quick & initial experience, my intention was to tune the AF of the Sigma as next step. As described in the initial post, I used the Spyder LensCal as my target and the Sigma software and USB Dock to update the firmware and AF settings of the 18-35mm lens. Sigma offers/requires 16 different correction values (4x focal lengths by 4x distances) for the ww zoom lens and as multiple tests are required per value, it took me more than an hour to adjust the central AF field of the lens (about 100x removing the lens from the camera and connecting to the USB dock and back)
Indeed, the AF seemed to work fine for the central AF field. Great (at least I thought so at that moment)
One of the primary USPs of the Nikon D500 is it very large coverage of the frame with AF sensors, 55 of them user selectable. After adjusting the central AF field, I checked for consistency with the far left AF field in the middle row (of the 5 rows available). Much to my surprise, the focus error was huge. About 10% of the focus distance. If your intended focus distance is 1 meter, the AF field exhibited a front focus of 10cm. Don't know about others, but for me this is beyond the level of acceptable tolerance if I want to use the large set of AF fields available in the D500.
Not knowing where this error was coming from, I also tested the D7200 with the Sigma Art and had more or less the same results like with the D500. I also checked Nikkor lenses like the AFS 14-24mm/2.8 and the focus was much, much better. So the AF system of the D500 seems to be in order.
Sharing my findings and reproduction by othersNot being sure, if I fundamentally did something wrong, I shared my findings in 2 german speaking forums, asking other users if they could reproduce the behavior of my Sigma 18-35mm/1.8. They could. By now, it looks that it is indeed a broader systemic issue of this lens with Nikon DX cameras. To a lesser degree (different focal range) my Sigma Art 50-100mm/1.8 had the same issue. Central AF field working well, the further going out to the frame edge, the bigger the focus gap becomes.
ArendV, I am sorry to read that you could also reproduce the error with your lens, but it is further confirmation about the issue.
Some users reproduced the results with lenses which they had sent to Sigma service with their camera. This indicates that Sigma Service doesn't do AF calibration with all AF fields in a camera, just the central AF field. Similar to the process a user does at home with the USB dock.
Some users also reported the AF issue with the Sigma Art 30mm/1.4
Some users reported the issue with D7100, D7200 and D500 cameras.
Potential reasonsAs of now, most of these thoughts of potential reasons are based on observing, guessing and reasoning - they need to be verified (or falsified).
My current guessing is that Sigma still has no proper licencse from Nikon for the f-mount. Less so the mechanical part (these patents seem to have expired), but more so the electrical/electronic part. As the innovation moves from HW centric world to a SW centric world, Nikon is relying and introducing more and more features which requires an every increasing interaction between the lens and the camera. What started as a simple protocol between the camera CPU and the lens CPU (LensId, min/max focal range for zooms, min/max aperture, ...) gets more and more the center piece of product integration (think PC-E lenses).
For AF tuning - there are 2 different approaches:
Nikon offers its customers AF finetuning with the camera firmware taking care of the AF tuning for a particular lens.
Sigma offers its customers AF finetuning with the lens firmware/CPU taking care of the AF tuning.
I do think, that the key difference is the concept of LensId in this integrated world going forward. With a few exceptions, the existing LensID's are assigned to unique optical designs in Nikkor lenses.
Here are a few databases on Nikkor LensId's:
http://www.rottmerhusen.com/objektives/lensid/nikkord.html and
http://www.turknikon.com/nikon-lens-numarasi-nikon-lens-id-1979When a Nikon camera can read the LensId of a particular Nikkor lens type, it "knows" the optical and electronic capabilities. Color index due to coatings, field curvature, Distortion, etc .... Doesn't need to come from the lens itself, a look-up table in the camera firmware for the different lens attributes is sufficient - with the LensId as the index into this table based information.
Like in the past with other Sigma lenses, Sigma Art lenses are reusing existing Nikkor LensId (there is to my knowledge no dedicated Sigma LensId's available). In the case of the Sigma Art 18-35mm/1.8, the lens reports itself as LensId "139". BTW, this is for Nikkor lenses the Nikkor AF-S DX VR Zoom-Nikkor 18-200mm f/3.5-5.6G IF-ED. So, from a camera CPU perspective the Sigma Art lens is considered to be a Nikkor Superzoom. The camera CPU has no way of knowing that a Sigma Art lens is mounted.
On the other side, the Sigma CPU has no way to get the information, that it is mounted on a (i.e.) D500. I am not aware that there is a command in the F-mount protocol that a lens can query this information. So the Sigma lens has no information, where for instance the AF fields are located in the frame of the particular body.
The Nikon AF tuning in the camera firmware with a Nikkor lens can at least combine multiple information sources to support the user with the AF tuning process. 1) The camera CPU knows its own AF fields, position, calibration info, etc ... 2) It know's which lens is mounted - via the LensId, 3) it can look up further optical/electronic descriptions of the mounted lens in its internal firmware database
The Sigma AF tuning process with the USB dock lacks the camera information. The sigma lens never knows, on which camera it is mounted. That's the reason, Sigma can only adjust its lenses for one body. Not for multiple cameras.
StatusI did return my 2 Sigma lenses, as I didn't want to spent too much time adjusting, sending in, getting back, checking, repeat as needed, etc ...
But the much deeper reason why I sent it back. Currently I don't see how this could get better with the current separation of information Sigma and its lenses have in the Nikon ecosystem. I did check a handful of Nikkor lenses on the D500 and their AF accuracy accross the frame where just great. I haven't had for a long time a body like the D500 on this matter.
The second issue is, that if I would build up a larger arsenal of Sigma Art lenses, every time a new Nikon camera comes on the market, the process starts from scratch. Not a compelling vision.
The third issue is, that the missing camera information in the current Sigma AF tuning method effectively limits accurate AF settings to one camera body. Too bad for those who want to switch their Sigma Art lenses between 2 or more bodies.
I do rather plan to spend my time doing photography, versus trying to become an expert in lens/camera interoperability between Nikon and Sigma.
A few users in the german forums sent mails to Sigma and asked for further information.
If there is interest here @NikonGear, I can provide an update in the future.
Closing remarksThe Sigma Art lens 18-35mm/1.8 is quite long in the market and has been extensively reviewed by the reviewer community. Why has this issue never been reported?
A few thoughts:
1) Are only the most recent camera models affected?
2) Do users for all practical purposes focus only with the central AF field?
3) Did users calibrate the Sigma Art lens with the USB dock on the central AF field and never thought that potential focus error with the non-center AF fields could be caused by the lens? Due to the broad internet folklore: The Sigma Art is sharp, sharp, sharp, sharp. So any focus error must be user error in their view.
4) Do user really use LiveView almost exclusively and aren't affected by the AF issue?
On one side it is a pity. The lenses are good optical performers. But I do not want to change my DSLR mode to LiveView only, just to accomodate the peculiarities of the Sigma lens interfaces and AF accuracy. Even if the absolute and theoretical resolution of some Nikkor lenses are below the values of the Sigma Art lenses, this advantage is of limited use, if the AF isn't working properly in daily life (test and reviewer scenarios exempted)
rgds,
Andy
