Sony A9 vs Techart Adapter vs Leica SL Autofocus Comparison - Part IV
Originally, I had wanted to publish this post on the Sony A9, while I was still in Hong Kong. I didn't want to be stuck in a situation where jet lag would render me an incoherent basket case. Unfortunately, life got somewhat real days before my departure, which prevented me from reaching my desired timeline. So, now I'm in New York, trying my best to finish this post under a fog of diminished capacity.
And I came so close too.
Before I continue, it might be useful to explain the relevance of comparing the autofocus between the Leica SL and the autofocus of Leica M-mount lenses adapted on the Sony A9. After all, it does seem rather frivolous, since it would be like comparing an apple to an orange that has been modified to be like an apple. Logically, what purpose would such a comparison serve? The apple will always be better than an orange under those testing conditions, while the orange will always be better as an orange.
You see, for many Leica rangefinder enthusiast, the dream of autofocusing Leica M-mount lenses has always been the unspoken holy grail. It has been, ever since the release of the Minolta 7000, the first SLR with an integrated autofocus system. But sometimes reality is just a bridge too far to make dreams come true. Leica M-mount enthusiast understood the limitations of rangefinder photography, and accepted that autofocus wasn't possible with M-mount lenses.
Still it would be so amazing if the peel of an orange were as edible and as crunchy as a red-ripe apple.
Eventually, technology made it possible to bridge the divide that previously made autofocus impossible with Leica M-mount lenses. With the rise of the mirrorless technology, dreaming the impossible dream no longer seemed unfeasible. It made Leica M-mount enthusiasts hopeful that Leica would finally grant us our wish.
Instead, Leica released the SL.
Admittedly, the initial announcement was disappointing. But over time, many M enthusiasts learned to embrace the Leica SL, despite the size and limitation of native lenses in the SL-system. Still, M-mount enthusiasts were never completely satisfied with the SL. It was after all too big. At best, the Leica SL complemented Leica M rangefinders, but it never replaced it.
In truth, accepting the Leica SL would be much easier if it weren't for the Sony A7. In a manner of speaking, the A7 is everything that the Leica SL could have been, in demonstrating how much of a lost opportunity the SL was. The A7 was so much more compact. Its autofocus worked better with native Sony FE lenses. And most importantly, it proved that autofocus was possible with M mount lenses - though with noticeable hiccups through a third party adapter.
*** WARNING *** This is a very long post. Just so you know.
The A7 wasn't without its faults, which is why many Leica M-mount enthusiasts never embraced it. But with the release of the Sony A9, it made me wonder. I've been gushing non-stop like a giddy school girl about how phenomenal the Sony A9's autofocus is from the moment I held one in my hands. Naturally as a Leica M and SL enthusiast, the A9 made me wonder how its improved autofocus would compare to the Leica SL, when adapting Leica M-mount lenses with the Techart E to M adapter.
To be clear, I am under no illusion to believe the autofocus of the Leica SL wouldn't outperform the Sony A9 adapting M-mount lenses. Native lenses will always perform better on a native system. But with how remarkable the autofocus performed on the A9, and with how unexceptional it was on the Leica SL, the obvious follow up question was by how much? Did the SL's autofocus perform noticeably better? Or was it just marginally better? Given sufficient uncertainty, I thought it made perfect sense to undergo this comparison.
Time to shove a circle into a square in making this orange crunch like a red-ripe apple.
I started the comparison with a series of simple tests comparing the SL and the A9 + Techart adapter, where Anna was seated stationary on the sofa. Actually, I really didn't have much of a choice, since she wasn't finished texting friends. Anyway, for the sake of properly evaluating the autofocus, I had set the aperture wide open, in order to rule out focus by depth of field.
Under the relatively controlled shooting scenario, both the autofocus of the Sony A9 + Techart adapter and the Leica SL performed within reason. Set to face detection, the Leica SL was more responsive than the Sony A9 + Techart adapter. But the Techart wasn't bad either. Compared to manual focusing M mount lenses, the autofocus experience of the A9 + Techart adapter was a better. Furthermore, the improved autofocus of the A9 seemed to have resolved many of the hiccups in focus hunting experienced previously with the A7.
But what if the available light wasn't ideal?
To evaluate the autofocus under poor or compromised lighting, I had set up a series of light conditions to compare with the lens wide open. On the first set, I compared the autofocus under strong backlight. On the second set, I compared the autofocus under reduced available light. On the third set, I compared a combination of strong backlight and reduced available light.
On the first set, both the SL and the A9 + Techart adapter experienced some minor focus hunting. I'm guessing that the reduced facial contrast in addition to the increased contrast between the highlight and shadow outlining Anna's silhouette gave the autofocus of both system pause to evaluate. But again, it was minor at best.
On the second set, both the SL and the A9 + Techart adapter did not appear to have experienced any noticeable issues. With that said, it was notable that the autofocus wasn't snappy. But it was reasonably quick enough to alleviate any suspicion that one was better than the other under reduced available light.
On the third set, both the SL and the A9 + Techart + adapter experienced focus hunting. Again, I am assuming the cause is a consequence of both reduced available light and strong backlighting. Though it should be noted that the autofocus experience didn't appear to be better or worse for either camera set up, in this situation.
Given what I observed, I am of the opinion that reduced contrast in the subject, as a result of compromised light conditions, didn't impact the responsiveness significantly of the autofocus. Having said that, it should be noted the focus hunting experienced with the SL and the A9 + Techart adapter was different. With the SL, the it was more an issue of determining the autofocus point. In contrast with the A9 + Techart adapter, it was more an issue of feathering the focus of the selected focus point.
Next test - on the street.
Up till now, I've been comparing the SL and the A9 + Techart adapter in a controlled indoor environment. Under such conditions, acquiring focus wasn't going to be difficult. After all, Anna was posing - or rather not moving. Testing indoors, there wasn't a crowd of faces competing for detection. And whatever difficulty there was in detecting contrast, the autofocus wasn't under any undue pressure to perform quickly.
But what if the testing environment wasn't controlled? What if I were to spill the comparison out onto the street, where the shooting situation wouldn't be controlled? Invariably, finding out how the autofocus performed in an uncontrolled environment was the next logical progression.
So, I asked Anna to blend into a crowd with other faces competing for detection. Moreover, I didn't want her to pose. I wanted her to walk with the crowd, to determine whether motion would impact the accuracy of face detection. And finally, I wanted her to move in and out of the shadow and sunlight, in order to determine whether changes in contrast and brightness impacted the accuracy of the autofocus.
The focusing distance was also a factor in testing. Because I needed to include other faces into the frame for testing, I had to shoot further away. And just to make sure that depth of field wasn't going to provide any margin for error, this test was conducted with the lenses shot wide open.
After conducting this comparison, it was evident that both the SL and the A9 + Techart adapter were too slow to acquire focus. In fact, they were both hopeless. Under such an uncontrolled situation, my hit rate was less than 10%. With movement, distance, competing distraction, and ever changing levels of light and contrast, both the SL and the A9 + Techart adapter failed to acquire focus consistently. With that said, there was however a difference in how they missed focus. Again with the SL, selecting the focus point was the main issue, whereas slow focus hunting prevented the A9 + Techart adapter from keeping up with the focus.
Maybe I should have asked Anna to take off her sunglasses?
Needless to say, the resulting images were horrible. The few I selected for this post have all been desaturated - given the color clashing nature of unplanned street photography. Furthermore, they've all been cropped for composition. Without control over the other walks of life, looking at each image was just an eyesore.
I think I would have tried to capture some better sample images for this comparison in an uncontrolled environment. Originally, I was going to do a second round. But then tragedy hit. I dropped my Leica SL 3ft onto the concrete sidewalk. The fall shattered my lens filter, and bent the filter thread on the 50mm f/1.4 Summilux-SL.
Needless to say, I decided to quit the outside portion of this comparison prematurely. For the moment, the street ceased to thrill me with its photographic possibilities. I returned back to the confines of my controlled space to regroup and recompose myself.
Plus Anna and I had to post the mishap on social media. We might as well, we figured.
After the fall, I needed a break from shooting more than one camera. Apparently, juggling between two cameras was just one too many for me to handle. Besides, I needed to do something different to get back into the groove of testing. Comparing cameras can become tedious after a while. But then, what to do next?
Up till now, I've only been testing apples to apples... or rather apples to oranges modified to be like an apple. And in doing so, I haven't tested the orange as an orange - albeit dressed like an apple. Unlike the Leica SL, the Sony A9 also comes with a tilting rear display which enables one to shoot from different angles. So then I wondered whether I could consistently get focus wide open if I shot the A9 + Techart adapter held above my head.
To put that to a test, I decided to break up the monotony by switching the lens on the Sony A9 to the 50mm f/0.95 Noctilux. I figured we might as well have some fun testing the autofocus with paper thin depth of field. Needless to say, it was. Under the confines of my controlled indoor environment, the A9 + Techart adapter focused without issue from a position hovering above my head - except for some minor focus hunting in feathering focus. With the benefit of the tilting rear display, I was able to visually confirm focus on live view. Easy peasy as the expression goes. The only issue worth mentioning was a lack of an electronic leveler on live view.
Just as I was getting back into my groove, Anna received a text. She had to meet up with her friends. With all the drama unfolding, I had lost track of the time. It was getting late. So then, we decided to wrap it up for the day. But since I was having so much fun after my mishap, I decided to walk Anna partway to her meet up point. Having said that, I switched the lens on the A9 back to its native FE 50mm f/1.4 ZA, given how poorly the Techart adapter performed under an uncontrolled environment.
The difference in performance was immediately noticeable, after lumbering around with the Techart adapter for most of the day. The A9 was just better with its own native lenses. The minimum focusing distance was closer, the tilting screen allowed for different shooting perspective, and the autofocus was snappy and always on point. It really was the proverbial no-brainer to shoot with Sony lenses on the A9. Consequently, it really made me wonder why anyone would adapt non-native Sony lenses on the A9. Oh yes... it's because we're Leica enthusiasts searching for the holy grail.
Given unfinished business with what we had set out to accomplish on this comparison, we decided to add a second day of comparison. This time we were going to compare the autofocus of the SL to the A9 + Techart adapter in a controlled environment graced with more than one face.
Clearly, shooting on the street didn't work out for me the day before. So I asked Anna to conscript a friend into our service for the day. Long story short, her friend who shall remain nameless cancelled at the last minute. My day was certainly starting off with a bang - shattered like my 50 Summilux-SL. But as the old saying goes, when a door closes, a window opens. Or rather in this case, it was Anna who was opening her iPhone contact list, to find a replacement.
As it turned out, in the haste of finding someone new to fill in for us, Anna managed to get two kind souls instead of the one to answer our call. With both friends seemingly willing to give us a hand - much to my surprise - who was I to play favorites. Besides, I felt somewhat guilty, seeing how Anna urged them incessantly to wake up so prematurely, this early in the morning. Besides, three was definitely better than two in making a crowd for the purpose of this comparison.
We would continue from where we left off from the day before. After all, we didn't quite finish the face detection comparison, given the crashing fall of my tragedy. The uncontrolled environment on the street was much too demanding for any reasonable autofocus comparison between the SL and the A9 + Techart adapter - much less achievable for presentable sample images (in light of both setup's limitations). But with the addition of Marina and Masha in today's shoot, we would be able to make that comparison without putting anymore gear in harms way.
I should have thought of this sooner.
The methodology for this face detection comparison was simple. Test the accuracy of face detection in autofocus, and determine whether face registration offered an advantage. Anna's face was registered on the Sony A9. So if I could consistently get her in focus, it would demonstrate that face registration works.
The gear used on this final set of comparison were (1) the Sony A9 + FE 50mm f/1.4 ZA, (2) the Sony A9 + Techart adapter + Leica 50mm f/0.95 Noctilux-M ASPH, (3) the Leica SL + 50mm f/1.4 Summilux-SL ASPH, and (4) the Leica M10 + 50mm f/0.95 Noctilux-M ASPH. I included the M10 in this test, since a manual focus sample with Anna always in focus was necessary for a baseline comparison.
It should be noted that I forgot to bring my 50mm f/1.4 Summilux-M ASPH. That was a goof up. On a brighter note, the 50mm Summilux-SL soldiered along despite licking its battle wounds.
To test the accuracy of face detection in a group, I conducted four sets of comparison. On the first set, I shot closer up with Anna in the middle and everyone seated on the same focal plane (at f/1.4). On the second set, I went a little further back and staggered Marina and Masha outside the focal plane (at f/1.4). On the third set, I had Anna seated furthest away from center with all three spread out at varying distances (at f/5.6.) On the fourth set, I had Anna seated furthest away with all three seated closer together at varying distances (at f/2.8).
It should be noted that I stopped down on the third and fourth set because I still wanted the image to look presentable, with all three appearing to be either in focus or close to it.
As expected on the first set, the autofocus selected Anna on autofocus face detection. On the second set, Anna was selected again. On the third test, Marina being the closest was actually selected all three times. Although there were instances when the A9's face detection locked onto Anna - with or without native lenses.
Given how consistent the A9 was with either a native or adapted lens, I didn't even bother with the adapted lens on the fourth test (which I regret in retrospect, given the point of this comparison). Having said that, this fourth set of comparison had the most interesting results. With all three seated closer together along the focal plane, the Sony consistently focused on Anna, while the Leica SL consistently focused on Marina who was closest. In my opinion, this demonstrates that face registration does work.
It is also interesting to note when Anna and Marina's head were turned away from the camera, the autofocus selected Masha. This was the case on the final image shown below with the Leica SL and the A9 (not shown here). In my opinion, this demonstrates the pecking order that face detection undergoes in finding its focusing point. With Anna's registered face turned away, and Marina's face closest, also turned away, the face detection algorithm locked onto Masha's as the best option left.
It has been an exhaustive two weeks of testing, evaluation, and writing with the added discomfort of jet lag. In that time, I believe I've gotten to know the Sony A9 quite well, with respect to how it compares to the Leica SL, from the perspective of a Leica M rangefinder photographer.
The Sony A9 doesn't replace the Leica M10. However, it almost replaces the Leica SL as my autofocus complement to the M10. For most of what Leica M photographers do, the Sony A9 isn't necessary. Admittedly, autofocus is fast. However, zone focusing with a rangefinder is faster still - even wide open (if you know what you're doing).
But when you start to adapt M-mount lenses on the Sony, the autofocus is no longer as responsive. From a Sony perspective, you kind of ruin the autofocus experience with the Techart adapter. Still, the autofocus experience isn't significantly worse than that of the SL. The SL with the 50mm Summilux-SL is only marginally better. Having said that, comparable lack-luster performance in adapted autofocus still isn't enough for me to switch over to the A9 from the SL.
So why doesn't the A9 replace the SL? It's actually a very silly reason. I prefer the focal range and maximum aperture of the Vario 90-280mm f/2.8-4 compared to the Sony 100-400 f/4-5.6. The only time I really need autofocus is when I need to zoom closer from afar. Having said that, the A9 has the advantage of being smaller and lighter - even with the massive 100-400mm lens. I suspect I might get one and see how it feels to travel with one. But without weather sealing on the A9, the SL edges out... which is a shame, because I really like the A9.
Butterfinger needs all the protection he can get, since I know I'll be shooting in the worst possible weather. It's just me.
By the way, both Masha and Marina have been such amazing troopers, in being the out of focus part of our Sony A9's face detection test. So to make up for it, I've taken a couple of parting shots with the Leica 50mm f/0.95 Noctilux-M for them... not that they need them... but why not... we had some time left over to spare.
After today, I don't think I will be reviewing the Sony A9 anymore in this way. I'm exhausted. Besides, it would be an overkill to do another one - with this being the fourth one. Still, I believe the entire process was worthwhile. As a Leica M photographer, the Sony A9 shows us what is possible. At the very least, we can only hope that the powers-to-be in Wetzlar are taking note. Mirrorless is the future, and with the technological gap now bridged, autofocus is no longer an impossible dream for M-mount photography. If I learned anything from this comparison, it's how time saving and convenient autofocus is.
I don't expect Leica to make the M an autofocus camera. But at the very least, they should come up with an autofocus adapter for the Leica SL. If Leica does that, it would forever end the discussion of whether one should get a Sony as a Leica M complement.
Images have been edited at varying degrees in Lightroom. Significant edits and cropping have been disclosed in the captions. All images photographed wide open, unless mentioned. Photos were selected on the basis of presentability.
Again, special thanks to Masha and Marina!