Testing Auto Mode on the Leica SL's Autofocus - with the Vario SL 24-90mm
This was suppose to be a Leica SL + 24-90mm Vario vs Novoflex adapted Canon 24-70mm autofocus accuracy comparison. But the day before the test, the autofocus on my Novoflex SL/EOS adapter stopped working. Needless to say, I wasn't happy - which wasn't exactly a good thing - since I had arrange to conduct my comparison in none other than the happiest place on earth - albeit on this side of the bigger pond. Yes, I'm talking about Disneyland, Hong Kong.
Admittedly, I could've canceled the shoot, given the change in circumstances. I mean, my Novoflex adapter is no longer an autofocus adapter. It's just a hunk of paperweight. But the thing is, I had arranged this outing already with Anna, and her friend Lessy, to accompany me - rain or shine - to Disneyland. And if I wasn't going to let a little ol' thing like rain get in the way of my shoot, why should unreliable technology stop me from pushing forward. Besides, I think Anna really wanted to go!
Only thing now is what to do in place of my original plan. If initially my intent was to test the autofocus accuracy of the Canon 24-70mm f/2.8L on the Leica SL with the Novoflex adapter, perhaps the logical fall back plan was to do something along the same objective as evaluating autofocus. Then it occurred to me that I really haven't fully tested the autofocus on the Leica SL, since my conversion from the M240.
Warm up shot. I needed a couple since shooting in auto mode was outside my comfort zone.
More practice shots. You know the only reason why I'm including this image is because Anna's friend Lessy is making that goofy face. On a side note: this image has gone through quite a bit of post, given the extreme contrast in highlight and shadow caused by the direct noonday sun. I have to say that the dynamic range on the SL is pretty good.
I think that this is part of the Toy Story World? It's really just a reference shot, given that the first part of our visit didn't really feel all that much like the Magic Kingdom?
Up till now, I've been shooting almost exclusively Leica M-mount lenses on the SL. I mean, how could I not? They're a match made in heaven. For the first time ever, I can focus and frame wide open with my Noctilux with relative accuracy. What's there not to like? But then it occurred to me. Maybe that wasn't the reason. Maybe, just maybe, shooting autofocus was outside my comfort zone. As crazy as that sounds, that might be the case.
In fact, that might even be the case for most Leica SL photographers. It makes sense, when you think about it. Most of us are converts from Leica rangefinder photography. And because of that, we may not necessarily be comfortable surrendering that measure of control to the camera.
Admittedly, I have used the autofocus on the Leica SL. However, I've limited myself to only static, servo, single point autofocus. In other words, I'm still selecting the focus point. But what if I were to let the camera decide for me? How well would it perform? Well, that made me curious. So then I decided that I was going to do the entire shoot in autofocus auto mode.
I needed some refuge from the intense noonday sun. I have no idea which world I had descended onto.
I was beginning to get used to the face detection of the autofocus.
At this range, face detection seemed to work very well.
Obviously, the lens that I used for this test was the Leica Vario Elmarit SL 24-90mm f/2.8-4 ASPH. I didn't really have much of a choice, given that Leica only has two autofocus SL lens at this time, and the other one is the 90-280mm lens, which is much too long in focal length (which quite frankly would not work for this review).
Well, I was satisfied with the change in objective. I mean, when you think about it, setting the Leica SL to auto mode makes sense in a place like Disneyland. Effectively, what I'm doing is turning this Leica SL into a giant point and shoot. And as we all know, the point and shoot is the camera of choice for the amusement park set.
So in keeping with the point and shoot comparison, I was limiting my Leica SL setup to f/5.6, 1/500s, and AUTO ISO, unless the lighting changed significantly. For the most part, the only variable I was allowed to change was the focal length on the zoom ring.
With Anna and Lessy further away, it became difficult for the Leica SL to make out enough contrast for face detection.
Another example where auto mode found difficulty in selecting focus. I don't recall where the autofocus finally acquired focus. But clearly it wasn't Lessy's face.
Another difficult autofocus situation. Under the shadow of the thick canopy, and at that distance, the autofocus couldn't make out enough contrast to select focus on either Anna or Lessy's face.
With the background and methodology of the day's shoot out of the way, the next part to discuss is the shooting experience in auto mode. To be perfectly frank, I felt like a mermaid out of water... part of that world. I'm really not used to shooting autofocus on a Leica, let alone auto mode. Even when shooting Canon, I'm selecting my own focus points. So to let the camera select focus took some getting used to.
The way auto mode works is like this. The camera automatically selects the closest subject with contrast. And if the camera detects a face, the face detection automatically locks on. And if you're on continuous focusing, like I was, the focus will continue locking on.
It sounds great on paper. But how would it play out in the real world - especially in a place like Disneyland, where there are many faces potentially confusing the camera's face detection, and where your subject is likely not to pay attention to you when you're taking photos of them. This is after all Disneyland. It's only natural that your subject's gazes won't be looking back at you, given distractions like a Cantonese speaking Mickey meeting and greeting - and no, I didn't take any pictures of Mickey.
The spray of water mist made focus in auto mode more difficult, since it again reduced the contrast of Anna and Lessy's face.
On tribal drums. This image has gone through considerable post, in editing. In the original capture, Anna's face was in shadow, while Lessy's face was in extreme highlight. As a result, there wasn't much contrast in either Anna or Lessy's face, making focus selection difficult in auto mode.
Another example where Lessy and Anna were far away from the camera, with the contrast in both their faces diminished as a consequence of distance.
The autofocus completely missed focus on either Anna or Lessy's face in this image. The autofocus selected focus on the contrasty leaves of the closer hedges. I had to manipulate the camera's position to focus on the leaves closer to Anna and Lessy, in order to reasonably get focus.
So how did it turn out?
Well, for starters, the autofocus had problem selecting focus when the subject was further than ten feet away, with the 24-90mm Vario lens. Whenever I tried to photograph Anna and Lessy with their entire body, from head to toe, in the frame, the Leica SL failed to select focus on them. That was infuriating, because it made capturing an environmental portrait of them excessively difficult - especially if I wanted more background scenery.
I can only rationalize this by assuming if the subject isn't close enough, then the amount of contrast on the subject's face diminishes beyond what the autofocus in auto mode could select. If that is the case, it would not be possible to get focus on Anna and Lessy's face in a head to toe body capture, while in auto mode.
Fact is, both Anna and Lessy are too tall at roughly six feet in height. If they were only four inches shorter, perhaps they would be close enough for the camera to detect sufficient contrast in their faces, for face detection to work. Or alternatively, I could just move in close enough for the autofocus to make out sufficient contrast. But that would crop out a portion of both their legs.
Lessy and Anna bolting off to make it to the Teapot ride. The autofocus didn't have a chance to acquire focus.
There wasn't enough contrast on either Lessy or Anna's face, especially with the teacup rotating and snaking round the perimeter of the ride. Instead, the autofocus locked onto the graphics of the teacup.
Even closer up, and yet the autofocus again locked onto the graphics of the teacup.
As for selecting focus on moving subjects, I found auto mode to be completely clueless. This I determined in my Teacup ride test. In all due fairness, it could be because I was too far from Anna and Lessy to get enough contrast on their faces for face detection. But then again, what use is auto mode if you cannot select a moving subject from a distance? If the Leica SL were a point and shoot designed to capture close up candids and group shots, I could understand limiting autofocus point selection to within ten feet. But with this being a professional level interchangeable lens camera, the autofocus should really be able to detect diminished contrast of subjects further away.
In my test, there was also the added difficulty of competing faces entering and leaving the frame. It virtually rendered the autofocus useless in auto mode. With so many faces, it would undoubtedly confuse the face detection. But I suppose that this test would have been difficult for any camera. Even if I were to use the Canon 5D Mark IV, and use its equivalent of auto mode, and in live view, I would still have problem locking focus, given that the subject is continually rotating and snaking out of view, throughout the duration of the ride.
However, I still found it difficult to lock onto focus even when the moving subject got close enough for face detection. It almost felt as if the camera needed a split second to confirm focus before it could lock on. And by that time, after that split second had expired, the moving subject would've moved on without focus ever locked onto it.
In the Its A Small World ride. Surprisingly, I was able to get reasonable focus on Anna's eyes. But more surprising is the dynamic range of the image file. The white balance wasn't this natural in the original capture. Also note that this was taken at 24mm, f/2.8, 1/60s, ISO 6400, and pushed two stops in post.
Still on the Its A Small World ride. Here, there was sufficient light for a good image capture. Unfortunately, Lessy didn't move up closer to be in focus (or alternatively, Anna didn't lean back). Either way, autofocus selected the closer subject, being Anna in this case.
The start of the Space Mountain ride - and no, I didn't capture anything else afterwards. Are you kidding me. It's a rollercoaster ride - albeit low grade - but still moving very fast. White balance is irrecoverable in this case, given that it was monochromatic - though in the magenta scale. Still, I was surprised that focus was acquired in auto mode. However, Anna and Lessy wasn't moving, and I was pretty close up to them.
Strangely, the autofocus in auto mode wasn't bad in selecting focus, when the subject was close up and under low light. This I was able to demonstrate during the Its A Small World ride. This is not to say that my hit rate was 100%. Far from it. But it was significantly better than trying to acquire focus on moving subjects, or of subjects crowded in an area with many competing faces confusing face detection.
Even at the start of Magic Mountain, the autofocus wasn't terribly bad in auto mode. I was still able to get focus at very close range, with limited contrast in very low light - albeit in tones of dark magenta. However, this is not to say that autofocus selection was quick. There was some hunting in the dark, and it did take some effort in manipulating the camera's position to persuade autofocus to select onto either Anna or Lessy's face. Still, the white balance was horrible, which is why I had to desaturate the image to black and white.
Obviously, this is anecdotal. But based on my findings, I have this sneaking suspicion that the autofocus of the Leica SL is designed more towards still life or portrait photography than action photography.
Bobbing up and down, within a reasonable distance, the autofocus was able to acquire focus in auto mode.
Anna taking a photo of Lessy.
After the Merry-Go-Round. This image has gone through considerable edit in post. With extreme backlighting, and both Lessy and Anna originally in shadow, the autofocus was still able to make out enough contrast on their faces to acquire focus in auto mode.
On my second attempt to acquire focus on moving subjects in auto mode - namely at the Princess's Merry Go Round - my hit rate found greater success. In that scenario, both Anna and Lessy were bobbing up and down. As for me, I was on the horse in front, shooting in a most awkward twisting back position, with a camera pointed at them, while also bobbing up and down.
In this test, I was able to lock on focus continuously. In retrospect, I am of the opinion that getting focus in this scenario was possible, because there weren't any competing faces to distract the camera's face detection, and because I was shooting close enough at a fixed distance for the autofocus to make out enough contrast. And at 24mm, I guess that the depth of field was generous enough to cover up any deviation in focus.
My only faux pas was not setting the shutter speed faster. But how was I to know that the bobbing motion was going to be that quick! In any event, this finding made me confident that autofocus in auto mode could reasonably select focus of moving subjects under more controlled conditions.
Another example where Anna and Lessy were originally in shadow. Again, autofocus was able to acquire focus in auto mode. Image edited in post to reduce the intensity of the shadow.
A quick turn around shot at the Marvel gift shop. The autofocus was able to select focus in auto mode. I guess I was close enough for the face detection to get sufficient contrast.
Finally, a bite to eat. Obviously no problem getting focus, at close distance, with sufficient lighting, and at the 24mm focal length.
Leaving the happiest place on earth this side of the bigger pond, by Über, with my body twisted back to face Anna and Lessy for one final snap. The autofocus had no problem selecting focus. But given the twist and turn out of the amusement park, it was difficult for my hands to shoot steady.
Overall, what did I think of the Leica SL in this review? Would I do this again? Would I shoot in auto mode? Would I opt for the Leica SL with the 24-90mm on any hypothetical follow up visit to the happiest place on earth? Honestly, I think I would have been better off with Anna's Fujifilm X-Pro2. It's smaller and therefore more flexible in making image captures at different angles of perspective. By comparison, the Leica SL with that massive zoom lens isn't exactly the most maneuverable camera to shoot from varying positions.
In the final analysis, I am still more comfortable manual focusing M-mount Leica lenses on the Leica SL. But with that said, I cannot say that I would be adverse to selecting auto mode, in the future. Now that I am more aware of it's limitations, I am mindful of what I can do and what I cannot do. I won't do any action photography with it. But for everything else, I believe the autofocus in auto mode will perform sufficiently well.
All images in this writeup have been optimize in Lightroom. I think one or two of the images have been cropped slightly for better framing.
Special thanks to Anna and also to Lessy for being such a wonderful tour guide! It wasn't her first time there. Apparently, she really likes Disneyland.
