http://petapixel.com/2016/02/11/techart-unveils-worlds-first-autofocus-adapter-manual-focus-lenses/
Very interesting. I'm thinking about my Zeiss 50/2 having AF.
Very interesting. I'm thinking about my Zeiss 50/2 having AF.
AF adapter for M lenses
- Thread starter jowei
- Start date
M. Gilden
Canon DSLR Moderator
I saw this earlier- I was wondering when/if we'd ever see something like this. I always wondered about a camera body that could shift the sensor position (much like IBIS) to something similar for focus of manual lenses, since I would think that is easier to engineer than a motorized lens mount. But the demonstration video is really promising! I would not have imagined it works this quickly, let alone supporting continuous autofocus!
But I have so many technical questions.
For example:
How does the camera/adapter combo know what position to jump to for focus? Each lens would have a slightly different range, would the adapter need to be calibrated for each one?
Is that what the bluetooth connection to a cell phone is for? Do you need to do that everytime a lens is attached (since it would have no other way to determine what is attached)?
What if the manual focus gear gets touched, wouldn't that throw the whole thing off and need to be recalibrated?
What about focus breathing? Would this combined with a prime that "breathes" be able to compensate for it and make it a sort of mini zoom?
And speaking of zooms, would they be incompatible because as a lens shifts its zoom, the focus range and positions shift as well (especially if a lens is not parfocal)- and the adapter would have no way to tell if that happened?
What about moving the lens too far away from the camera body that the image circle starts to vignette? Remember, the focus gear moves certain elements back and forth, but the rest of the lens is designed to make sure image circle still covers the frame it was designed to. While physically moving the lens can achieve the same shift in focus, would it undermine the lens' ability to keep full frame coverage?
My mind is literally racing with questions, possible uses and technical hurdles. Whatever the case, this is VERY cool tech!
Bern Caughey
Veteran
Demo of the Contax AX's moving film plane. Since it was an SLR the entire mirror box needed to move too.
http://youtu.be/MjgV34PzKTA
"CONTAX AX – This had a unique autofocus system that worked with manual focus lenses by moving the film plane inside the camera. A side benefit of this arrangement is that it allowed the AX to feature a macro mode which worked much like a built-in 10 mm extension tube, allowing for a magnification ratio higher than 1:1 without the use of bellows or extension tubes. By using of a special adaptor made by Kyocera, Hasselblad V-series lenses could also be used on all CONTAX C/Y mount cameras. Thus users could perform auto-focus of Hasselblad V-series lenses with AX."
https://en.m.wikipedia.org/wiki/Contax
M. Gilden
Canon DSLR Moderator
Ah, yes. Someone pointed that out on another forum. I had heard about it once but totally forgot about it. One second, if the entire mirror box moved along with the lens, wouldn't the focus not change in the viewfinder? I was under the impression that only the front of it moved.
Either way, what I'm trying to understand now is how the Contax determined focus in the AX. I would assume that it used phase detection or some other sort of similar distance calculator, as cameras from the 90s usually did, but then how would it know precisely how far to move the lens to focus at a particular distance? As I understand it, phase detection and other similar technologies would calculate the distance of the subject, and then tell the lens or focus motor to snap into a particular position ASAP, which it often does in one swift movement. Wouldn't lenses of different optical design and focal length have a different distance throw? In other words, some lenses would hit 2.5m focus at one position, but another perhaps would require sitting closer or further, so how did the Contax AX know exactly how far to hold the lens to focus at a particular distance?
If it was using contrast detection to analyze if the shot is in focus yet, wouldn't it have been much slower and frustrating to use? I mean, E-mount lenses can operate in contrast detection on our Sony cameras because they were designed to do the "contrast detection dance" at high speed, where it sort of hunts back and forth doing trial-and-error adjustments. But lenses not designed for that, as we've seen by adapting Canon lenses to a Sony camera with an electronic adapter, hunt around slowly (unless you have the optional phase detection support in the camera like the A7R2). Surely a load-bearing motor that moves an entire lens or lens+mirror box would be even more so, incapable of the high speed dance required for contrast detect systems?
Yet, there it was, working rather well according to reviews of the Contax. And there's that video of this M-mount adapter, focusing in what MUST be phase detection, because there's no way a third party adapter can push an entire lens around hunting back-and-forth to focus as quickly as they show it in contrast detection. In fact, don't they say that this only works with cameras like the A7R2 (which offers PDAF), and almost purposely leaves out the A7S or original A7?
It must be phase detecting and snapping the motor into a particular position, but my question is HOW does it know the correct position?!
Are there any engineers in the house???
Samuel H
Major Contributor
Not an engineer, but ready to guess stuff I don't know about:
My first thought is that this might need strong AF microadjustment because this manufacturer is unlikely to work with very small tolerances. But on mirrorless the PDAF sensors are on the sensor (yes, I know) so no, there's no AFMA available, and you won't need it if the thing can jump to the correct position.
The thing is: will it jump there? Or will it move a bit too far, or fall a bit short? If it did so, the PDAF sensors would tell it that it's out of focus and it has to move slightly one way or the other.
And yet different lenses will need different amounts of motion: my 8mm fisheye goes from 30cm to infinity in just a couple of mm, whereas my 50mm needs five or ten times as much movement.
But this may be easy to solve: you can program the adapter so that it has a "motion scale" parameter that is recalibrated if on the last AF actuation it fell short or went too long. So, only the first time you focus after changing lenses you'll see the thing hunting a bit, then everything works fine at the first try.
That's how I would approach this problem, no idea what they did or if it would work.
Also: my 300mm needs more than 5cm of displacement for focus. Using "shift-sensor AF" will limit the focal lengths you can use.
My first thought is that this might need strong AF microadjustment because this manufacturer is unlikely to work with very small tolerances. But on mirrorless the PDAF sensors are on the sensor (yes, I know) so no, there's no AFMA available, and you won't need it if the thing can jump to the correct position.
The thing is: will it jump there? Or will it move a bit too far, or fall a bit short? If it did so, the PDAF sensors would tell it that it's out of focus and it has to move slightly one way or the other.
And yet different lenses will need different amounts of motion: my 8mm fisheye goes from 30cm to infinity in just a couple of mm, whereas my 50mm needs five or ten times as much movement.
But this may be easy to solve: you can program the adapter so that it has a "motion scale" parameter that is recalibrated if on the last AF actuation it fell short or went too long. So, only the first time you focus after changing lenses you'll see the thing hunting a bit, then everything works fine at the first try.
That's how I would approach this problem, no idea what they did or if it would work.
Also: my 300mm needs more than 5cm of displacement for focus. Using "shift-sensor AF" will limit the focal lengths you can use.