Lens aberration impacts TechArt LM-EA3’s auto focus accuracy, part 3

 

Sony Phase Detection Image Sensor

The Sony mirrorless camera use two kinds AF methods, ‘contrast AF’ and ‘phase detection AF’. Earlier model like A7 and A7R use only the contrast AF and later ones A7 II, A7R II and A9 use the phase detection AF in addition to the contrast AF, which is user selectable. The TechArt LM-EA3 adapter uses the only phase detection AF thus does not work on the A7 or A7R.

 

 

 A phase detection image sensor is made a part of pixel’s height different from others, so that the path of light (=phase) differs in between normal pixels and phase detection pixels.

 

 

 

 

When ideal lens (no spherical aberration) is used

(Left) Normal pixels are in focus     (right) Phase detection pixels are out of focus: These two cases are distinctly different

As shown in the above figure, light coming to normal pixel and phase detection pixel are very different due to height difference of them if ideal lens used. AF algorithm calculate correct focal point based on these signal difference, and drives AF motor to bring the lens to the right position. This is the principle of phase difference AF.

When an old lens with spherical aberration is used

(Left) Not all of normal pixels are in focus   (right) Phase detection pixels are also in focus: These two cases are mixed in signal processing and confuse AF algorithm

Lens with (a lot of) spherical aberration has no signal difference in between normal pixels and phase difference pixels as illustrated above, and AF signal is noisy to prevent accurate focal plane calculation. As a result, AF algorithm place the lens in wrong position. This is exactly what happen to the Sigma XQ 135 / 1.8 lens. When use as f1.8 (hell lots of spherical aberration), it focus to front side consistently, meanwhile stop down to f2.8 minimize spherical aberration & become pin-point accurate focusing.

 

Front focus symptom due to Phase Detection sensor confused with spherical aberration, when focus on ‘Rollei’ letter using TechArt Adapter, the actual focus goes around the purple-colored book ‘classic cameras’.

In fact, I was able to realize this problem due to Sigma XQ 135 lens’ behavior(?). It is a telephoto lens with very shallow depth of field which makes any small focus error visible with the naked eye & being as 1960’s legacy, comprise a lot of lens aberration at wide open.

The other lenses I’ve used, the Canon FD 55mm f1.2 and Nikon 55mm f1.2, also have this problem but they do not show the symptom clearly like Sigma 135mm lens and not consistent enough to suspect the lens aberration as a root cause.

At wide open f1.2, both lenses looks ‘not so good image quality’ instead of focus shift at a glance. In reality, the adapter gives very slight front-focus consistently at f1.2 aperture thus image looks blur but I thought it might be the lens performance at f1.2 anyway or the subject or camera moved a little after focusing or simply I made mistake when I put focus point to the subject.

Nope, not at all. Those 55mm lenses perform well even with wide open, at f1.2, as long as manually focused through magnified screen. I have been using TechArt LM-EA3 adapters for the last 1.5 years, but I’ve only use moderate angles like 35, 50, and 85 mm lenses which is not easily noticeable this issue, then couple of weeks ago I try 135mm f1.8 & banged up.

Sigma XQ 135 / 1.8 at f2.8. To avoid front-focus problem I set aperture to f2.8 throughout photo session & found satisfactory result at AF accuracy

When I tested the Sigma lens for the first time, I suspect the lens is too heavy for the adapter so focusing helicoid was not operating properly, but as soon as the aperture stop down to f2.8, it started to focus precisely and firmly every time press the shutter. When open the iris back to f1.8, make the same front-focus again.

With this kind of consistent symptom, it can’t be the adapter, nor a camera, not even photographer’s mistake as a reason. It is the lens itself. Only difference is aperture opening f1.8 vs f2.8 and associated spherical aberration which becomes too much at wide open for AF algorithm.

 

 

(Continue and conclude at Part 4)

 

 

 

글쓴이 Leo_KHIMME

사진, 사진기, 렌즈, 여행, 새로운 곳, 새로운 기술 & 신산업혁명
대만 타이페이 거주
Oldies but Goodies 오피넛 멤버