1Point AF?

[dgktkr]

Advertisement (gone after registration)

I've been struggling to get a working understanding how 1Point AF functions in the SL. For starters, 1Point implies to me, one pixel. But that can't be, because the whole system depends on contrast in differences neighboring pixels, doesn't it? OK, so maybe it means a small region. To use this mode properly, some of us wonder how big that region is. To try to answer that question, I did some experiments.

 

I set up two focussing targets (small Siemens stars) in my kitchen. The lens is the 24-90 zoom. OIS is off. The camera is on a tripod. Manual focus mode is used, and AFs is triggered with the joystick. The far focus target is 2.46m away from the sensor plane. The near target is 2.18m away.

 

The first photo is of the LCD on the SL after AFs has been triggered. The AF action registered as successful with a beep from the camera and a green cross.

 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

The white cross is over the darker wood of a kitchen cabinet. The out of focus Siemens star is in the plane of the kitchen cabinet, which is 28cm further away from the camera than the target that is in focus. The lighter wood is an edge-on cutting board.

 

As you can see, the camera signaled success focussing on the kitchen cabinet, even though it actually focussed 28cm closer to the camera! One might be tempted to say that the region that the AF is using is a region that is maybe 5 times the size of the cross.

 

If I move the cross further from the image of the cutting board, the camera signals AF failure by not beeping and turning the cross red.

 

To determine if the camera behaved differently in the portrait orientation, I turned it 90 degrees and repeated the experiment. 

 

 

The results are somewhat similar to the first experiment. The AF system wants to focus on the edge of the cutting board, which is closer to the camera. AF success is indicated, even though the object most in focus is not indicated by the cross.

 

One difference is that neither target is in good focus.

 

dgktkr

 

[LocalHero1953]

I agree that so-called 1-point autofocus isn't: I suspect it just means it is a small area centred on one of the SL's focus points. But I don't think there's a simple answer to how big the area of focus is. I think it depends on the degree of contrast visible. And something which is within the area of focus but initially way out of focus and blurred will not be picked up.

 

If you are in 1-point AFs mode and switch to tracking, you get a box, not a cross. I suspect that is more like the actual area used for 1-point autofocus. But it would be more helpful if Leica's manual explained how the technology is implemented instead of leaving us to guess.

[steppenw0lf]

"One point" is to be read as opposite to "multi-point". Of course it is not related to a single "pixel".

You would not be able to see such a tiny area on the EVF, and results would be completely random if you chose a target not completely uniform, because the slightest movement would select a different point on the target. And if you selected a different resolution, would that then change the size of the AF pixel ?

 

The one thing to learn is: Select your type of AF carefully and check if it is suitable for your photographic object. Use manual override to make a second or third photo - if you want to be on the safe side.

 

I do not need to know the exact size of the AF area or point, in the same way that I never needed to know the exact size of the area used for spot metering in the older Leicas. (I think it was 1% of the viewfinder, but I am not sure, because it never did matter). 

If I would need the exact size to get decent results, this means that the method chosen is inappropriate for the target.

 

P.S.

I wonder, is there anything like automatic AF bracketing ? (Slight variations in distance, and the amount customizable. Probably not, because this would mean to admit failure to focus perfectly. But I think manual override is easier anyway).

[ramarren]

...

I wonder, is there anything like automatic AF bracketing ? (Slight variations in distance, and the amount customizable. Probably not, because this would mean to admit failure to focus perfectly. But I think manual override is easier anyway).

 

 

Not so much focus "bracketing" but focus "stacking" ... making a series of exposures focused slightly off from each other, then combining them in post to achieve greater depth of focus than can be achieved with a single exposure and DoF. Olympus now offers that in-camera with the latest firmware on the E-M1 model and certain lenses. 

 

All this discussion of auto focus makes my use of manual focus most of the time seem so much easier and more consistent... !

[pop]

.... results would be completely random if you chose a target not completely uniform ....

This might be close to the truth, but not in the way you might expect. All AF has to rely on there being an edge within the "AF point"; hence, the term "point" is rather misleading. All AF system I have used so far seem to favour vertical edges when holding the camera in landscape orientation.

 

Focussing on a uniform area would not succeed because there would be no way of telling whether the uniform area was sharp or not. 

 

However, the indicator in the F which marks the spot might be in a wrong or at least in a misleading place.

[LocalHero1953]

Sure, we don't "need" to know how AF and spot focusing works. But as someone who dismantled my first clock when I was 8, i want to know how things work. It helps me do things better. The handbook for my car, a current VW Golf, which has auto everything, including adaptive cruise control and emergency braking, explains in some detail how all these things work and their limitations, qualified in every case (cf Scotty from Star Trek) by the comment "but this system cannot defeat the laws of physics".

Undoubtedly YMMV.

[dgktkr]

Advertisement (gone after registration)

...

 

The one thing to learn is: Select your type of AF carefully and check if it is suitable for your photographic object. Use manual override to make a second or third photo - if you want to be on the safe side.

 

...

 

 

Isn't it difficult to select carefully if one doesn't have a good working knowledge of the operational characteristics? For me, having a rough idea of the size of the AF sensitive area would help me choose where to position that area in a scene for best results.

 

I wonder if threads about the 90-280 zoom focussing on the trees in the foreground instead of distant buildings might be related to this one. And perhaps Diglloyd's excitement about "focus shift" with the 24-90 zoom.

 

dgktkr

[ramarren]

Experiment, test, understand how the system works in operation with all modes and options.

Then you can make your choices over what mode and options to use in what situation. 

 

I've never seen a manual for any complex device with a large number of modes and options sufficiently cover what I needed to know to understand the systems' operation for my use completely. Testing and experimentation targeting my typical uses have always been essential. 

[edwardkaraa]

I'm not using AF on the SL yet, but doesn't the cross mean tracking AF?

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[steppenw0lf]

Isn't it difficult to select carefully if one doesn't have a good working knowledge of the operational characteristics? For me, having a rough idea of the size of the AF sensitive area would help me choose where to position that area in a scene for best results.

 

 

If you do not know anything about the system characteristics, you have to try all possibilities. Then you will see which gave the desired result, if any.

Next time you will know more and after a while of experimenting you will get better at choosing the right method.

Remember ? That's how you started into life...          

[steppenw0lf]

Sure, we don't "need" to know how AF and spot focusing works. But as someone who dismantled my first clock when I was 8, i want to know how things work. It helps me do things better. 

 

If this is your trusted method, then you should maybe start with dismantling the camera, or the AF system to be more precise ...  maybe also the lens ...    

 

Actually you left out the interesting part of the story ("I also like repairing things"  )   -  did the clock ever work again, afterwards ?

Typically that is the proof of a deeper understanding. 

 

Many others I know would automatically experiment with the camera, and would probably never touch a manual that gave them the dimensions of the AF area. (The hint to look into the manual would be considered a "mortal offence" and a reason good enough to cancel a life-long friendship.)

[LocalHero1953]

Actually you left out the interesting part of the story ("I also like repairing things"  )   -  did the clock ever work again, afterwards ?

Typically that is the proof of a deeper understanding. 

No. At the age of 8 I lacked the understanding, skill or wish to put it back to get again. And most of the cogs flew into the corners of my bedroom, and were never seen again.

But since then l have learned not just to take clocks and cameras apart, but to put then together again. Often in working order.

[steppenw0lf]

I agree that in this case I would also like to know the exact dimensions of the AF field. (Because the SL is special for me, and it is highly precise and .... blablabla)

But if this was a "asian" camera, would I request that ? Probably no, because I would not expect to get an answer.

I heard from the grapevine (or simply it is a rumour) that the AF system of the SL is from Panasonic. So do I expect a clear answer from them ? Not really. And I am not even sure, that Leica has fully detailed specifications.

But this (not knowing detailed AF specifications) is also the case for my Canon and Nikon and Sony equipment. And it has not kept me from using these cameras.

The question never even occured to me - so Leica has to regard this maybe as a compliment. The EVFs or OVFs are rather modest, so I was busy selecting the right clipping, and had no time for additional questions.

[pop]

I think that a bit of experimentation might go a long way towards finding the answer to that question.

[ramarren]

I'm not using AF on the SL yet, but doesn't the cross mean tracking AF?

 

 

The SL has three focusing parameters: 

• Mode: AFs, AFc, MF
• AF mode: Static, Dynamic [Tracking], Auto [Face]
• AF field size: 1 Point, Field (all focus points), Zone (3x3 points)

{AFs:Static:1 Point} is indicated by a small cross. 

{AFs:Dynamic:1 Point} is indicated by a rectangular field box. 

... and so forth. The instruction manual was written before the SL firmware was finalized and has many small mistakes in it. 

 

When I use AF with the SL24-90 lens, I generally use {AFs:Static:Zone} or {AFs:Auto:--}*. These give me reliable focusing for most subjects. When they are inconsistent, I switch to MF with AF on demand by pressing the joystick button. 

 

* Auto [Face] mode does not take a field size. 

[profus]

What about those 529 AF points ( as announced for fw 2.0 ):

 

Leica released firmware update 2.0 for their SL mirrorless camera system. Some of the improvements are improved AF, 1/16,000 shutter speed, 529 AF points (up from 49)

 

Are those really available?

[Leicaiste]

Reading this topic, I am now wondering if the AF of the SL will be good enough for the future Leica Summilux-SL 50 mm f/1.4 ASPH. @ f1,4.

[steppenw0lf]

Reading this topic, I am now wondering if the AF of the SL will be good enough for the future Leica Summilux-SL 50 mm f/1.4 ASPH. @ f1,4.

 

I'd like to have your problems ...     simply wait and see.    

Stephan

 

In the meantime use a Summilux-R 1.4/80 with focus peaking. Maybe this will give you some confidence again.

Or maybe you have a Contax 1.2/85. Or have you tried a Noctilux (any version) ? It works quite nicely with focus peaking. (ok, maybe not if it is really dark).

By the way, a 280mm with f 4 has less depth of field than a 1.4/50 .

[steppenw0lf]

What about those 529 AF points ( as announced for fw 2.0 ):

Marketing is for making you buy things, it's not for making you understand things.

Maybe you should ask Panasonic, I think it is their invention (at least they made the base for it). Any hints ? Maybe any Japanese Leica users, who are able to read Japanese documents ?

[Advertisement]

Hallo,

Look here: Leica L-Mount

[profus]

Marketing is for making you buy things, it's not for making you understand things.

 

 

Aware of that, but still...they shouldn't be able to write just about anything and get away with that without any explanation. Or...?