Focus Accuracy

[roblumba]

I'm new to rangefinder focusing and rented an M6 with a 35 Summilux and noticed that the focus recompose was way off. I heard that rangefinders are supposed to be accurate for wide angle. But how do you guys deal with focus and recompose?

Let's say I have a scene with a person close to me and to my left. I focus using the center (of course), and then move the camera to place the person off to the left. However, now the focus is actually in the wrong place.

I can think of two solutions.
1. Guestimate the new focus distance (short distance), do focus bracketing and pray.
2. Focus and compose on the center, and crop later.

Both methods seem like a downgrade from my 1DMarkII method of choose a focus point that is closer to the subject. In the case of the 1DMarkII, the 45 autofocus points have very good coverage over the frame because of the 1.33 crop factor. So this is actual very accurate.

Any better ideas or help would be greatly appreciate as I'm seriously considering the M8 (on the waiting list), and want to make sure I can focus properly, especially at wide aperatures.

 

[gdewitt]

I'm unclear on why the focus is in the wrong place? If you haven't moved closer or farther from the subject, how is the focus in the wrong place?

 

[roblumba]

Focus wrong place.

Focus wrong place.

Because if you draw a straight line from me to the subject, let's say it's 5 feet. This is like the radius of a circle. As I rotate the camera to the right, the focus plane rotates with me at a distance of 5 feet from me. When I stop rotation, the plane of focus is still 5 feet from me. However, the subject is no longer located on that plane of focus. They are inbetween me and the plane of focus.

 

[Plasmat]

Whatever object you focus on in the first place will still be in focus no matter where it is in the frame if you move the camera while you are physically standing in the same place.

If you focus on an object, say, five feet from the plane of focus, that will be the point that is in focus, unless you refocus the lens on something else. Depth of field will more than compensate for most situations if you're stopped down a few stops.

Maybe you THINK it looks out of focus in the viewfinder for some reason, such as that you're wearing bifocal eyeglasses/ (not likely but possible).

What makes you think otherwise? I think you're confused about how a rangefinder camera operates.

 

[Plasmat]

I'm thinking. Still thinking.

 

[roblumba]

Confused

Confused

I hope I am confused. I was just looking at some of the photos I took over the weekend and that was one of the conclusions I came to. The chance my focus was off is very high since this was my first time with a rangefinder. However, I know there are some difficulties with focus recompose. I wish I could draw the diagram of what I am talking about. I found a website that describes the problem that I am talking about.

http://www.mhohner.de/recompose.php

It talks about AF cameras, but the same problem applies to rangefinders.

 

[rogue_designer]

I hope I am confused. I was just looking at some of the photos I took over the weekend and that was one of the conclusions I came to. The chance my focus was off is very high since this was my first time with a rangefinder. However, I know there are some difficulties with focus recompose. I wish I could draw the diagram of what I am talking about. I found a website that describes the problem that I am talking about.

http://www.mhohner.de/recompose.php

It talks about AF cameras, but the same problem applies to rangefinders.

Ah - now I see what you mean. Yes, the effective plane of focus can shift somewhat, especially if you use a very strong adjustment in the frame (left or right). This will be a problem with any camera that does not have variable focus points - not just rangefinders.

You can compensate with depth of field by not useing the widest apertures, or you can guestimate the new distance to your primary subject. But long story short - the more of a recomposition shift, the more your focus will be suspect - You can focus very accurately indeed with an RF, but that requires that you keep your subject pretty close to the plane of focus you chose. 😉

Partly this is a shift in compositional styles between an RF and an SLR. I find that with an SLR I am much more meticulous about the relations between elements, focus, etc. With a rangefinder, I tend to be much more spontaneous, and give myself a little "wiggle room" to compensate for slight focus shift, or possibly inaccurate framelines (more common than you would think).

It's the photography equivalent (for me) of the difference between a flemish old master painter, and an impressionist. Different methods, different results, but both well suited to their purpose.

 

[roblumba]

Usually, I'm already standing in a position that frames the person. So I don't need to move them into the frame. So when I am in position, I need that person to be the focus of the frame. How do I do that with a rangefinder?

 

[Plasmat]

We are nitpicking here. Unless you're shooting wide open with a Noctilux, a Zunow f1.1, or a Canon F0.95, I doubt very strongly whether even moving your whole upper body would be detectable in the image sharpness, and even then it's highly questionable.

Operator error in focussing from shot to shot, or camera shake would have far more effect.

Theoretically, yes, moving ones body could throw off the focus in the arc, but in practical picture-taking, I think it's a moot point.

But yes, if you put it that way, you are minisculy changing the plane of focus.

 

[roblumba]

Focus Plane moved

Focus Plane moved

I just noticed this effect with the 35 Summilux at F2. The object was my son, and he was approximately 3 feet from me on my far left, and then I rotated to the right and noticed the focus plane had moved about 1 foot beyond him. I think this is probably a rare case as most of my photos are not like this. P

erhaps it is a moot point and I would need to have an adjustment in my style. It's just one of the tricks I like to play with the 1D MarkII. The focus points cover a large part of the frame and I'm able to accurately focus on objects in the extremes of the frame with a lense that's wide open (f2.8).

This problem is exaggerate at very close distances. The Canon 24-70 2.8L is actually capable of minimum focus distance of 15 inches. That makes for some very interesting compositions. But the DOF is only a few inches at that f stop and distance. Compared to a 28 elmarit with a min focus distance of 2.7 meters. The DOF would be magnitudes greater at that focus distance. I think .5 ft DOF? Still the DOF is small enough that a person will notice the difference.

I think what's most distracting is when the focal plane is actually visible in the photo and it's not the subject! This is when the shift in focal plane is annoying. Let's say I focus on the eyes and instead I get razor sharp ears or arms. Not very pleasing.

 

[roblumba]

Focus plan is a sphere

Focus plan is a sphere

Check out that link I posted. The problem is that the focus plan is flat and perpendicular to the line that is create when we aim at something and focus upon it. However, the distance between the camera and the focal plane is fixed. When we rotate, we take the entire focal plane with our rotation and swing the focal plane away from the object we were once focused upon!

😕😱

 

[rogue_designer]

I think the problem comes in, that I don't think that the focus plane stays entirely flat. If it did, then somebody on the far right of the image might be several feet further away (on the diagonal) - than someone directly in front - it should be somewhat radial no?

Maybe that varies with lens design?

 

[Topdog1]

There is no problem here

There is no problem here

If you view the world as a sphere that surrounds you, then the focal plane is a plane that lies tangent to that sphere where your focus lies, i.e. it makes a 90 degree angle with a radius of the sphere having you as the center. As you rotate around the sphere, assuming you don't move your feet, the tangent plane rotates with you, moving neither in nor out. Hence the original focal plane remains the same distance from you as the new one. Ergo, there is no problem here. 🙂

Regards,
Ira

 

[roblumba]

Tango with the Tangent

Tango with the Tangent

Okay, I can understand why the camera would actually have a focus sphere rather than a focus plane. After all, the light entering the camera is entering from multiple angles, with each having it's own distance from the lens and thus independent of where the camera is actually pointed and more dependent on the set focusing distance of the lens.

Perhaps the problem with rotation lies in the fact that we don't normally rotate the camera on the lens axis, but we rotate from our body axis. There are actually special panaramic tripod heads that are designed to rotate the camera properly when stitching together multiple panaramic shots. Perhaps this is the problem that I am experiencing. So really, I am the problem! 😱

 

[roblumba]

Ferider used the center?

Ferider used the center?

So Ferider, did you actually use the center focus point on that pic and recompose?

 

[rolo]

Roland ,
you have a 1 in 27 chance of being in focus.

Made the number up, but it will seem like that.

It's precisely why Canon push the 45 focus points in the 1D so much. Within 15ft you'll not be sharp if you move the lens at <f2, without compensating.

Depth of fieldwill carry you through.

Better maybe to keep the composition and crop.

 

[didjiman]

...
Perhaps the problem with rotation lies in the fact that we don't normally rotate the camera on the lens axis, but we rotate from our body axis. There are actually special panaramic tripod heads that are designed to rotate the camera properly when stitching together multiple panaramic shots. Perhaps this is the problem that I am experiencing. So really, I am the problem! 😱

It is probably you that is the "problem." However, it probably has less to do with you rotate the camera, but how you MOVE your camera or body. Some people may unconsciously move the body a bit (for example pitch forward) when they turn etc.

 

[Topdog1]

The center of the original focus plane is still the same distance away after you rotate. It has to be, because the tangent plane of focus touches the sphere at only one point. All points on the rotating plane of focus are the same distance from the center of the sphere. What changes is the distance of the off axis rays. If you rotate right, the rays to the right of the tangent point get closer while the rays to the left get farther away. So your central point of focus is still in focus but the outlying points are not. If these outlying points are within the depth of field, everything stays sharp. If not, their blur circle will be exagerated beyond the point of acceptable focus and they will appear less sharp than if you hadn't rotated.

So, the center stays sharp but the oulying edges may be softer than they would otherwise have been if you hadn't rotated, if they are outside the depth of field of the lens at the f-stop you were shooting.

Simple, no?

Regards,
Ira

 

[roblumba]

focus blurs

focus blurs

The objects just inside the dof are overexposed by 1/3 stop on the M8.

... ahh just kidding!

 

[jaapv]

1.A sensor gives a more defined DOF. Also it is judged on 100% crop.😉
2.I attach a diagram about focus-recompose.
3.There is also focus shift. A lens will shift the plane of focus when you stop down. The Noctilux for instance is well known for this effect. It will shift the plane of focus towards the camera when closing the aperture.