How do I focus a rangefinder when the object of focu is outside of the rangefinder patch?

Hello all!

I think this is a stupid question, but I don't know the answer, so I am asking! I have used a 35mm rangefnider, but usually just zone focus with it. On the flip side, I am using a Mamiya Press MF camera and trying to take much more careful photos. I usually shoot with small apertures, but not always. My rangefinder patch is very clear in most lighting conditions.

However, what if the item I want to focus on isn't where the patch is? As an example, what if I want to take a portrait, but the person isn't in the center of the photo, and I am shooting at say f2.8? My thought has been to move the camera, focus on the person with the rangefinder patch, then compose as I want -- but does that work? Doesn't moving the camera slightly change the focus? I exclusively shoot on a tripod with this camera, so I tend to have control and time and so on. Still, I am very curious about how this is supposed to work.

Thanks!

Focus on the object that you want to be in focus, then swivel the camera to frame as you want. There is a theoretical difference because the off-axis distance-to-subject is slightly different than the on-axis distance-to-subject for a flat field, but in practice for reasonable amounts of subject off-axis and reasonable subject distances (not trying to do macrophotography), this theoretical difference should be covered by depth of field and it isn't worth worrying about.

reddesert said:

Focus on the object that you want to be in focus, then swivel the camera to frame as you want. There is a theoretical difference because the off-axis distance-to-subject is slightly different than the on-axis distance-to-subject for a flat field, but in practice for reasonable amounts of subject off-axis and reasonable subject distances (not trying to do macrophotography), this theoretical difference should be covered by depth of field and it isn't worth worrying about.

Click to expand...

Great, that is what I thought and have tried, but haven't gotten film back yet. Realistically, I think my camera can only shoot at f5.6 or something, and this is almost never relevant for the big landscape photos I take -- but I was curious. Thank you!

The amount of correction needed depends on the swing angle:

corrected distance = RF measured distance * COS (swing angle)

For example, if you swung the camera 10 degrees then the corrected distance would be 0.985 the measured distance, a correction of 1.5%. For a 45 degree swing the correction would be 0.71 or 30% (it should be said that this is placing the focused object at the edge of a 35mm frame and using an 18mm lens).

This gives an added use for a tripod's protractor scale.

reddesert said:

Focus on the object that you want to be in focus, then swivel the camera to frame as you want. There is a theoretical difference because the off-axis distance-to-subject is slightly different than the on-axis distance-to-subject for a flat field, but in practice for reasonable amounts of subject off-axis and reasonable subject distances (not trying to do macrophotography), this theoretical difference should be covered by depth of field and it isn't worth worrying about.

Click to expand...

The above statement captures the method that I have used for many decades when using a range finder camera.

The shift of focus distance that occurs when focus-recompose methdology is employed is most significant when at closer distances with wider FL, and the larger angle of change (during recompose) is greater.
For example,

1. 100mm lens seens 20 degree AOV on the long dimension of 135 format. If you focus-recompose by 5 degree angle shift (1/4 the frame height) or 10.4" shift of scene for 8' shooting distance, that introduces almost 0.61" distance error in the focus distance...with a DOF zone of 1.34', the f/2.8 DOF can mask the focus error.
2. 50mm lens seens about 40 degree AOV on the long dimension of 135 format. If you focus-rcecompose by 11 degree angle or 10.4" shift of scene for 4' shooting distance, that introduces a 1.02" distance error in the focus distance...with a DOF zone of 1.34", there is more focus error than f/2.8 DOF zone can mask.

both these situations frame the same amount of subject in the frame, assume the same amount of shift in the framing in order to focus within the rangefinder area. The same issue affects SLRs with center focus aid, as it affects rangefinders, it isn't unique to rangefinders. SLRs do have benefit of groundglass surround being able to also focus, albeit with less precision.

'Focus re-compose' is what the technique is called.
Very common practice with centre focus spot & split-screen focus cameras.

Unless you are working with extremely narrow depth of field, "Focus and then re-compose" only creates an issue when you are so close that parallax is also an issue.
Even with an SLR, I often focus on an edge or a line or a detail that is close to, as an example, the subject's eyes, and then re-compose.

Now introduce the lens' field curvature into the equation and you will have people believe rangefinder cameras can only be used at infinity and f16

Yeah, and great camera and lens. Shoot wide open if you want. 5.6 is going to be in the sweet spot.

To beat this dead horse until it really can't get up, I think we mostly agree that in realistic cases you don't have to refocus. (Field curvature typically goes in the direction that would reduce the need to refocus, as well.) With some approximations, it is possible to calculate a fairly simple formula to compare focus offset to the depth of field or depth of focus.


Here, the blue ellipse off-axis shows the subject. If we focus on it at a distance d_focus, then swing the lens by an angle theta, the plane of focus (magenta dashed line) is slightly behind the desired subject plane (assuming a perfectly flat field lens). The amount of offset in subject space is x = d_focus * (1- cos theta) as was mentioned in a previous post. In the image space, the plane of focus is slightly in front of the desired image plane, and recomposing moved the image sideways by a distance h = d_image sin theta. If we assume that the subject is at a further-than-macro distance, then d_object >> f and d_image ~= f, so the image moved sideways by h ~= f sin theta.

It is easiest to compare the focus offset in the image space to the depth of focus. The extension past infinity, e = f^2 / (d_object - f). You can derive that by rearranging the lens equation 1/f = 1/d_object + 1/d_image. For non-macro distances, d_object >> f, so

extension e ~= f^2 / d_object.

Recomposing shifts the d_object by a small factor, (1 - cos theta), where cos theta is a bit less than 1, and sin theta is a smallish number. Then, we can approximate the change in extension as:

focus shift, delta-e ~= f^2 / d_object * (1 - cos theta)

This is how much the image moves forward-back in the image space. And it moves sideways by h ~= f sin theta. For smallish angles, we can approximate (theta = angle in radians, usual Taylor series expansions of cosine and sine, approximations are probably good to 10% or better):
cos theta ~= 1 - theta^2 / 2.
sin theta ~= theta

So focus shift ~= f^2 / d_object * ( theta^2 / 2) = h^2 / (2 * d_object).

And we want to compare this to the one-sided depth of focus = N * c, where N = f-number and c = circle-of-confusion. So we get a minimum subject distance d_object to stay within the depth of focus for a given sideways image shift h:

d_obj_min = h^2 / (2*N*c)

What's neat is that most of the variables like focal length have dropped out. Let's plug a couple of reasonable examples into this. Assume that we are recomposing by a sideways shift h that moves the subject from the center of the frame to halfway between center and edge (because in reality we wouldn't shift the subject all the way to the edge). So we shift by h = 1/4 of the frame width.

On 35mm format: h = 9mm, c = 0.03mm, and say N=2 for an f/2 lens used wide open. Then d_obj_min = 675mm.

On 6x6 format: h = 14mm, c = 0.05mm, and say N=2.8 for an f/2.8 lens used wide open. Then d_obj_min = 700mm.

So for reasonable assumptions, even for a wide open lens, if the subject is farther away than about 1 meter, the focus shift caused by recomposition is covered by depth of field / depth of focus. If the subject is very close, like macro, then some of the approximations are not valid, but for macro it would be best to use an SLR and focus on the subject on the ground glass rather than recomposing, anyway.

If your camera is clamped to a tripod and you need to reframe do you move the camera or the whole tripod? I always move the tripod, so if I need critical focus on a subject which is not center, its a lateral movement of the entire system so the distance to the plane of the subject doesn't change. I see this as a compositional issue, which will also reduce the possible DOF issue.

Reality check - focus and then reframe. LOL

reddesert said:

So for reasonable assumptions, even for a wide open lens, if the subject is farther away than about 1 meter, the focus shift caused by recomposition is covered by depth of field / depth of focus. If the subject is very close, like macro, then some of the approximations are not valid, but for macro it would be best to use an SLR and focus on the subject on the ground glass rather than recomposing, anyway.

Click to expand...

To further beat the rotting horse corpse, it should be recognized that reddesert' analysis is based upon the false assumption of 'manufacturer's DOF' assumption about the size of the CofC, c = 0.03mmm (rounded, actually 0.01" or 0.025mm) ...in fact, the human eye corrected to the optomitrists' goal for corrected vision (20/20 in US, 6/6 in Europe) is about 3x more precise, that is, one can detect 'sharp' vs. 'out-of-focus' in a CofC that is 1/3 the size of 'manufacturer's DOF' assumption about the size of the CofC.

So focus-recompose detection of focus error induced by an angle change of the lens, (rather than a lateral move of the camera), might NOT be acceptable to your eye in reality. In my post 6 of this thread, I used the 20/20 vision detection capability of 'out-of-focus'. So using 'manufacturer DOF' the DOF zone is 2.52" deep at f/2.8, yet it is only 0.83" deep assuming 20/20 vision.

reddesert said:

Focus on the object that you want to be in focus, then swivel the camera to frame as you want. There is a theoretical difference because the off-axis distance-to-subject is slightly different than the on-axis distance-to-subject for a flat field, but in practice for reasonable amounts of subject off-axis and reasonable subject distances (not trying to do macrophotography), this theoretical difference should be covered by depth of field and it isn't worth worrying about.

Click to expand...

This works with manual focus, as on your rangefinder. If someone is using an autofocus camera like an SLR, you have to hold the shutter button down halfway to lock in the focus before moving the camera to reframe. If you;re SLR camera has one, you might want to use the button that locks in the auto exposure setting when the setting seems right.

wiltw said:

To further beat the rotting horse corpse, it should be recognized that reddesert' analysis is based upon the false assumption of 'manufacturer's DOF' assumption about the size of the CofC, c = 0.03mmm (rounded, actually 0.01" or 0.025mm) ...in fact, the human eye corrected to the optomitrists' goal for corrected vision (20/20 in US, 6/6 in Europe) is about 3x more precise, that is, one can detect 'sharp' vs. 'out-of-focus' in a CofC that is 1/3 the size of 'manufacturer's DOF' assumption about the size of the CofC.

So focus-recompose detection of focus error induced by an angle change of the lens, (rather than a lateral move of the camera), might NOT be acceptable to your eye in reality. In my post 6 of this thread, I used the 20/20 vision detection capability of 'out-of-focus'. So using 'manufacturer DOF' the DOF zone is 2.52" deep at f/2.8, yet it is only 0.83" deep assuming 20/20 vision.

Click to expand...

Since I move the focusing spot to the alternate focus place and back, there is no focus error. Therefore the focus error analysis is most properly classed as a WOMBAT*.



WOMBAT* Waste Of Money Brains And Time. -- a term I learned while working at Kodak in the Harkeye Building.