Jump to content

Measuring for focus - perpendicular distance or trigonometry?


Recommended Posts

I've worked on a few smaller productions where people have used a tape measure to pre-measure distances to actors' marks so they can ocus to them during a take.

 

From memory (it was a while ago) they measured from the mark on the camera to the actors' mark, so the line of measure is usually off at some angle from the perpendicular to the film plane.

 

I think the distance scale on lenses specificy perpendicular distance between the film plane and the plane of focus. That would be shorter than the measured distance typically.

 

Do focus pullers use trig to figure out the distance they should pull to on the lens, having only measured the direct distance? Or do they measure perpendicular distance?

Edited by Mei Lewis
Link to comment
Share on other sites

I have not seen any 1st ACs use or calculate trig on a regular basis - it will never be time effective

 

ACs use intuition and experience to adjust

 

While you can calculate the precise distance to the film plane, on the day it wont matter as much - the actor may overshoot the mark, or lean in further, or the dolly will have to adjust because someone popped into the frame, etc

 

You could use trig if the camera is way up on the crane and subject is too far for cinetape to read - use laser disto to get length to subject and height of the camera

Link to comment
Share on other sites

  • 4 years later...

Nearly five years later and I've been on many sets, but still don't know the answer to this one.

I know what I do, but that's probably considered a hack by most ACs.

Here's a concrete example. The difference in distances can be quite significant. Here the diagonal measured distance is 2.2m, whereas the focus distance should be set to 1m.

 

focus-trig-small.jpg

Edited by Mei Lewis
Link to comment
Share on other sites

  • Premium Member

I find this the most incomprehensible topic of modern cinematography. The camera has a reflex finder system but it is not used to determine focus points. It is fully clear to me that one person can’t frame and pull focus perfectly all the time, so an assistant takes care of quick or also elongated focus adjustments. But measuring tape when you can accurately focus on a GG and set marks from there?

In the dark, in dim light, filtered, no question, then focusing by the finder can become very difficult, sometimes impossible. Still, with a professional mirror shutter reflex camera I’d rather rely on what I see than fumble around with sagging tape, even oblique as put up by OP. And I know what too long focusing is, have had it as a producer when a camerawoman gave us the curtain behind the interviewed person sharp, the face soft. At ten feet. We had one take. She had an Arriflex 16 SR on tripod. Every cameraperson ought to be able to master that.

I don’t understand it. Was the invention of the reflex finder in vain? Camerapeople all weak-sighted? Measuring tape was already unnecessary with the rackover systems of Bell & Howell (1912) and Mitchell (1917). ?

Link to comment
Share on other sites

34 minutes ago, Simon Wyss said:

I don’t understand it. Was the invention of the reflex finder in vain? Camerapeople all weak-sighted? Measuring tape was already unnecessary with the rackover systems of Bell & Howell (1912) and Mitchell (1917). ?

Sorry Simon, but you've evidently never been an AC. Measuring with a tape is done because it gives you actual distances, rather than vague marks on a focus wheel. If you know your focus distances, then there is no need to recheck focus after lens changes. Using a tape measure means that the focus puller can do his or her job without looking through the camera, without distracting the camera operator. Focusing by eye requires a target to focus on, using a tape does not.

Before the advent of digital video and high quality, sharp monitors, this is the way it was done, efficiently and effectively, for many years.

  • Like 1
Link to comment
Share on other sites

If you are shooting film, the tape measure is the necessary tool for following focus.  The reality is that extreme angles are only possible with wide lenses, where the difference in focus setting is not so precise.

There are times, when even using a focus pulling crew member, that the camera operator may see something out of focus and grab the focus knob and make an adjustment by eye.  And usually, the operator, with his/her eye in the viewfinder is the person to confirm that a shot is in fact, in focus.  If there is any doubt about the focus mark for the lens, due to extreme non-centering of the subject, it is always possible for the 2nd assistant to go to the actors mark with a focus target that's easy to see, and set the focus by eye through the viewfinder.

All of this of course goes out the window when using a remote focus device on a Steadicam or crane etc.  And here the operator can not see fine focus in the video assist.

Lastly, in near 40 years on the set, I've never seen a focus puller measure to any spot other than the spot where the actor will be standing.  I've never seen trigonometry used for focus marks either.  And I've never shot 35mm film without a focus puller.  There have been a few special occasions where I've operated a camera and pulled focus by eye though.

If you have any doubt about the correct way to measure, I suggest that in camera prep, you have the focus team run a tape measure to a mark and then pan the camera so that the target is near the far side of the frame and see if it still looks in focus through the viewfinder.  Because you won't be actually shooting, you can use a good amount of light so that you get a clear view of the focus.  If there are any discrepancies, note them, by lens, for use durning filming.  

I think you'll find that if you test motion picture lenses by aiming at a flat object, that generally the flat object stays in focus across the frame.  Which means that a straight tape measurement, without trigonometric compensation, will give the correct focus mark for off center subjects.

  • Like 1
Link to comment
Share on other sites

I thought Mei was referring to the problem from the field curvature.  On the object side of the lens the focal plane can be curved or partially corrected to almost flat. I always assumed that the focus marks on the barrel were for objects on the lens axis.  I know that there were some even quite old lenses (Zeiss planars) that were designed with quite a flat focal plane. What the shape of the focal plane is for the latest cine lenses vis a vis say,  super speeds I don't know. The lens techs will know.

So there is quite a good question in there about whether or when, and how the good focus pullers need to allow for the field curvature.

 

 

Edited by Gregg MacPherson
Edit: spelling
Link to comment
Share on other sites

  • Premium Member
7 hours ago, Mei Lewis said:

Nearly five years later and I've been on many sets, but still don't know the answer to this one.

I know what I do, but that's probably considered a hack by most ACs.

Here's a concrete example. The difference in distances can be quite significant. Here the diagonal measured distance is 2.2m, whereas the focus distance should be set to 1m.

 

focus-trig-small.jpg

That example is a pretty extreme one - you’d need a lens with a field of view of more than 120 degrees, so something like an 9mm lens on S35, and be wanting to focus on a subject at the very edge of the frame. I doubt that happens often, but even if it did the depth of field of a 9mm lens would probably cover most things except at close distances. 

More realistically, for a wide angle you might have an 18mm and a subject a bit in from the edge, in which case at say 6ft from the perpendicular the subject might measure 6’9. Now with an 18mm lens on S35 even at f/1.4 your depth of field will cover from a bit over 5’ to a bit over 7’, easily covering 6’9. The longer the focal length the narrower the field of view, and so the discrepancy in measured distance from centre to edge of frame diminishes.

Modern lenses tend to have very flat fields, an older wide angle lens may have a more pronounced curvature to the plane of focus, in which case objects at the edge of frame may need more care to achieve sharp focus.

  • Upvote 1
Link to comment
Share on other sites

12 hours ago, Gregg MacPherson said:

So there is quite a good question in there about whether or when, and how the good focus pullers need to allow for the field curvature.

I’ve never heard an AC mention field curvature in that context, much less try to compensate for it. Lens manufacturers don’t publish that kind of data, if it even exists in a readily understandable form.

In any case, focus pulling is as much an art as a science. Unless you are shooting static objects from fixed positions, there are generally so many variables due to movement that exact mathematical calculations would be pointless.

Link to comment
Share on other sites

I wasn't referring to field curvature.

AFAIK they *try* to design lenses so that the focus plane _is_ a plane. They're far from perfect, but I *think* the effect is much less than the one I'm talking about here.

I've not seen the curves for any cine lenses, but I have seen them for stills lenses. 

Lensrentals have some interesting charts the curvatures are typically quire complex:
https://www.lensrentals.com/blog/2014/09/field-curvature-and-stopping-down/

Link to comment
Share on other sites

On 12/20/2019 at 9:55 PM, Bruce Greene said:

...

If you have any doubt about the correct way to measure, I suggest that in camera prep, you have the focus team run a tape measure to a mark and then pan the camera so that the target is near the far side of the frame and see if it still looks in focus through the viewfinder.  Because you won't be actually shooting, you can use a good amount of light so that you get a clear view of the focus.  If there are any discrepancies, note them, by lens, for use durning filming.  

...

 

I've done effectively the same experiment many times with a full frame stills camera, while trying to use the 'focus recompose' technique for getting focus. It doesn't work well for wide apertures and close subjects, so I went about figuring out why, and realised it's because the plane of focus stays parallel to the sensor plane, and therefore moves behind the subject if you reframe them from the centre to the edge of frame.

Just tried it again with a 24mm f1.4 lens on full frame, rotating in my office chair, and the subject, a tea cup about 3 feet away, visibly goes out of focus.

The linked article mentions the issue I'm asking about (it's point 4) so it's well known in still photography.

 

On 12/20/2019 at 9:55 PM, Bruce Greene said:

...

I think you'll find that if you test motion picture lenses by aiming at a flat object, that generally the flat object stays in focus across the frame.  Which means that a straight tape measurement, without trigonometric compensation, will give the correct focus mark for off center subjects.

I think that a flat object _parallel to the sensor plane_ stays in focus across the plane (ignoring field curvature, mentioned elsewhere), but that's why measuring diagonally won't work for off center subjects.

But it really seems like from all the answers that it's 'good enough' in most cases. Maybe that will change with increasing use of large format sensors in video (Alexa LF etc).

Link to comment
Share on other sites

Interesting thread, but way too technical for me. A lot of my focus work is zone focus. I can't be doing equations to figure it. Big budget films use stand-ins that would be used for focus and lighting set ups. I don't think they would have to be concerned with all this.

I think the closer you are the more critical the focus difference is between the 2 methods when it comes to this discussion. For instance a test of the 2 measuring methods produce a 59" distance when measured level from the camera vs 82" for the distance angled to the ground under the focus point. If the subject was far away I don't think there would be much of a difference.

 

Edited by Daniel D. Teoli Jr.
Link to comment
Share on other sites

3 hours ago, Mei Lewis said:

I wasn't referring to field curvature.

My bad. What I should have said is that some interesting thoughts and questions about the field curvature are implicated in your topic. The plane of focus on most of the lenses being used will not actually be flat, but a deformed shape generically described as having field curvature. Not necessarily a simple arc in its section.

The science is really interesting, and some interesting info turns up easily on the web that non academics can read. In a piece by Roger Cicala, the guy Mei refers to, he looks at the field curvature on a CP2 lens. May be worth a look...

https://www.lensrentals.com/blog/2017/11/testing-lenses-best-individual-focus-mtf-curves/comment-page-1/

Scroll down to what he calls Best Individual Focus (BIF). Looks clear that one can extract much better performance from a lens if one knows the shape of the field curvature. Not to ignore the obvious point, that longer lenses can be focused by eye, at least they could be on film cameras, I don't know the digital systems.


Whether or how science can trickle down into something useful on set is another issue.  If the plane of focus was always flat and the focus puller was worried about Mei's offset from the lens axis, then the correction could be very quick, a few seconds with the calculator on your phone. Multiplying by the cosine of the estimated offset angle would be my choice.

More interesting though....Imagine a laser device measuring distance that is aware of its angular relationship to the lens axis, which has an on board database of field curvature profiles for common lenses, and can calculate an accurate focus distance for the lens, based on the measured distance to the offset object.

Link to comment
Share on other sites

Thanks Greg, and thanks for the link.

Having read the link I now think that field curvature is also an issue, in addition to the "measuring the wrong distance" problem my original question was about.

I mostly work on small productions and operate and focus pull myself, using the image being recorded, looking at an electronic viewfinder or screen. This mostly avoids potential issues with field curvature and trigonometry/measuring to the wrong point , but is susceptible to other issues e.g. lack of resolution on screen/viewfinder to know if something is truly in focus.

Link to comment
Share on other sites

 

 

 

13 hours ago, Daniel D. Teoli Jr. said:

...

I think the closer you are the more critical the focus difference is between the 2 methods when it comes to this discussion. For instance a test of the 2 measuring methods produce a 59" distance when measured level from the camera vs 82" for the distance angled to the ground under the focus point. If the subject was far away I don't think there would be much of a difference.

 

Yes, the closer the more it's an issue.

But this isn't about measuring from the ground or not. The intention with my original post is that everything is happening at roughly the same distance from the ground, which is fairly typical, the camera is most often at roughly eye level and it's the eyes that need to be in focus. So distances are measured in a plane parallel to the ground and at roughly camera/eye height.

That's why I indicated 'top view' on both my pictures. I'm ignoring to up-down dimension completely, for simplicity, and because I think it's mostly not significant.

Link to comment
Share on other sites

Here's one pf the images related to field curvature from Greg's link to lens rentals. https://www.lensrentals.com/blog/2017/11/testing-lenses-best-individual-focus-mtf-curves/comment-page-1/

I think I understand most of the rest of that page, but not this diagram. Specifically I don't get why there are two graphs. What do 'tangential' and 'saggital' refer to here?

This page https://www.optics4kids.org/what-is-optics/refraction/aberrations explains the terms but I can't see how they relate to these graphs.

It seems to me that because a lens is circular only one graph would be needed, showing information about focus at various distances out from the line that goes through the center of the lens. I had previously thought that's what the left hand graph was, because I'd seen similar graphs on their own. But now I don't know!

 

Zeiss-cp.2-50-SS-T5.6sml.png

Link to comment
Share on other sites

Also, I *think* the horizontal axis is the distance across the sensor from its centre, and 20mm for that makes sense, because 40mm is roughly the diagonal of a full frame sensor, for which the .cp2 lenses are designed.

But what's the vertical axis? It's described as the focus position in mm, is that referring to distance in front of and behind the sensor plane?

If that is the case, can we assume similar pattern out in front of the camera, where we can actually measure things?

I really want to be picturing these charts like this, but I'm fairly sure this is wrong!

focus-recompose-error-v2.jpg

Link to comment
Share on other sites

13 hours ago, Mei Lewis said:

Here's one pf the images related to field curvature from Greg's link to lens rentals. https://www.lensrentals.com/blog/2017/11/testing-lenses-best-individual-focus-mtf-curves/comment-page-1/

I think I understand most of the rest of that page, but not this diagram. Specifically I don't get why there are two graphs. What do 'tangential' and 'saggital' refer to here?

This page https://www.optics4kids.org/what-is-optics/refraction/aberrations explains the terms but I can't see how they relate to these graphs.

It seems to me that because a lens is circular only one graph would be needed, showing information about focus at various distances out from the line that goes through the center of the lens. I had previously thought that's what the left hand graph was, because I'd seen similar graphs on their own. But now I don't know!

 

Zeiss-cp.2-50-SS-T5.6sml.png

I  too am puzzled by how different the results are for the Tangential and Sagittal planes. I thought that tangential and sagittal planes were just orthogonal vertical and horizontal planes intersecting the lens axis. I thought that "spherical" lenses would have almost identical results there for tangential and sagittal. 

So clearly, I don't understand the axial definitions or how to interpret the results yet. Maybe Simon can steer us. He loves a good quasi academic correction.

Roger Cicala seems like a guy who might entertain an explanation.

I visualize the field curvature on the object side of the lens. I think the graphs here are trying to illustrate the sectional shape of the field curvature by mapping the image results, showing the needed shift along the lens axis to achieve focus at any point in the image. The relevant distance shift is up to 0.2mm, which seems like a lot at the image plane. I'm just trying to translate that into distance increments on the object side of the lens.
 

Link to comment
Share on other sites

  • Premium Member

The difference between tangential and sagittal sharpness in a lens is one of the main aberrations a spherical lens naturally produces, called astigmatism. All MTF curves map the contrast ratios of line pairs that are both tangential (in the direction of a wheel hub) and sagittal (in the direction of wheel spokes) because they always vary. A lens design always needs to balance the two different curves of tangential and sagittal aberration to create an overall correction.

In that Lens Rentals article, Roger is mapping how the MTF of both measurements changes when you shift the lens slightly forward or behind the point of best centre focus, and finding where the best MTF is as you move away from the centre. So it's mapping the field curvature in image space (behind the lens), but that will have a corresponding curve in object space (in front of the lens).

So in that example, if you shift that 50mm Zeiss by 0.05mm you'll get a sharper image in the outer 1/3 of the image. Shifting the lens is changing where the lens focusses out in front, so depending on which part of the image you want sharpest there will be a curved plane of sharpest focus. 

It's worth remembering those MTF curves are measuring different frequencies of line pairs that in the real world can correspond to minute shifts of what we perceive as sharpness. The slight softening of the outer edges of a test chart is usually something not noticed in a moving image of 3 dimensional space, especially in modern lenses. Still photographers can get a little carried away with this sort of pixel peeping.

Different lenses will have different amounts of field curvature, but in terms of adjusting focus to compensate for it most ACs I've met don't worry too much about it, if at all. In many cases I think it's too subtle to be noticed. Sometimes a lens with fairly extreme curvature (perhaps an older anamorphic or wide) will have its focus marks set for a point a bit off-centre, so that the focus is more evenly spread across the field. If an AC is zooming right in to the centre to check focus they may complain that the focus is a bit out, so I will show them on the projector how the focus scale has been set. Once I explain that if I set the focus marks for maximum centre sharpness the rest of the frame would get progressively softer as you moved towards the edges, they understand the rationale. 

Link to comment
Share on other sites

On 12/23/2019 at 6:04 AM, Dom Jaeger said:

All MTF curves map the contrast ratios of line pairs that are both tangential (in the direction of a wheel hub) and sagittal (in the direction of wheel spokes) because they always vary.

Thanks Dom!

That wheel comparison has really cleared up my understanding of what tangential and sagittal refer to in this case. I think!

Link to comment
Share on other sites

 

I just chanced on this video

Which shows a quick way to examine the field curvature of lens using a photo of some tarmac and photoshop's 'find edges' filter. It actually refers back to another lens rental blog post
https://www.lensrentals.com/blog/2019/11/stopping-down-some-bargain-primes-and-zooms/

Here's the example the video makes (it's dark and raining outside here, will directly test myself another time).

aop-v2.thumb.jpg.c4ebad4fbbc27f5e56ad9b0836e25d97.jpg

Link to comment
Share on other sites

On 12/20/2019 at 4:36 PM, Stuart Brereton said:

Sorry Simon, but you've evidently never been an AC. Measuring with a tape is done because it gives you actual distances, rather than vague marks on a focus wheel. If you know your focus distances, then there is no need to recheck focus after lens changes. Using a tape measure means that the focus puller can do his or her job without looking through the camera, without distracting the camera operator. Focusing by eye requires a target to focus on, using a tape does not.

Which AC's job is it to calibrate the flange distance on each lens? I tried doing it on one of mine and could get it close but never exact. Or is that the job of the rental house when preparing a camera/lens package?

I'd love to go purely off of measurements because I'm sick of having to squint-eye the monitor all the time.

Edited by Max Field
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...