How to align camera and flat object ?

[QT Luong]

For the purposes of lens testing, I need to photograph a flat target so that the film plane and the target are strictly parallel. I have found that for critical MTF measurements, even small deviations affect the results.

So far, what I am thinking of doing is materializing the normal (line perpendicular) from the center of the target with two Ts, and then attach a laser pointer to one of them. Then I will know that if the laser beam strikes the lens in the center and I point the camera in such a way that the center of the target is in the center of the image, the two planes will be exactly aligned.

I am aware that there are specialized tools such as the zig-align, but I'd prefer not too spend too much money on that project. One should not assume ground-glass focussing and movements, since I might use the procedures with a variety of cameras. Any other suggestions ?

[David A. Goldfarb]

Not as precise as a laser alignment tool, but another option is to use a level for shots on a copy stand and a compass for tests with the target on a wall. I use a Suunto Tandem clinometer-compass in this way sometimes.

[Michael Gudzinowicz]

There's the old-fashion cheap way...

Take a mirror, mark the center with a 1/4" dot of tape, and place the mirror on the center of your target (coplanar). Set the camera to neutral, stop down a bit, and focus alternating on the reflection of the camera lens in the mirror and the dot, adjusting the camera position to line the dot up with the center of the reflected image of the lens.

[Richard Boulware]

A simple mirror about 2X3", like the one women carry in their purse will work if it is placed at the center of your target subject and secured with some Scotch Magic Tape. Then set up your camera where you can see your own reflection in the mirror, with front and rear standards at normal. At that point you will be within one or two degrees of arc of being parallel. Sometimes taping a tiny light on your camera or lens will help see when your exact reflection hits the center of your GG. When you see the bright light in your ground glass, from the light shining forward from your lens, you will know you are in business.

Another trick I use is to use a laser parallel alignment tool for enlargers. Tape this to the center of your subject target and then use your camera and GG to find the center of the laser beam.

Works like a champ....both tricks.

Best of luck..

[Bill_1856]

You already make great pictures. Quit wasting your time testing lenses -- it is the first step down from photographer to camerajunkie.

[Henry Ambrose]

To make your laser T-square is not going to be that easy. I'm guessing that you are thinking about rotating it while lasing a far off wall to calibrate it to square. How are you going to attach and adjust the laser pointer so that your measurement is precise? You'll need some time and tools to make this T-square laser jig. And your flats that touch the target may rock unless everything is super flat.

It seems to me that a three legged laser pointer jig would be easier to make. Find a piece of sheet metal, cut it into an equilateral triangle and bend the three corners down to make legs. Don't make a little sissy tool. Go big! 18 -24 inches on a side and bend down about 6 inches for the legs. Fix the laser pointer in center position in this three legged jig. Get it pretty straight and lock it in place. Putty epoxy might be good for this. If your laser has a threaded housing you could nut it to the sheemetal in a hole drilled in the center. Just make sure its secure.

To calibrate, keep rotating that three legged jig on a flat surface and adjust the leg length by filing or grinding the legs until you acheive plane by watching the circle the laser draws on the far wall as you rotate. A laser that shoots out both ends 180 degrees opposed would be nice for centering while calibrating and later in use.

Got a machine shop handy? Make a plate with a hole for your laser and three drilled and tapped holes for threaded legs and you have -- an adjustable three legged jig. Imagine a 3/4 inch aluminum plate about 12" in diameter with nice engine turning on both sides. Knurled bronzed lock nuts and knobs on the legs. Oh yeah - cut some lightening holes in a nice pattern like a bicycle chain ring. Engrave your name on it in a nice script.

A pinhole cap for your lens is the next thing to make. Screw it on and line up the dot of light in the center of the screen without looking directly into the laser. Probably fairly easy on a view camera and damned tough on a DSLR.
Or, place a mirror over the lens and line up the reflection with the laser at the target.

Give up and order a Zig Align.
; >)

[Jonathan Brewer]

The tool specifically created to do this kind of work is the animation stand, they can get very involved in complexity and price, although I believe you can track down one of the simpler models, you probably want an older style animation style stand, used.

Google 'oxberry animation stand' which will give you an idea of what they look like, they were designed for 'old style animation'.

[Jonathan Brewer]

Actually I mis-spoke, more to the point, the animation stand was specifically created to keep things in register and the film plane paralell to whatever you're shooting from image to image, which is why it was the 'bread and butter' tool for the old style 'full' animation technique.

[Jonathan Brewer]

OMT....................they are also excellent for photographing flat art.

[Jonathan Brewer]

Speaking of full animation, if this is before your time, get the Max Fleisher 'Superman' classics from the 1930's, Art deco and high energy fun.

[Leonard Evens]

The mirror method seems the simplest and most practical, but there is one potential problem. Your pointer towards the lens, no matter how it is arranged has to be exactly flat on the subject plane. That means it should have a large base of contact. Otherise slight errors could be magnified. Thus carpenters don't rely on T-squares when checking such matters for large surfaces. They use various simple tricks from geometry.

Here are some other things you might do to get around that problem.

If it is feasible, put a large rectangle centered in the field of view on the subject plane. Then measure the distances to the corners from the center of the lens. These should be very close to equal. (If you want me to, I can calculate how much the difference in lengths translates to in terms of a displacement from parellism.) A related method which could be used to check the miror or laser pointer method would be to measure the distance from the center of the subject to the lens along what you've established as a right angle, measure the distance in the plane to some reference point, and then measure the distance from that reference point to the center of the lens. The Pythagorean Theorem tells us that the square of the diagonal measurement should be equal to the sum of the squares of the other two measurements. Carpenters use this method by choosing dimensions so the triangle should be a 3. 4. 5 right triangle.

If the subject plane is vertical, once you have the rear standard leveled, you can also check the lengths of diagonals of the image of such a rectangle in the subject plane. If these are essentially indistinguishable using a scale in mm, and the rectangle comes close to filling the frame, it would seem to me that you would be close enough for all practical purposes. If the subject plane is too far away, you could tape a tape measure or other ruler to it in the center and in the corners and measure the lengths on the gg of equal lengths in the subject. These should be very close to equal.

Another related thing to check is that your standards are as close to parallel as possible. There is often some play in the default positions of the standards, and they can be slightly out of parallel. The easiest way to do this is to turn the tripod head so the camera is pointing vertically downward, using a level on the rail as a guide to see if it is plumb. Then check to see how close the standards are to horizontal with a level. Check on the gg for the rear level and on the front of the lens barrel for the front standard. For wide angle lenses, small displacements from parallelism of the standards can tilt the exact subject plane enough to disturb focusing in the corners or edges.

[Ralph Barker]

As one who is also a maker of sawdust, I'd lean toward Leonard's measurement approach. Use a plumb bob to ensure both the target and the standards are vertical, and then careful corner-to-corner measurements to check alignment. The key issue, I think, is not to assume the wall is plumb, nor that the floor is square to the wall.

[Jonathan Brewer]

I'm a woodworker also, but I think the suggestions posted here 'pale' in comparison as to how these problems have already been addressed by the animation stand, reference the history of the Acme animation stand, and their multi-plane animation stand which was used to produce 'Snow White'.

In terms of all of the calculations and measurements that have been suggested by the folks above, there's always the potential to be off trying to figure this out for the first time, where you have folks who produced tools like the Acme animation stand, who were professionals at this from way back.

You may have trouble finding a good animation stand, and it might not be a cost effective proposition, but in terms of what your proposing, particularly taking cameras on and off, which implies repeatability, the animation stand would be the most precise tool for the job.

It'll be your first time trying this, as opposed to getting a tool produced by folks who have several decades to perfect a tool to get this right, good luck, I'm going to the beach.

[Paul Fitzgerald]

QT,

Do you have a mirror and a flashlight?

Put the mirror at your test target and the flashlight to the view finder and adjust the camera to set-up a 'hall of mirrors'. If you have a grid or micro-prism, you can set exact focus.

For a view camera have a mirror cut to fit in place of the film holder. Scribe an X from corner to corner thru the backing of the mirror. Remove the lens board, insert the mirror and hold the flashlight to the gg. and set-up the 'hall of mirrors'. Lock down the back and install the lens and set-up the 'hall of mirrors' again. If you have a grid on the gg. you can set exact focus.

You can get 12" mirror tiles at Home Base, Home Depot, Costco, ect. You DO NOT need front surface mirrors for this.

Fast, easy and cheap.

[robc]

I like the sound of mike Gudzinowicz method since it works for any camera where you compose through the lens. But will it work for a range finder? Possibly not due to paralax error.
One point though, with a view camera, if the lens is not perpendicular to the gg then all bets are off. So before lining up the resolution target align your lens / gg accurately. I'd be interested to hear how this can be done simply. Possibly place a mirror on lens barrel and laser on GG with lens open to check reflection is same point as laser exit. Probably need to focus so that returning point is sharp which should be at exact focal length of lens. For fixed back cameras then I guess you have to trust that lens is perpendicular to back.

[Leonard Evens]

"One point though, with a view camera, if the lens is not perpendicular to the gg then all bets are off."

I think the method I described above for ensuring the standards are parallel suffices for most practical purposes. I don't think most view cameras are machined precisely enough nor have sufficiently fine controls to justify the use of more elaborate measures. QT wants to do this in order to make careeful MTF measurements. My guess is that he would be better off repeating his setup procedure, including making sure the standards are parallel, several times, and then averaging the measurements for each position in the field. That way he would use statistical theory to compensate for the inevitable random errors. Any systematic erros would show themselves in differences in the four corners. Of course that would assume that the lens is symmetric to within tolerances smaller than anything he might measure, but I think that is a safe assumption, at least with a modern lens which has not been abused. Lack of symmetry in the lens would likely produce other rather obvious degradation of the lens.

[robc]

Sorry Leonard, I hadn't read your post fully.

[Henry Ambrose]

What I did was to swing a tape from the lens edge to the wall on a diagonal about 25 feet away then swing the tape off in the other diagonal and mark that. A string won't cut it because it stretches - wire or a steel tape will do the trick. Measure carefully to find the center between these two marks. This mark is also the center of your target. From there plumb bob down the wall to transfer to the floor, plumb bob down from lens to floor for your second mark. Snap your chalk line there and coat it with clear coat to keep it from getting erased. You now have a line that puts you in rough position for moving closer or farther when different focal lengths and formats. Just plumb your lens to the chalk line each time you begin for a close start. It makes no allowance for plumbness of the wall. I "measured" that by noting that the plumb line was parallel to the wall. A water level might be good for getting same height of camera and target but still no promise of planarity.

What I got from this was close but not quite there. I could stiil get good focus in one corner most of the time but not all corners. I could compare several frames and interpolate (swag) the compound results but nothing that would allow a definitive comparison between cameras. Not good enough to prove or publish but good enough that I could make a decision about which was better and which might not be so good.

So in my first post I suggested the laser with mirrors to attempt to increase the precision. There will be plenty of error in reading your plumb line, tape and levels. It would never show up if you were building a wall but it will in your test target film.

Leonard:

Would you do some math on the resolution of the numbers of the triangle for finding center and perpendicularity? The two legs were about 20-25 feet measuring out from the camera to the subject wall. What does that translate to given some small errors in measuring? Perhaps +_ 1/16th inch for a sloppy measurer?

[robc]

it now occurs to me that if Mike Gudzinowicz method is used then, for view cameras, by first aligning the target as described and then placing the laser pointer on the ground glass center with the lens open, the reflection from the mirror on the mtf target should show on the gg. This should show the minutest deviation of the gg not being at 90 deg to the lens. It does of course require that the laser pointer has a flat barrel head which can be placed on the gg and that the laser is perpendicular to the laser barrel end. Now how do we test that?

I am only advocating this method since if a laser pointer is already owned then there is no additional cost or construction involved.

[Darin Cozine]

It seems that using a mirror on the test target is the best blend of affordability and accuracy.
There were some suggestions of using an animation stand, but I would think they are too small for testing LF lenses.

So that will ensure that camera is in the correct position and facing the center of the target.

However the film plane and the lens plane still need to be parralel to each other and to the target if you are going to be testing edge sharpness. This complicates the matter.

Assuming the plane of focus was perfectly flat, and you focused on the center point. If your target had many radial lines, then theoretically one of the lines will be in sharp focus and the others will be out of focus (to a varying degree). This should make it obvious which adjustments to make so that the entire target is in proper focus.

One note on lens testing though. Lenses that have a slight curvature of field will appear much worse on edge numbers than other lenses. However when photographing 3d objects, the curvature of the plane of focus may actually match the object better than a perfectly flat plane. Also, many photographs have a focal point in the center area of the image. Thus I have allways interpreted lens test by weighting the center numbers more than the edge numbers.

*disclaimer.. I am an amature photographer with about 5 years experience, and in no way an expert*

-Darin

[Jonathan Brewer]

'There were some suggestions of using an animation stand, but I would think they are too small for testing LF lenses.'........................................................no way, the biggest of these were humongous tanks, that could hold motion picture cameras in perfect register.

[Darin Cozine]

Any of them 12 feet long or larger?
Most lens testing is at 10x enlargements.. so a normal lens on an 8x10 is 12 inches.. x 10 is 12 feet.

[Jonathan Brewer]

Who said it had to be? Not the guy who started this thread, you're the only one who's brought up 12 feet, or that 'most' testing is at 10x enlargements, which wasn't part of the original question, and he didn't say what kind of lens he was testing either, ...................................................here's the original question,............................... 'For the purposes of lens testing, I need to photograph a flat target so that the film plane and the target are strictly parallel. I have found that for critical MTF measurements, even small deviations affect the results.'

And that's exactly what you do with an animation stand, photograph a flat target registered on a 'cell', with the film plane and 'flat art' paralell to each other, if it's a matter of what you said later on, after he asked the original question, instead of what he said, then it's a different proposition,

[Leonard Evens]

Henry,

In response to your question "Would you do some math on the resolution of the numbers of the triangle for finding center and perpendicularity?", I'm afraid I don't quite understand what you are doing. Perhaps I'm missing something, but I can't visualize how you can arrange two opposite diagonals which are lined up precisely enough so the line between their endpoints can be bisected to obtain the center point. I can calculate angular errors, but I have to have a fairly precise specification. For example, if you know two sides of a triangle are precisely equal, and you know the altitude and base reasonably well, I can tell you how far off from perpendicular a line to a presumed midpoint would be depending on the relative error in the placement of the midpoint. Suppose the altitutde is about 20 feet, and the base is about 10 feet, then the presumed right angle from the presumed midpoint would be off by about one half the error in the placement of the midpoint relative to the base. The anglular error here would be in radians. So if the base were 10 feet, and the midpoint off by 1/16th of an inch, the relative error would be (1/16)/120 = .000520833. Half of that would be .000260417 radians or about .015 degrees, which is a little less than one minute of arc. View camera standards and tripods certainly can't be set to that degree of accuracy, so this error would be small compared to those arising from the setup of the camera. Also, even if you assume the camera were set up absolutely perfectly and that the subject plane departed from perpendicularity by such an angle, then the departure of the subject plane from the exact focus plane some tens of feet from the center of the field would be a small fraction of an inch. Since the relative error in the position of the image plane is the magnification times the relative error in the image plane, this is much smaller than the invevitable focusing/film placement errors.

As I noted previously, instead of trying to be more precise than the equipment allows for, one should use repeated measurements each done after stanrting from scratch, and then average. The hope is that the setup errors, which are beyond our ability to measure effectively, are random and will pretty much cancel out in the average.

[Mark Woods]

This animation stand suggestion isn't very pratical. For one thing they're generally fairly large. You'd need a truck to move one. They're not cheap, unless you find one being sold for scrap by a company that's changing it's business model. And if you did get one, you still have to mount the camera correctly on the head! The cameras used have pins in them to insure that they don't twist and the film plane is parallel to the art work/table.

BTW, Snow White was shot on a custom built stand that Disney fabricated. You can see it in the lobby of the building where the library is on the Disney Burbank campus. It's about 12' tall and has about a dozen trays that can move for more realistic portrayal of fantasy land.

QT, why don't you just take the lens to an existing MTF machine and do your tests?

MW

[Jonathan Brewer]

BTW, Snow White was shot on a custom built stand that Disney fabricated............................That's not quite correct, Disney collaborated together with Acme to produce the multi-plane animation stand that Snow White was shot on, which premiered Dec 21,1937.

[Jonathan Brewer]

In fact in 1924, Adolf Furer bought the Acme tool and manufactuing, a tool and die shop on San Fernando RD. In 1928, Walt and Roy Disney contracted w/Acme to modify a Bell & Howell for stop motion, in 1936 they collaborate on the above multi-plane animation stand.

[Ralph Barker]

Hmmm, that's very interesting. All this time, I thought only Wiley Coyote bought from Acme. ;-)

[Paul Fitzgerald]

Good morning,

"Any of them 12 feet long or larger? Most lens testing is at 10x enlargements.. so a normal lens on an 8x10 is 12 inches.. x 10 is 12 feet."

--darinwc , do you work for Microsoft ;-) That would be 10 feet, OOPs.

"One point though, with a view camera, if the lens is not perpendicular to the gg then all bets are off. So before lining up the resolution target align your lens / gg accurately. I'd be interested to hear how this can be done simply. "

--rob, place 1 mirror at the film plane that has been scribed thru with any pattern and light it from behind thru the gg. Place another mirror flat to the lens with a center hole scribed thru the mirror plating to the plain glass. Now look thru from the lens, it will look like sighting down a pipe or a 'Slinky' until it's dead-on. It's easier to do than describe.

"The mirror method seems the simplest and most practical, but there is one potential problem. Your pointer towards the lens, no matter how it is arranged has to be exactly flat on the subject plane. That means it should have a large base of contact. Otherise slight errors could be magnified. "

--Leonard Evens, by back-lighting thru the gg. the light is diffused, no pointer to worry about. Standing at the camera you can see when everything is exactly aligned in the target mirror. When the pattern is aligned, it really lights up. By removing the lens board you can align the film back and lock it down (watch it move). Then install the lens and align and focus the lens the same way and lock that down (watch it move). Move the focus adjustments and watch it move all over the place, which is why you align it when it's focused.

None of this is hard, it's exactly the same as aligning an enlarger

[Henry Ambrose]

Sorry Leonard, I knew exactly what I meant! <grin>

Maybe this will explain:

Start with a camera position facing a large, flat (we hope) wall. From the camera pull a tape from the center of the lens to a point on the wall well to the left, mark the wall with a pencil and then pull the same tape distance off to the right of the camera to the wall to obtain an equal distance measurement to the first. Measure between the two marks on the wall and mark the middle of that distance. That mark is the center line and when extended out to the camera is perpendicular to the wall.

Here is the triangle:
Corner A is the camera position, B is the left mark on the wall, C is the right mark on the wall, D (where the test target is located) is the midpoint between B&C.
Lets say: AB=25 feet, BC=25 feet, CA= 25 feet.

If I mismeasured by 1/16 of and inch between AB and CA, how much angular error is there in terms of the camera film plane and the test target? Or perhaps another question is: If the target is 2 feet square how much error will show up at one or both edges?

I hope this makes sense.
I can probably do this myself but I think you will have additional insight that will add value to the thread.

[Leonard Evens]

Henry,

I think I see what you are saying. The mark you make will lie on a line which will end up in the center of the field, but it won't usually be itself in the geometric center of the field. Assuming the placement of the two marks B and C is absolutely precise, it is simple to do the calculation and for an error of 1/6 inch in 25 feet, I get about 5 seconds error in the angle. But since the angle need not be horizontal, the deviation of the line on the ground from perpendicularity would be less. As I noted this ignores any errors in the placement of the marks B and C. I could analyze that also, but my guess is that it wouldn't materially change the result. The deviation from perpendicularity would be negligible.