Sticky: Recent Updates and Important Information

Sticky: Recent Updates and Important Information

Monday, January 21, 2013

Editorial: Parallax Correction

A couple of comments on the Kickstarter comments page asked about parallax correction, which is an issue characteristic of all TLR cameras. There are a number of ways to correct for parallax error - moving the entire viewfinder image, Rolleiflex-style, a moving needle to note the top of the frame, a la Mamiya, and finally fixed correction markings in the viewfinder. To minimize complexity, Duo will use the last option. 

But the problem still remains of how to calculate the error itself. We could of course measure it empirically by shifting the camera up and down by the lens separation of 75mm, or by comparing the VF image to the one projected onto ground glass in a roll film back. Either would work, but why not use some math to characterize the parallax error throughout the entire focusing range?

First we need to know the angle of view of the lens we're using (it'll help us calculate the subject size, and compare it to what's seen in the viewfinder). Mamiya literature says 41 degrees, 20 minutes for the 105mm Sekor, but of course this is on a 6x6 frame (56mm, nominally). Usually angle of view is measured diagonally across the frame, which by itself is not really useful. But, we can find the vertical and horizontal angle of view using a little bit of grade school trigonometry. We find it to be 30.6 degrees.

We know that the image projected in the viewfinder and onto the film should differ by the lens separation, 75mm. However, we need to know what this error looks like when the image is projected onto the ground glass. It seems safe to assume that the error in the viewfinder and on the subject itself will be proportional to the image dimension (i.e. the height of the viewfinder image, assumed to be 56mm, and the height of the true image - the subject - let's call that "a"). 

Again, using some trigonometry, we can form a relationship between the distance from the camera and the size of the subject. From that, we can relate the subject size to the parallax error in the viewfinder (delta v).

 And that's that - mostly painless. We find, as we expect, that the parallax error for subjects relatively far away is minimal, but becomes pronounced at around 1 or 2 meters from the camera. 

These numbers line up nicely with my experience, as well as some tests I performed by shifting the camera up and down by the lens separation. The three figures highlighted in yellow are probably the most useful for everyday photography, and are the ones I'll be including in the viewfinder. *Well, since the camera CAN focus down to about 1 foot, I'll add that parallax marking.

And here it is, the updated viewfinder cover (there are two pieces to the viewfinder - the ground glass that the image is projected on, and an acrylic protector with frameline engravings). It got a bit busier with 6x6, 6x7, ad 6x9 engravings. Parallax markings are shown on the right, in METERS. Now you may be thinking: "but that only helps if you're shooting Polaroids, and even then, only in portrait orientation!" 

You may have a point, though. But it would get extremely busy to add parallax markings to all SEVEN framelines. The parallax markings shown work best as reference. If you use your imagination, you can transpose those markings to the other framelines. 

1 comment: