Experimentally measuring FOV for DIY Rifts

[xef6]

With all the DIY HMD stuff floating around here, I began wondering about how to determine the specs for a given device. Does anyone know how to check how wide/tall the FOV is?

My only idea would be to show a grid on one eye of the HMD, and look at a real object with the other eye. By comparing the size of the grid in the HMD to how large a measurable object appears, you should be able to estimate FOV, right? FOV as in http://en.wikipedia.org/wiki/Visual_angle. Given the distance from the eye to the object, and how wide/tall the object is, it should be possible to estimate, right?

Any input is welcome

[EdZ]

I can't think of any reliable empirical method other than a calibrated wide-FoV camera placed where the eye would be.

[brantlew]

I was thinking about a camera too, but how do you know you are not just measuring the FOV of the camera? I think the original idea of an overlay grid is pretty good? Draw a grid on a flat wall in the real world. Then render a grid in your HMD, but include simple controls so you can stretch the grid interactively. Keep putting the HMD on and off, comparing the real grid with the virtual grid and making adjustments until they look the same. Now you can just measure the real world coordinates of the visible grid and your position and estimate the FOV. Or something like that?

I'm sure Palmer, TheLostBrain, or some others knows the best way to do this.

Edit: A similar but maybe logistically simpler method might be to get a projector and display the same image on the wall and in your HMD - making adjustments until they are the same scale, then do your measurements.

Edit2: Just a thought. I guess instead of squashing the image in software you could just physically walk forward/backward in the room until the projected images matched.

[EdZ]

I was thinking about a camera too, but how do you know you are not just measuring the FOV of the camera?

Hence the 'calibrated' bit. The maximum FoV you could measure would the the maximum FoV you could image: a circular fisheye setup should be able to measure up to 180° in both axes.

I suppose if you replaced the focal plane sensor with a simple projection surface, and told the HMD just to show a full field of white, you could draw a graticule on the projection surface and measure the projected shape (and thus the FoV) without an expensive sensor. You may need an additional magnifying lens to see the graticule depending on the focal length of your fisheye lens though, and the reading would hardly be automatic and repeatable.