EDIT: It's possible I should have just asked whether any digital video format included per-frame metadata. As asoundmove did here:
And like many others have asked elsewhere for some years now.
I've been asked to measure the surface area of ground floor windows on a kilometer-long street lined with stores. I'm also supposed to videotape the storefronts, as if from a (slowly) passing car, and superimpose the surface area of glass for each store. Maybe on the glass, using motion-tracking software.
Videotaping the stores from a car is not a problem (hood or roof mount, I guess), nor is "attaching" a number on the glass with software afterwards for each store on the way by. Likewise, measuring the glass with a tape measure is easy, but very time-consuming. I'd like to acquire the measurements from the videotape, even if I have to shoot once for a nice angle, and again for measuring, where the camera points straight at the glass.
And if it works, I'll be reproducing the work for much more than the one kilometer of storefronts, so efficiency is my goal.
EDIT: Actually, one of my first thoughts was that I could just read the distance off the focus ring of my old SLR, but this doesn't provide accuracy to within a meter, as far as I can read it.
Then I wondered why I'd never seen camera-to-subject distance in EXIF data. Turns out it was because I wasn't looking hard enough.
- Get the measurements from an individual frame for each store, using the pixel-to-meter ratio, but I'd have to also measure the distance from the camera to the glass each time. Not a really big nuisance with an inexpensive laser distance measuring device like the Bosch DLR130K. I'd need an assistant to measure and record the distance for each store while I was driving, or vice versa.
This method would definitely work, but it's only a short step, really, from using a tape measure on the window of every store.
Attach a meter-long stick to the edge of one window, photograph the storefront, walk to the next store, repeat. This is also like using a tape measure, but I my attempts have shown it to be more efficient to get the pixel-to-meter ratio from the stick and measure the rest in Photoshop.
Somehow figure out how to use OpenCV with my two GoPros for stereo triangulation. Like in this video, the source code for which has been removed.
Should it work, this approach has the advantage of getting video and distance simultaneously. Not to mention 3D, should there ever be any use for that.
Use any number of range imaging methods, none of which I actually see myself pulling off.
So now I'm hoping someone has dealt with this kind of thing before, and can show me the real way to do it! :)