tatest wrote:
I think I see a lot of barrel distortion in that photo, making it difficult to draw any conclusions about what is level. That's pretty typical for the low-cost zoom lenses in digital cameras, to go from severe barrel distortion at one end of the zoom to severe pincushion distortion at the other. Some manufacturers try to correct this in post-processing of the digital image, looks like your camera does not.
With this much distortion, I think it would be pretty difficult to say much about what is level based on pixel coordinates. Don't pick just two points, follow the edge of the building roof all the way across. I think the change in Y coordinates as you go across is going to show you the roof as bow-shaped, opening downwards. Any two points, even if the camera is level, Y is going to depend on where you are on the distorted curve.
I agree on everything you have posted. The way I get around these issues in the intended application (wheel camber) is to:
1. Center the camera lens horizontally in line with the near vertical camber angle I wish to capture.
2. Position the camera lens vertically on a stand (which is parallel to the ground) in line with the wheel axel.
3. Position the camera about 5-6 feet from the wheel.
4. Restrict the area of the Win Paint x,y coordinates to the center of the picture, in other words avoid the extremities.
All my the fender shots show every distortion you have pointed out. The saving grace is they need to be only within +/- 2 degrees or so to still be useful.
My mall shot was meant to illustrate the relative reliability of measuring the center only. Naturally, I am assuming the mall is level and the 0.11 Degrees is my error in interpreting the bubble level.
At any rate, my target measurement estimation is only meant to provide a GO (less then -4.0) or NOGO (greater then +4.0) trip to a professional alignment service
Again, I do appreciate your input here, please respond if you think I could improve on this technique.
