Improving a Camera Calibration

A calibration is an important step when using PhotoModeler or PhotoModeler Scanner, especially when looking for maximum accuracy.

A calibration usually is fully automated and requires no intervention, but if you’re looking for maximum accuracy, and the benefits of improving the calibration would outweigh the benefits of optimizing ALL project factors (as shown here:, this article describes a few tips to take a calibration from OK to really good.

So, if all other project/accuracy factors are optimized (ie number of points, photos, angles etc), then you might want to try to redo and improve the calibration, bearing the following in mind:

– Make sure the calibration sheet is firmly attached to floor using tape. Better yet get the calibration grid mounted onto a very rigid board. This prevents any warp – ie. if the sheet or part of the sheet moves or warps, even the slightest amount, the calibration can be thrown off a bit. On a very high resolution camera, a slight draft as you’re walking around it could bubble up the grid in places, and even a hair width movement can account for 1+ residual value.

– Turn off/down the camera’s ‘sharpening’ setting if it has it.

– Turn off the camera’s image stabilization setting if it has it.

– If your targets have a bit of colour ‘fringing’ around some of the targets, you might want to try the camera’s black and white mode if it has a good one

– Some lenses (eg Canon) have an internal focus ring so it can’t be totally fixed, but for others try to fix the focus ring during calibration with thick rubber bands to fully stabilize. Even at a manual focus, the lens can get jogged a bit and lens elements may move a bit. This one is not really solvable if the lens has an internal focus mechanism, but to minimize the lens shift due to focus, find the appropriate focus using auto focus with your camera in position (best is tripod), then turn auto-focus off, stabilize the lens and take all photos.

In summary, there are two main causes of calibration problems: a) the grid changes between photos (eg not fixed to surface); or b) something about the camera changes between photos (eg zoom or focus).

Many of these issues are subtle, but can have a detrimental effect on a calibration.

Also, if in your eventual project you’ll intend to use Field Calibration, you may not need to fully optimize a calibration since the calibration may be a good enough starting point and the Field Calibration process will fine tune and improve the solution. Information on Field calibration can be found here:


©Copyright 2016 Eos Systems Inc.