What do you think of this chart to measure Lens Distortion?

[Guillermo Luijk]

From great to crap, I just want to learn and have fun. I have drawn the following char to print and shoot to measure lens distortion:

http://guillermoluijk.com/lens/lens_distortion_chart.png

I start from two assumptions:
- The lens distortion has radial symmetry (so visible de-centred lenses would be an issue)
- It requires shooting precision to locate the end of the arrows (mainly the two arrows with the stripped blocks) exactly at the end of the frame (needs a 100% coverage viewfinder)

The idea is to measure the size all those blocks (I plan to use RAW's G data) get along the sensor diagonal after shooting which will be the distortion profile. Both hemi-diagonal values will be averaged to reduce measurement errors. Then undo the distortion profile and interpolate is with a smooth curve (spline) that can be applied to correct the distorted lens. Checking if it works will be easy by shooting any checkered or linear subject.

The inner circle serves to make shooting easier since a tiny piece of it will be visible on four corners. I think it cal help locate the arrows in the extreme corners before shooting.

Regards

[Guillermo Luijk]

BTW does anyone know if geometrical distortion usually changes with the lens focusing distance? I gues not, or minimally, otherwise lens profiles wouldn't correct properly all images. But never wondered.

Regards

[Guillermo Luijk]

The chart was a pain to use. It's very difficult to adjust the corners exactly at the arrow's ends.
Tested with a macro lens with (nearly) zero distortion, the only one of mine with enough magnification to fill the A4 print. I definitively need to print at A3.

http://guillermoluijk.com/misc/chart.jpg

http://guillermoluijk.com/misc/canonfd35mmmacro.png


New chart: this one just requires the user to centre the chart in the frame:

http://guillermoluijk.com/lens/lens_distortion_chart2.png

Regards

[Rhossydd]

I start from two assumptions:
- The lens distortion has radial symmetry (so visible de-centred lenses would be an issue)

There's your first problem. Almost all lenses suffer a degree of asymmetry in their performance and that will vary with aperture too. Zooms particularly move around a lot as focal length changes.

In practice not much of an issue for still use, but in other fields can become a proper problem.

[kers]

I measure the lens distortion with ptGui. As long as you have the nodal point rright, you can make a complete circle with overlapping images- or sphere; and Ptgui calculates the rest.