else its buggy.

actually use the blink detection o remove invalid pupil positions

save the eye capture settings in a pickle or npy. This makes life easier.

investigate if the grab and xmos grab fns could be threads instead of separate 
processes. 

After pattern is detected for calibration routine, the green dot remains in the 
'world' window.  This can be easily fixed.  

Possible tools to use:
 - gaussian difference 
 - RANSAC fitting
 - Kalman filtering for good initial guess for RANSAC

glumy uses the glut event loop. it does not exit nicely.

h and v blank should be computed automatically. 

Hello,

We are students in Health Safety Environment and for our final project year we 
must make a eye-tracking. But we encounter a problem: How can photographic film 
behind the camera lens?

Thank you for your answers,

Regards,

Stacy.

if a camera exposes 0 controls. we get an error.

RANSAC for calibration fitting.

using scipys ransack methods.

this should be easy.