2022-05-11: APP 2.0.0-beta 1 has been released !!!
Download links per platform:
LNC Order vs Iterations tradeoff
What is the tradeoff between using a higher order of LNC with a low number of iterations vs a lower order LNC with higher iterations? When would you choose more iterations or a higher order?
Thanks anyone for input!
The two parameters speak to two different aspects of LNC:
Degree determine the "shape" of the correction: 1st degree means it can only correct a linear gradient (bright at one end, dark at the other), whereas 2nd or higher degree can fit a curve on the gradient. (a "standard" vignette can probably be estimated well by a 2nd degree). Changing the degree has (basically) no impact on the performance/speed, as it's only determining "how do I correct the gradient I'm detecting".
The iteration # determines how many time you run the computation, by fitting a first correction (iteration 1), then looking at whether that can be improved (iteration 2) and on and on. That has a direct impact on the performance.
--> The degree will determine the complexity of the correction you can compute, the iterations how well that correction is fitted.