Lens distortion changes as a lens pulls from FocusMin to INF.
Vignetting is similarly dynamic, tho more subtle.
To create a lens model, you need distortion maps for a whole lot of witness marks btw min/INF. Traditionally this requires a stack of large 32bit exr images. A lot of info to work with interactively.
Here are a few complete sets to kick around for your own RnD:
An app like Unreal is all about interactivity.
nstead of requesting data from a stack of heavy images, they can be converted to raw distortion values, which are much faster to access. When converted to roughly a dozen float values for SPH (bit more for ANA), less data needs to be swapped on the fly, resulting in a much more interactive tool.
In Unreal you can find it within a Unreal Lens Asset