Why deal with mesh files to do the warping, why not derive the warping tranformation explicitely?
A few reasons. Perhaps the main reason is that it’s very difficult (if not impossible) to derived a closed solution. While it is possible for an idealised situation, it becomes much more difficult when one has to deal with certain realities such as lens offset, rotated projector, possibly rotated mirror and so on.It also turns out that some components are not perfect and some are additionally difficult to measure. So even if a closed expression was found there would still be some trial and error as well as manual tweeking if one wanted an optimal result. Instead of loading each application with the equations (or the current simulation) and an interface for testing parameters, I decided to do that once and all subsequent applications just read the single calibration mesh that describes the warping.
A few reasons. Perhaps the main reason is that it’s very difficult (if not impossible) to derived a closed solution. While it is possible for an idealised situation, it becomes much more difficult when one has to deal with certain realities such as lens offset, rotated projector, possibly rotated mirror and so on. It also turns out that some components are not perfect and some are additionally difficult to measure. So even if a closed expression was found there would still be some trial and error as well as manual tweeking if one wanted an optimal result. Instead of loading each application with the equations (or the current simulation) and an interface for testing parameters, I decided to do that once and all subsequent applications just read the single calibration mesh that describes the warping.
