In continuing to work with MadMapper in the dome, we loaded up a dome master calibration grid. Last test, we used images with MapMapper and had good results. We decided use the grid to achieve more accuracy.
We first placed the grid on the dome using DigitalSky 2 software so we could see what it should look like. Then we turned on each projector individually and took images of the grid by projector. This allowed us to have an image of what part of the grid each projector is projecting.
We loaded the grid onto MadMapper and began with the zenith, which proved to be fairly easy. Our reference images where very helpful in the process. Next, we attempted to map the grid on the front projector. The results were excruciatingly inaccurate. The lines warped and bent and got fatter and thinner as we manually adjusted the grid to fit the reference picture.
MadMapper seems to works best with surfaces that have hard edges which the spherical dome simply does not have. We spend hours just trying to get a decent image and at the end of the day we left feeling frustrated, but understanding just how complex the task actual is.
Given the difficulty of calibrating the dome projections using MadMapper, we will be focusing the majority of our efforts on DomeGL. (Although Ethan is still convinced he can correctly map MadMapper) We’ll be investigating the possibilities of combining the best aspects of both systems by creating a standalone DomeGL projection application with Syphon support. This would allow us to use DomeGL with a number of application frameworks such as Max/MSP/Jitter, openFrameworks, Processing, Quartz Composer, and Unity3D.