I places a series of orange spheres on the rock and then photographed the site from 5 different angles using a calibrated camera. The markers were located in the reference shots and then processed using the Camera info into 3-space and then exported as a mesh to Rhino.
Modeling
Using the imported mesh (dark green) and some measures from the building I rebuild the site in Rhino. The platform itself was a half day modeling job but I extended it with some scripts to calculate quantities. Here I would love to experiment with parametric and generative models that take points as inputs to adapt structures. Something for Catia or GenrativeComponents.
Completion
It took me about a week part time to build the platform. After completion I measured the result and compared it to the original measures and found an error-margin of between 4 and 7%. This was a bit much but is probably due to the crappy quality of the camera I used and the lack of resolution in tagging the markers. The technology is really promising, especially if one could use unsupervised tracking with video as input. It can allow for much greater adaptation of pre-manufactured buildings to complex sites with a minimal macro-environmental impact.
