Graves

thingiverse

If they can't visit those famous graves, the graves can come to visit them.\nuploaded some point cloud files that others can use to generate higher resolution STL files if their computer is capable of doing so.\nit's good practice to clean up this file a bit more before reconstructing it.\ni plan to upload reconstructions when I get my hands on a machine with enough RAM to handle the task.\nedited : I tried to rebuild more and gained some lessons learned:\n1) close-up photos provide much better detail levels than many high-resolution overview photos.\n2) A 64-bit machine, based on DDR3 or DDR4, with 16 GB quad-channel or more RAM is a must-have.\nSome helpful links can be found at www.cs.jhu.edu/~bolitho/Research/StreamingSurfaceReconstruction/ and http://www.cs.jhu.edu/~misha/Code/PoissonRecon/Version6.11/, but the PVMS alone won't work due to requiring over 4 GB of RAM for feature matching.\ni haven't attempted any slicing of such scans yet...\nof course, those are over 2 meters high models, but I bet someone with a Delta printer will finally try making at least a 1:2 replica ;)\n3) PVMS does not set up normals, even if camera positions are valid. No idea why - perhaps it's a bug. One needs to reconstruct normals in MeshLab, but it's far from perfect and many artifacts kick in :/\n4) Poisson surface reconstruction in MeshLab is broken (infinite loops). Use one of the above or something completely different.\n\n \n \n Instructions for Not Printing