Easy to compile with VS2017 using the VS2013 project provided and slight changes (adding preprocessor directive _SILENCE_STDEXT_HASH_DEPRECATION_WARNINGS ). However, it is only a partial reference implementation of the paper. While the paper sells speed over concurrent approaches, the reference monothreaded CPU implementation does not allow to reproduce the timings given in the paper. The implementation also does not reconstruct the surfaces (only the point projection is implemented) so the surfaces shown in the paper cannot be reproduced from this implementation. The implementation is also missing the normal projection step.
If you want to contribute with another review, please follow these instructions.
Please consider to cut/paste/edit the raw JSON data attached to this paper.