TODO.txt

[  ] window with preview (scaled blit of buffer)

[OK] command line params

[  ] separate passes (normals, depth, AO)
   [  ] pre-pass with 1 ray and depth=1. 
   [  ] save to separate files (with suffix _n, _z, _ao)
   [  ] linear z? log z?
   [  ] exr with floats?

[OK] render one whole image x samples-per-pixels.
  [  ] accumulate and save (another thread?) between each sample (or group of samples? 8?)
  [OK] each thread renders one whole image instead of rendering tiles.
  [OK] subdivide the number of samples by the number of threads available.
  [OK] option to output different images instead of writing in the same?
  [  ] keep a full accumulation buffer to do the average at the end for a better result. The previous outputs are previews.
  [  ] time estimation using the time of the first image rendered x nb_samples / nb_threads

[  ] montecarlo
[  ] microfacet material.
[  ] normal, spec, ... textures.
[  ] triangle geometry: 
  [  ] obj reader?
  [  ] triangle mesh aabb.
  [  ] ray/triangle intersect.


REF:

Robust BVH Ray Traversal................http://jcgt.org/published/0002/02/02/
Watertight Ray/Triangle Intersection....http://jcgt.org/published/0002/01/05/
SSE RayBox Intersection Test............http://www.flipcode.com/archives/SSE_RayBox_Intersection_Test.shtml
nanort..................................https://github.com/lighttransport/nanort
Embree..................................https://embree.github.io/
FastNoise-SIMD..........................https://github.com/Auburns/FastNoiseSIMD