Matt Pharr

image differences with MIS compensation and approximate BSDF PDFs

2

With ``` Sampler "halton" "integer pixelsamples" 8 Film "rgb" "float iso" 200 "integer yresolution" [ 800 ] "integer xresolution" [ 1200 ] Scale -1 1 1 LookAt 3.69558 -3.46243 3...

1M emissive triangles use 9GB of GPU memory

4

(That's pushing 9kB per light source!) Repro: in pbrt-v4-scenes toplevel, render the file: ``` WorldBegin AreaLightSource "diffuse" Shape "plymesh" "string filename" "landscape/geometry/mesh_00110.ply" ```

Windows build improvements

4

There are a number of ways the Windows build could be improved: * Parallel builds would be nice * The organization of the source files in the MSVC project browser...

The option is currently ignored... (Via @Vutshi in #134)

Windows GPU performance much worse than linux

16

e.g. crown is 7.3s on a 3090 and 97.6s on a 2080 (though the 3090 is faster, it's not that much faster!). Looking at the `--stats` output, the issue seems...

pbrt --upgrade needs support for the Disney material

2

pbrt-v4 no longer includes the Disney material, but `pbrt --upgrade` doesn't do anything with it. We should try to do some sort of reasonable mapping to the new material models.

BDPT regression in pbrt-v4

2

With the pavilion-night example scene: pbrt-v3:  pbrt-v4: 

CPU/GPU image differences in zero-day scenes

1

A few things are coming out differently, for unclear reasons. e.g. the pipes in frame52.pbrt. CPU:  GPU: 

The version of ptex in src/ext is currently a few years old; however, updating to the latest version causes the Windows build to break due to something related to pkg-config.

Due to code that still needs to be written in `gpu/accel.cpp`, the GPU path does not currently support: * Curves (which could be done using OptiX's built-in curves) * Animation...