Matt Pharr

I believe that changing the camera (and updating the film's output filename) would suffice, but haven't tried it myself. What do you mean "could not make it work"? What happened...

This is basically due to a bad design decision (that will be fixed in pbrt-v4). Most lights in pbrt-v3 take multiple parameters to specify their emission: e.g. the diffuse area...

Hm, I'm not sure about that blue by the pillows. My best advice would be to render just one of those pixels and add debugging print statements that describe the...

You are correct; pbrt currently incorrectly flips S in that case. :-( Unfortunately, fixing this makes all bump maps go in the opposite direction (down rather than up and vice...

Hm, we do check the second quadratic solution here: https://github.com/mmp/pbrt-v3/blob/master/src/shapes/hyperboloid.cpp#L100, in the case where the first one is at t < 0, and then here: https://github.com/mmp/pbrt-v3/blob/master/src/shapes/hyperboloid.cpp#L116 if the intersection point...

The second of these no longer causes a crash, presumably due to the parser rewrite, which doesn't have some of the fixed-size arrays that the old parser did.

Nice find--thank you! Now fixed.

This breaks the Triangle.Reintersect test for reasons that are no immediately obvious. Pushed a95b0bc97f79785dbbf2576019d730084677b492. Reopening for now.

Now it's just the BSDFSampling tests that fail.

One issue is that the variant of Quadratic() for EFloats computes the discriminant using double precision, completely ignoring the inbound error bounds with the A, B, and C parameters. If...