sample count looks cool

adaptive sampling fixed the point light material

A roughness to glossiness scalar applied to make the result match similar to the sample image

Fixed UV issues, and using a custom 360 photo I took a few years ago as the skybox, to demonstrate how I UV align spheres in my code.

For some mysterious reason... Changing the UV scaling to half, for plane objects, creates some NaN color when rendering the point light which is rendered as an in-world sphere in my code. Part of my optimizations from before involved redoing how I was hanlding checks for my BVH to remove a division, which made the collision messy for flat boxes planes, so I redid the planar collision too to be simpler. Somewhere along the way the UV coords got messed up. But what does that have to do with the color outputted by the sphere?

Submitted late as I did a ton of optimization to speed things up, and redesigned lights to be able to project a texture as their light. Image shows soft shadows. I still have not implemented adaptive sampling, without it rough reflections and refractions take too long to render with my current code.