Book 3: Not actually sampling from uniform PDF

[spkersten]

This listing in the section about "Using a Uniform PDF Instead of a Perfect Match":

raytracing.github.io/books/RayTracingTheRestOfYourLife.html

Lines 1631 to 1656 in 62ce2e6

	class lambertian : public material {
	public:
	...
	bool scatter(
	const ray& r_in, const hit_record& rec, color& alb, ray& scattered, double& pdf
	) const override {
	auto scatter_direction = rec.normal + random_unit_vector();
	// Catch degenerate scatter direction
	if (scatter_direction.near_zero())
	scatter_direction = rec.normal;
	scattered = ray(rec.p, unit_vector(scatter_direction), r_in.time());
	alb = albedo->value(rec.u, rec.v, rec.p);
	~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ C++ highlight
	pdf = 0.5 / pi;
	~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ C++
	return true;
	}
	double scattering_pdf(const ray& r_in, const hit_record& rec, const ray& scattered) const {
	~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ C++ highlight
	return 0.5 / pi;
	~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ C++
	}
	~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

doesn't seem to actually sample using a uniform PDF. It still samples from rec.normal + random_unit_vector(), so cosine-weighted. The PDF value, pdf = 0.5 / pi;, is set as-if it was sampled uniformly. The final image looks good, because the error is cancelled by a wrong scattering_pdf of 0.5 / pi. That should remain cosine/pi, irrespective of which sampling PDF is chosen.

[spkersten]

Off topic: Since I'm only reporting things that I think are wrong, which feels rather negative, I want to say that I'm really enjoying these books. This is one of the nicest programming projects I've done so far.

[trevordblack]

Nah, you're good. I need to go through and do a final check on book 3. I'm just procrastinating as I know that it's still quite rough

[trevordblack]

I merged your PR before realizing that I shouldn't have

[trevordblack]

It looks like your code is right, but I might be making infrastructure changes.

The mistake that your code is fixing should never have occurred, so I need to do some code refactoring to simplify.