Using HitRecord::t before declaring it

[Krenodeno]

In Chapter 6, at listing 13, hit_record definition is

struct hit_record {
    vec3 p;
    vec3 normal;
};

and then in listing 14, there is reference to a t member that do not exist... yet.

bool sphere::hit(const ray& r, double t_min, double t_max, hit_record& rec) const {
[...]
            rec.t = temp;
            rec.p = r.at(rec.t);
            rec.normal = (rec.p - center) / radius;

hit_record's member t is defined later as the time:
I know that we’ll also want motion blur at some point, so I’ll also add a time input variable.
Therefore I cannot see any relation to time in the calculations of temp listing 18:

auto a = r.direction().length_squared();
auto half_b = dot(oc, r.direction());
auto root = sqrt(discriminant);
auto temp = (-half_b - root)/a;

If I'm correct, at this point t is more like a depth buffer, as to see listing 19.

[trevordblack]

This is a bug. Thank you for catching it.

Unfortunately, rec.t refers not to the time, but the parameterization of the ray. Namely:

orig + t*dir.

t needs to be put back into the initial definition of hit_record. And we need to consider renaming it since it's overloading with the time variable later on

[trevordblack]

Just use

struct hit_record {
    vec3 p;
    vec3 normal;
    double t;
    bool front_face;

    inline void set_face_normal(const ray& r, const vec3& outward_normal) {
        front_face = dot(r.direction(), outward_normal) < 0;
        normal = front_face ? outward_normal :-outward_normal;
    }
};

for hit_record, where t is the parameterization of the ray (not time). Neglect the comment I know that we’ll also want motion blur at some point, so I’ll also add a time input variable. as it was a brain fart on my part.

There is a new PR to fix this issue.

Thank you for bringing it up

[Krenodeno]

Glad I could help !

[hollasch]

Resolved in PR #429

[junkimu]

t still needs to be added back to listing 18: "The hittable class with time and side"

[hollasch]

Thank you for catching this!

[hollasch]

It would be good to use a better name, perhaps ray_parameter.

[hollasch]

See also #418 regarding our nomenclature for time0 and time1. This should be considered as well when making these changes.

[hollasch]

Resolved in PR #625