BxDF tests #165
BxDF tests #165
Conversation
…Examples-and-Tests into bxdf_unit_tests
…ng/Nabla-Examples-and-Tests into bxdf_unit_tests
|
why do you have random things from sorakrit here instead of your own branch against his or I literally can't see whats yours and what's @keptsecret's. |
| retval.T = nbl::hlsl::normalize<float32_t3>(rngFloat301(retval.rng)); | ||
|
|
||
| retval.T = nbl::hlsl::normalize<float32_t3>(retval.T - nbl::hlsl::dot<float32_t3>(retval.T, retval.N) * retval.N); // gram schmidt |
There was a problem hiding this comment.
you can still generate T almost colinear to N and end up with numerical issues, personally I'd use the frisvad basis function + random rotation around N (only one random number needed)
| retval.eta = 1.3;// rngFloat01(retval.rng) + 1.0; | ||
| retval.ior = float32_t3x2(1.02, 1.0, // randomize at some point? | ||
| 1.3, 2.0, | ||
| 1.02, 1.0); |
There was a problem hiding this comment.
you can only have one:
- eta for BSDFs (things that refract) because refraction can only be done monochrome (each wavelength refracts at different angle)
- ior for BRDFs (things that reflect) because reflection is alawys the same
Also eta is just a ratio of two refractive indices, and using matrix<float32_T,3,2> is very bad semantically because it assumes we'll have 3 spectral buckets.
Ideally we should do tests in monochrome, so just a float eta should be enough or vector<float,2> ior
There was a problem hiding this comment.
Do we want to keep ior as a matrix in the BxDF?
For the BRDFs https://github.com/Devsh-Graphics-Programming/Nabla/blob/6fc78a46f466d176f1b3a4c84f18f7448d3cd894/include/nbl/builtin/hlsl/bxdf/reflection.hlsl#L361
There was a problem hiding this comment.
no dont keep it was a super bad idea, I also want to make everything templated so that vector<T,3> isn't implied for the colours and we could do other number of spectral buckets (1, 2 or 4 for spectral rendering)
| SBxDFTestResources retval; | ||
|
|
||
| retval.rng = nbl::hlsl::Xoroshiro64Star::construct(seed); | ||
| retval.u = float32_t3(rngFloat01(retval.rng), rngFloat01(retval.rng), 0.0); |
There was a problem hiding this comment.
the third random number is sometimes necessary to discern discrete BxDF branches:
- reflection vs refraction in e.g. glass
- diffuse vs specular in e.g. plastic
- choosing a leaf BxDF in a mixture/tree of BxDFs
| ray_dir_info_t dV(int axis) | ||
| { | ||
| float32_t3 d = (float32_t3)0.0; | ||
| d[axis] += h; | ||
| ray_dir_info_t retval; | ||
| retval.direction = V.direction + d; | ||
| return retval; | ||
| } | ||
|
|
||
| float h = 0.001; |
There was a problem hiding this comment.
- what do you need
dVfor? - this is not the way to vary
Vparameter
| uint32_t pcg_hash(uint32_t v) | ||
| { | ||
| uint32_t state = v * 747796405u + 2891336453u; | ||
| uint32_t word = ((state >> ((state >> 28u) + 4u)) ^ state) * 277803737u; | ||
| return (word >> 22u) ^ word; | ||
| } |
There was a problem hiding this comment.
this is useful in its own header like xoroshiro is
…ng/Nabla-Examples-and-Tests into bxdf_unit_tests
…ng/Nabla-Examples-and-Tests into bxdf_unit_tests
…ng/Nabla-Examples-and-Tests into bxdf_unit_tests
No description provided.