Solstice Solver

Version 0.10

Replace CMake with a POSIX Makefile

The build procedure is now written in POSIX make and can be configured via the config.mk file. A pkg-config file is also provided to link the library as an external dependency.

Compared to the CMake alternative, this Makefile adds support for static libraries and an uninstall target. It also enables compiler and linker flags for various hardening features, improving the security and robustness of generated binaries. More broadly, the motivation for this rewrite is to rely on a well-established standard with a simple feature set, available on all UNIX systems - reducing portability concerns and maintenance burden while remaining significantly lighter.

Also raise the minimum required version of dependencies.

Version 0.9

Fix self-intersection on meshed mirrors.
Raise the minimum required Star-SampPling to 0.12. This version fixes compilation errors with gcc 11 but introduces API breaks.
Raise the minimum required Star-3D to 0.8.
Fix compilation warnings detected by gcc 11.

Version 0.8

Register into the estimator the final state of the RNG used during the simulation. Add the ssol_estimator_get_rng_state function that returns this state.

Version 0.7.3

Update the version of the RSys and StarSP dependencies.
Fix a compilation warning with GCC7 and above.

Version 0.7.2

Fix the gaussian sunshape.

Version 0.7.1

Fix the creation of a glossy BSDF that uses a pillbox microfacet distribution.

Version 0.7

Add the Gaussian sun shape.
Add the microfacet distribution parameter to the mirror material: one can choose either the Beckmann or the pillbox distribution.

Version 0.6.1

Rename the ssol_sun_pillbox_set_theta_max function in ssol_sun_pillbox_set_half_angle.

Version 0.6

Fix the integration for non parallel sun: the angle between the principal sun direction and the sampled direction was not correctly taken into account leading to a wrong initial weight for the optical paths.
Fix the integration with shapes having perturbed normals: perturbed normals must be taken into account in the bounces of the optical paths only, not in the energy computations.
Fix the distribution of the pillbox sun: the pdf was wrong.
Fix the ssol_sun_pillbox_aperture function and rename it to ssol_sun_pillbox_set_theta_max. The submitted parameter, i.e. theta_max, is the angular radius but was treated as the angular diameter.
Update the ssol_solve API: add a parameter that controls the number of realisations than can fail before an error occurs.

Version 0.5

Improve performance by up to 50% by optimizing the allocation of the BSDF along the optical paths. Performance gains are mainly observed in situations where the optical paths are deep, i.e. when they bounce on many surfaces.

Version 0.4.2

Energy conservation property might not be ensured when the optical paths were fully absorbed.
Handle infinite optical paths, i.e. paths that bounce infinitely due to the material properties and/or numerical inaccuracies. Use a Russian roulette to stop the optical random walk without bias.

Version 0.4.1

Fix a wrong "path inconsistency" check. The paths going from a dielectric to infinity were wrongly detected as inconsistent.

Version 0.4

Add the SSOL_PATH_ERROR type used for the paths that travel unforeseen mediums.
Fix the cosine factor estimation that did not take into account the shadowed realisations.
Ensure the energy conservation property for dielectric materials. Previously, some energy was lost even for dielectric materials with no absorption.

Version 0.3

Full rewrite of the estimated values. The global results report the cosine factor, and the overall flux that is: absorbed by the receivers, atmosphere, or other entities; occluded before it reaches a primary entity; missed because it does not reaches any surface. The per receiver results include the incoming/absorbed flux in 3 situations: all phenomenons are taken into account; the atmosphere is disabled; the material propagate the whole incoming flux, i.e. they absorbed nothing.
Update the ssol_solve API. Streamed binary outputs are removed.

Version 0.2.2

Fix the estimation of the cosine factor for the sampled instances: it was not correctly reported and was thus always equal to 0.

Version 0.2

Add normal maps to describe spatially varying normals in the tangent space of the surface.
Add support of spectral data to the atmosphere and the materials.
Fix the per primitive irradiance estimate by dividing the result by the area of the primitive in order to have watts per square meter.