- 14 9月, 2021 9 次提交
-
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
This speeds up light BVH builds by ~20%.
-
由 Matt Pharr 提交于
The lights now all use an InternCache for their DenselySampledSpectrum values, saving a significant amount of memory for scenes with many area lights that have the same emission spectra. (Partial fix to issue #183.)
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
This way things like seg faults will no longer cause pbrt to just silently terminate, which has been fairly confusing behavior...
-
由 Matt Pharr 提交于
- Use a std::atomic for the bytesUsed statistic in the BufferCache (a legit bug, though harmless.) - Hold associated mutexes when consuming futures. Strictly speaking, all of the worker threads should be idle when these are consumed and it's only the main thread that's running, so there should be no need...
-
由 Matt Pharr 提交于
Fixes #179, via @shadeops.
-
- 13 9月, 2021 2 次提交
-
-
由 Jim Price 提交于
-
- 07 9月, 2021 1 次提交
-
-
由 Matt Pharr 提交于
Assorted renamings in the VolPathIntegrator implementation.
-
- 05 9月, 2021 1 次提交
-
-
由 Matt Pharr 提交于
-
- 03 9月, 2021 3 次提交
-
-
由 Matt Pharr 提交于
Partially seeding an RNG using just ray direction is risky business with an orthographic camera. Things go extra bad when the object enclosing the medium has camera-axis aligned planes in which case doing the rest of the RNG seeding with tMax to the intersection point rounds out the borkage. Fixes #175.
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
-
- 01 9月, 2021 6 次提交
-
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
Fixes #176.
-
由 Matt Pharr 提交于
s/BumpEvalContext/NormalBumpEvalContext/ Factor Bump() into BumpMap() and NormalMap()
-
- 31 8月, 2021 5 次提交
-
-
由 Matt Pharr 提交于
Switch to DiffuseBxDF in DiffuseMaterial Remove "sigma" parameters with --upgrade
-
由 Wenzel Jakob 提交于
-
由 Wenzel Jakob 提交于
-
由 Wenzel Jakob 提交于
- Adds chi^2 tests for various microfacet model configurations (conductive, dielectric, anisotropic, isotropic, dense->less dense and vice versa, etc.) - Increased the directional resolution of the chi^2 test
-
由 Wenzel Jakob 提交于
This commit fixes issues in the computation of Jacobian of the rough dielectric half-direction mapping and the criterion for discarding back-facing microfacet normals with respect to both the incident and outgoing direction argument. These changes now enable the rough dielectric to pass the chi^2 tests.
-
- 30 8月, 2021 2 次提交
-
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
Rename texture mapping classes to not have "2D" and "3D" suffixes. Otherwise just naming and formatting changes.
-
- 29 8月, 2021 3 次提交
-
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
pZownZ's x and y components are 0, so....
-
- 28 8月, 2021 2 次提交
-
-
由 Matt Pharr 提交于
We need to transform dpdx and dpdy to texture space first... (Not doing that could lead to overblurring or aliasing, depending on the circumstances.)
-
由 Matt Pharr 提交于
Just variable renaming and formatting...
-
- 27 8月, 2021 6 次提交
-
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
This takes a mutex (that should be held at call time). It returns the future's value if it's available and otherwise unlocks the mutex, tries to do some work, and then returns with the mutex re-locked. This allows mutex-protected future harvesting to work without deadlock...
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
-
由 Matt Pharr 提交于
-