Cycles: improve ray tracing precision near triangle edges

Detect cases where a ray-intersection would miss the current triangle, which if
the intersection is strictly watertight, implies that a neighboring triangle would
incorrectly be hit instead.

When that is detected, apply a ray-offset. The idea being that we only want to
introduce potential error from ray offsets if we really need to.

This work for BVH2 and Embree, as we are able to match the ray-interesction
bit-for-bit, though doing so for Embree requires ugly hacks. Tiny differences
like fused-multiply-add or dot product intrinstics in matrix inversion and ray
intersection needed to be matched exactly, so this is fragile.

Unfortunately we're not able to do the same for OptiX or MetalRT, since those
implementations are unknown (and possibly impossible to match as hardware
instructions). Still artifacts are much reduced, though not eliminated.

Ref T97259

Differential Revision: https://developer.blender.org/D15559

1 files changed, 24 insertions, 0 deletions

diff --git a/intern/cycles/kernel/bvh/util.h b/intern/cycles/kernel/bvh/util.h
index b67c9394bea..a57703a8b8c 100644
--- a/intern/cycles/kernel/bvh/util.h
+++ b/intern/cycles/kernel/bvh/util.h
@@ -33,6 +33,30 @@ ccl_device_forceinline float intersection_t_offset(const float t)
   return __uint_as_float(bits);
 }
 
+/* Ray offset to avoid self intersection.
+ *
+ * This function can be used to compute a modified ray start position for rays
+ * leaving from a surface. This is from:
+ * "A Fast and Robust Method for Avoiding Self-Intersection"
+ * Ray Tracing Gems, chapter 6.
+ */
+ccl_device_inline float3 ray_offset(const float3 P, const float3 Ng)
+{
+  const float int_scale = 256.0f;
+  const int3 of_i = make_int3(
+      (int)(int_scale * Ng.x), (int)(int_scale * Ng.y), (int)(int_scale * Ng.z));
+
+  const float3 p_i = make_float3(
+      __int_as_float(__float_as_int(P.x) + ((P.x < 0) ? -of_i.x : of_i.x)),
+      __int_as_float(__float_as_int(P.y) + ((P.y < 0) ? -of_i.y : of_i.y)),
+      __int_as_float(__float_as_int(P.z) + ((P.z < 0) ? -of_i.z : of_i.z)));
+  const float origin = 1.0f / 32.0f;
+  const float float_scale = 1.0f / 65536.0f;
+  return make_float3(fabsf(P.x) < origin ? P.x + float_scale * Ng.x : p_i.x,
+                     fabsf(P.y) < origin ? P.y + float_scale * Ng.y : p_i.y,
+                     fabsf(P.z) < origin ? P.z + float_scale * Ng.z : p_i.z);
+}
+
 #ifndef __KERNEL_GPU__
 ccl_device int intersections_compare(const void *a, const void *b)
 {