Fix intersection distance offset in Cycles #107364

Sergey Sharybin · 2023-04-26T11:03:11+02:00

Sergey Sharybin commented

2023-04-26 11:03:11 +02:00

The intersection distance offset in Cycles could have returned
a denormal floating point value for the input values of 0 (and
for the denormal input value).

This could lead to a situation when ray is unable to be advanced
when it hits an edge between two triangles: the intersection will
keep bouncing between two adjacent triangles. This is because the
ray->tmin is compared inclusively, and 0 >= .

The solution is to return the smallest possible normalized floating
point value from the intersection_t_offset if the input is zero
or a denormal value (which is covered by the same t == 0 check).

This fix is hard to measure on the user level. The old code did
not cause any infinite traversal loop because of the way how the
integration is organized (some kernels offset ray.P, others check
for the number of bounces). It is possible that this fixes some
corner cases of noise: i.e. if some shadow rays falsefully were
considered occluded due to reached maximum number of bounces.

The actual problematic case was discovered during working on a
prototype which had an in-lined intersection loop with the ray
tmin offset.

The intersection distance offset in Cycles could have returned a denormal floating point value for the input values of 0 (and for the denormal input value). This could lead to a situation when ray is unable to be advanced when it hits an edge between two triangles: the intersection will keep bouncing between two adjacent triangles. This is because the ray->tmin is compared inclusively, and 0 >= <denormal zero>. The solution is to return the smallest possible normalized floating point value from the intersection_t_offset if the input is zero or a denormal value (which is covered by the same t == 0 check). This fix is hard to measure on the user level. The old code did not cause any infinite traversal loop because of the way how the integration is organized (some kernels offset ray.P, others check for the number of bounces). It is possible that this fixes some corner cases of noise: i.e. if some shadow rays falsefully were considered occluded due to reached maximum number of bounces. The actual problematic case was discovered during working on a prototype which had an in-lined intersection loop with the ray tmin offset.

Sergey Sharybin added the

Module

Render & Cycles

label 2023-04-26 11:03:11 +02:00

Sergey Sharybin added 1 commit 2023-04-26 11:03:18 +02:00

buildbot/vexp-code-patch-coordinator Build done. Details

2f9211ed3f Fix intersection distance offset in Cycles

The intersection distance offset in Cycles could have returned
a denormal floating point value for the input values of 0 (and
for the denormal input value).

This could lead to a situation when ray is unable to be advanced
when it hits an edge between two triangles: the intersection will
keep bouncing between two adjacent triangles. This is because the
ray->tmin is compared inclusively, and 0 >= <denormal zero>.

The solution is to return the smallest possible normalized floating
point value from the intersection_t_offset if the input is zero
or a denormal value (which is covered by the same t == 0 check).

This fix is hard to measure on the user level. The old code did
not cause any infinite traversal loop because of the way how the
integration is organized (some kernels offset ray.P, others check
for the number of bounces). It is possible that this fixes some
corner cases of noise: i.e. if some shadow rays falsefully were
considered occluded due to reached maximum number of bounces.

The actual problematic case was discovered during working on a
prototype which had an in-lined intersection loop with the ray
tmin offset.

Sergey Sharybin added this to the Render & Cycles project 2023-04-26 11:03:25 +02:00

Sergey Sharybin commented

2023-04-26 11:03:31 +02:00

@blender-bot build

Sergey Sharybin requested review from Brecht Van Lommel 2023-04-26 11:03:39 +02:00

Brecht Van Lommel approved these changes 2023-04-26 11:10:11 +02:00

Sergey Sharybin merged commit 02b5c04b2b into main

2023-04-26 11:36:43 +02:00

Sergey Sharybin referenced this issue from a commit

2023-04-26 11:36:43 +02:00

Fix intersection distance offset in Cycles

Sergey Sharybin deleted branch cycles-denormals-offset