Cycles/EEVEE: change point light to double-sided sphere light #108506

Weizhen Huang · 2023-06-01T11:32:59+02:00

Weizhen Huang commented

2023-06-01 11:32:59 +02:00

for energy preservation and better compatibility with other renderes. Ref: #108505

Point light now behaves the same as a spherical mesh light with the same overall energy (scaling from emission strength to power is (4\pi^2R^2)).

Cycles

Comparison

Mesh Light	This patch	Previous behavior

The behavior stays the same when radius = 0.

This patch	Previous behavior

No obvious performance change observed.

Sampling

When shading point lies outside the sphere, sample the spanned solid angle uniformly.
When shading point lies inside the sphere, sample spherical direction uniformly when inside volume or the surface is transmissive, otherwise sample cosine-weighted upper hemisphere.

Light Tree

When shading point lies outside the sphere, treat as a disk light spanning the same solid angle.
When shading point lies inside the sphere, it behaves like a background light, with estimated outgoing radiance

L_o=\int f_aL_i\cos\theta_i\mathrm{d}\omega_i=\int f_a\frac{E}{\pi r^2}\cos\theta_i\mathrm{d}\omega_i\approx f_a \frac{E}{r^2}

with f_a being the BSDF and E measure.energy in light_tree.cpp.
The importance calculation for LIGHT_POINT is

L_o=f_a E\cos\theta_i\frac{\cos\theta}{d^2}

Consider min_importance = 0 because maximal incidence angle is \pi, we could substitute d^2 with \frac{r^2}{2} so the averaged outgoing radiance is f_a \frac{E}{r^2}.
This only holds for non-transmissive surface, but should be fine to use in volume.

EEVEE

When shading point lies outside the sphere, the sphere light is equivalent to a disk light spanning the same solid angle. The sine of the new half-angle is the tangent of the previous half-angle.
When shading point lies inside the sphere, integrating over the cosine-weighted hemisphere gives 1.0.

Comparison with Cycles

The plane is diffuse, the blue sphere has specular component.

Before		After
Cycles	EEVEE	Cycles	EEVEE

TODO

Point light should be renamed as sphere light at various places.
Overlay and gizmo should adapt to the new size, but they share the same code with spot light, so maybe wait until changes for spot light.

for energy preservation and better compatibility with other renderes. Ref: #108505 Point light now behaves the same as a spherical mesh light with the same overall energy (scaling from emission strength to power is \(4\pi^2R^2\)). # Cycles ## Comparison | Mesh Light | This patch | Previous behavior | | -------- | -------- | -------- | | ![mesh_1024](attachments/2900954c-57f8-49c2-b6f3-8fb559b820ac) | ![sphere_1024](attachments/148241ca-9350-48b6-be04-3933e015424c) | ![point_1024](attachments/d9b19d54-2b00-4986-ba8c-c4b28f687f09) | The behavior stays the same when `radius = 0`. | This patch | Previous behavior | | -------- | -------- | | ![sphere_64](attachments/aa05d59a-146a-4f69-b257-5d09a7f41d4e) | ![point_64](attachments/69a743be-bc15-454b-92d8-af02f4e8ab07) | No obvious performance change observed. ## Sampling When shading point lies outside the sphere, sample the spanned solid angle uniformly. When shading point lies inside the sphere, sample spherical direction uniformly when inside volume or the surface is transmissive, otherwise sample cosine-weighted upper hemisphere. ## Light Tree When shading point lies outside the sphere, treat as a disk light spanning the same solid angle. When shading point lies inside the sphere, it behaves like a background light, with estimated outgoing radiance \[L_o=\int f_aL_i\cos\theta_i\mathrm{d}\omega_i=\int f_a\frac{E}{\pi r^2}\cos\theta_i\mathrm{d}\omega_i\approx f_a \frac{E}{r^2}\], with \(f_a\) being the BSDF and \(E\) `measure.energy` in `light_tree.cpp`. The importance calculation for `LIGHT_POINT` is \[L_o=f_a E\cos\theta_i\frac{\cos\theta}{d^2}\]. Consider `min_importance = 0` because maximal incidence angle is \(\pi\), we could substitute \(d^2\) with \(\frac{r^2}{2}\) so the averaged outgoing radiance is \(f_a \frac{E}{r^2}\). This only holds for non-transmissive surface, but should be fine to use in volume. # EEVEE When shading point lies outside the sphere, the sphere light is equivalent to a disk light spanning the same solid angle. The sine of the new half-angle is the tangent of the previous half-angle. When shading point lies inside the sphere, integrating over the cosine-weighted hemisphere gives 1.0. ## Comparison with Cycles The plane is diffuse, the blue sphere has specular component. | Before | |After || |---|--|--|--| |Cycles|EEVEE|Cycles|EEVEE| |![](attachments/5824c494-0645-461a-b193-d74e02f353b8)|![](attachments/d2e85b53-3c2a-4a9f-a3b2-6e11c6083ce0)|![](attachments/a8dcdd8b-c13c-4fdc-808c-2563624549be)|![](attachments/8c3618ef-1ab4-4210-9535-c85e873f1e45)| # TODO - [ ] Point light should be renamed as sphere light at various places. - [ ] Overlay and gizmo should adapt to the new size, but they share the same code with spot light, so maybe wait until changes for spot light.

point_light_1024spp_2.63s.png

16 KiB

cycles_before.png

12 KiB

eevee_after.png

8.0 KiB

eevee_before.png

8.0 KiB

sphere_light_64spp_7.png

17 KiB

point_light_64spp_7.png

17 KiB

cycles_after.png

13 KiB

sphere_light_1024spp_2.51s.png

17 KiB

mesh_light_1024spp_7.7s.png

22 KiB

👍 4

Weizhen Huang added the

Module

Render & Cycles

label 2023-06-01 11:32:59 +02:00

Weizhen Huang added 7 commits 2023-06-01 11:33:11 +02:00

8839881b69 Change disk sampling to solid angle sampling for point light

1498b94516 Change point light to one-sided sphere light

955477406b Sample sphere light from inside

8ec1fcabd3 Cleanup: rename variables in #ray_sphere_intersect for clarity

d0196da81e Cleanup: avoid computing squared root when possible

74db71e83c Use different sampling techniques on surface and in volume when inside sphere light

bb7817232e Cleanup: remove todo

Weizhen Huang referenced this pull request

2023-06-01 11:33:40 +02:00

Fix wrong conversion between power and radiance of emitters #108505

Weizhen Huang referenced this pull request from blender/cycles

2023-06-01 11:54:15 +02:00

Fixed bug with un-normalised ellipse area lights not having proper area calculation and erroneously reverting to quad lights. #4

Weizhen Huang force-pushed point_light_to_sphere_light from 0153e574cd to abcacaa58e

2023-06-01 14:02:52 +02:00

Compare

Weizhen Huang added 1 commit 2023-06-02 16:32:51 +02:00

56c0c48a1b Fix `ls->Ng` not pointing from light center to sampled position

Weizhen Huang added 1 commit 2023-06-08 14:49:48 +02:00

2848f4a646 Merge remote-tracking branch 'origin' into point_light_to_sphere_light

Weizhen Huang force-pushed point_light_to_sphere_light from 00b3bd8f23 to 19bf32dcc6

2023-06-08 15:24:00 +02:00

Compare

Weizhen Huang added 1 commit 2023-06-09 20:08:25 +02:00

0af8ac3874 Adjust comments regarding point/spot light in EEVEE

Weizhen Huang referenced this pull request

2023-06-15 15:39:28 +02:00

Cycles: Change sun lamp to have uniform intensity at high angles #108996

Lukas Stockner referenced this pull request

2023-06-16 00:36:14 +02:00

Cycles: Change sun lamp to have uniform intensity at high angles #108996

Weizhen Huang added 2 commits 2023-06-19 12:52:38 +02:00

4891a4ffef Merge branch 'main' into point_light_to_sphere_light

9e816b8cef Eevee: change point light to sphere light

Weizhen Huang changed title from ~~WIP: Cycles: change point light to sphere light~~ to Cycles/EEVEE: change point light to sphere light

2023-06-19 13:39:04 +02:00

Weizhen Huang added the

 @ -23,2 +18,2 @@
   if (radius > 0.0f) {
     ls->P += disk_light_sample(lightN, rand) * radius;
   float3 lightN = P - klight->co;
   const float d_sqr = len_squared(lightN);

 @ -159,2 +176,2 @@
     points[0] = Px * -ld._area_size_x + Py * -ld._area_size_y;
     points[1] = Px * ld._area_size_x + Py * -ld._area_size_y;
     if (ld.type == LIGHT_POINT) {
       /* Sphere light spans a larger angle than a disk light with the same radius (sin -> tan). */

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Cycles/EEVEE: change point light to double-sided sphere light #108506

Cycles

Comparison

Sampling

Light Tree

EEVEE

Comparison with Cycles

TODO