In this test we will compare Geo Light sampling technology in 3Delight OSL, Arnold and RenderMan. We will use a test with a geometric area light that is composed of a relatively large number of faces. The test should allow us to find the convergence rate of each algorithm as well as general performance.
The Scene
The test scene allows for procedural generation of floating area lights using the "Tubes Per Step" PaintEffects attribute.
The scene is designed to procedurally generate a number of area lights floating on top of a "city".
- Only diffuse reflectors. The original test scene had a diffuse+specular surface but that generated a considerable amount of additional noise in Arnold. Since we want to concentrate on light sampling here, we decided to go for a simpler setup.
- We will disable any adaptive sampling so to make sure we have very close ray-counts.
- We will use only direct lighting to estimate the geometric area light contribution. In the statistics files for each renderer, one can see that we have only one path length.
- All Renders are done in Maya 2016.
Methodology
We are interested to find out about:
- The quality of the light samples drawn by each renderer. How fast do we converge to ground truth by increasing the number of samples ?
- Performance. How fast is the sampler to produce an image of a given quality ?
We will start by creating a "ground truth" image for each renderer. This image is generated by using a very large amount of samples so there is no more apparent noise. We will then render several images with varying amount of samples and measure the RMSE between these images and ground truth. Timings will be taken at each render. Having this data will allow us to draw a conclusion about convergence rate and general performance.
The Renderers
| Arnold | RenderMan/RIS | 3Delight OSL | |
|---|---|---|---|
| Version |
|
| |
| Technology | Unidirectional path tracer. | Using unidirectional path tracer. Other options are available but not useful for this test. | Unidirectional path tracer. |
| Shaders | C++ | C++ | OSL |
Notes About Sampling Parameters
Arnold — For light samples, Arnold uses effective sample counts that are proportional – within a constant – to the square of the user specified value. As we will see, this makes sense from a UI standpoint since the variance follows the inverse of the same rule in the case of Arnold. This makes the light samples slider linear in term of perceived noise. In the Arnold tables below, we will specify the effective samplers per pixel along with the user samples.
RenderMan – We had troubles extracting consistent quality from RenderMan. In Arnold and 3Delight, light samples are the single "go to" parameter to control image quality when only direct lighting is considered. In RenderMan/RIS, we had to match light sample count with BxDF sample count to achieve acceptable quality and convergence rates. Using light samples only (or BxDF samples only) produced slowly convergent renders. In RenderMan tables below, "N samples" means N samples for both light and BxDF. We did all the test with the "advanced (4)" light sampler — other samplers did not provide acceptable results for this test case. Note that we used the path tracer with once bounce instead if the "direct lighting" algorithm since the latest would crash once in a while.
3Delight – We have only one control for the general quality of the render. In the case of direct lighting, 3Delight "understands" that samples are best used for light sampling and that's what it does. As tests will show, those samples have a linear impact on perceived noise levels.
Results
Arnold
| Samples (effective) | 2 (1.23) | 4 (4.91) | 8 (19.64) | 16 (78.56) | 32 (314.29) | 64 (1257.18) |
|---|---|---|---|---|---|---|
| Image |  |
|
|
|
|
|
| Time | 1s | 2s | 6s | 21s | 1:21 | 8:12 |
| TTFP | 0s | 0.35s | 1.2s | 3.2s | 11s | 41s |
| Shadow Rays | 0.678 M | 3.26 M | 10.8 M | 43.4 M | 173.6 M | 694.5 M |
| RMSE | 0.15699 | 0.100115 | 0.0501787 | 0.0242515 | 0.0117413 | 0.00693426 |
3Delight
| Samples | 2 | 4 | 8 | 16 | 32 | 64 | 128 | 256 |
|---|---|---|---|---|---|---|---|---|
| Image |
|
|
|
|
|
|
|
|
| Time | 5.72s | 8s | 12.82 s | 22.18 s | 40.92 s | 86.91 | 165.69 s | 302.98 |
| TTFP | 3 | 3 | 3 | 3 | 3 | 3 | 3 | |
| Shadow Rays | 2.45 M | 3.13 M | 4.51 M | 7.27 M | 12.78 M | 23.8 M | 45.88 M | 90.1 M |
| RMSE | 0.0933142 | 0.0658266 | 0.0441248 | 0.0290439 | 0.0185575 | 0.0117566 | 0.00752546 | 0.00449892 |
RenderMan
| Samples | 1 | 2 | 4 | 8 | 32 | 64 | 256 | 512 | 1024 |
|---|---|---|---|---|---|---|---|---|---|
| Image |
|
|
|
|
|
|
|
|
|
| Time | 6.74s | 7.23s | 7.99s | 9.42s | 18.51s | 29.40 | 1:38.08 | 3:15.77 | 6:23.39 |
| TTFP | 1s | 1.5s | 4s | 4.7s | 9s | 15.5s | 51.2s | 97s | |
| Rays | 1.47 M | 2.94 M | 5.88 M | 11.7 M | 47.02 M | 94.14 M | 376.3 M | 751.3 M | 1499 M |
| RMSE | 0.151125 | 0.121487 | 0.0953649 | 0.0728148 | 0.0373876 | 0.0265388 | 0.0138148 | 0.00854045 | 0.00396 |
The following plot gives us a good idea on the algorithm sophistication of the different light samplers.
 |
The following plot gives us a good idea the time required to achieve a certain quality. From the user perspective, this is an important quantity.
 |
The following chart shows how much time it takes to each renderer to build the light acceleration data structure depending on sample count.
 |
Conclusions
- 3Delight generates light samples that are algorithmically better than both Arnold and RenderMan. In short, for N samples:
- 3Delight Varince ~ 1/N
- Arnold Variance ~ 1/sqrt(N)
- RenderMan/RIS Variance ~ 1/sqrt(N)
- 3Delight is slower to generate these samples. Meaning that for draft renders Arnold/PrMan are faster. For final renders 3Delight becomes increasingly faster with the number of samples.
- Both Arnold and RenderMan produce biased images at low sample counts. More specifically: images are darker. This explains higher RMSE with low sample counts. 3Delight manages to keep the same energy overall independent on sample counts.
- Arnold, 3Delight and RenderMan rely on acceleration data structures to sample the geometric area lights. In Arnold and RenderMan, the algorithmic complexity to build those data structures is tied to the number of samples as well as the complexity of the light. In 3Delight, only to the complexity of the light matters (time to first pixel for 3Delight was 3 seconds no matter how many samples there are).