Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

  • 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 ?
  • How do the algorithms scale with more complex area light geometry ?

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 renderer. Having this data will allow us to draw a conclusion about convergence rate and general performance.

Notes About Sampling Parameters

For light samples, Arnold use the effective sample counts that are  proportional – within a constant –  to the square of the UI value.  As we will see, this makes perfect sense is the variance follows this same rule in the case of Arnold. This makes the light samples slider linear in term of perceived noise.


3Delight

RenderMan


Results - RMSE

...

Chart
width600
titleRMSE vs. TIME
typexyLine
yLabelRMSE
xLabelTime
Time5.72s812.8222.1840.9286.91165.69302.98
3Delight0.09331420.06582660.04412480.02904390.01855750.01175660.00752546

0.00449892

Time1262181492
Arnold0.156990.1001150.05017870.02425150.01174130.00693426
Time6.747.237.999.4218.5129.4098.08383.39
RenderMan0.1511250.1214870.09536490.07281480.03738760.02653880.01381480.00854045

Results - Scalability

For this test we will render 3 images that have a similar RMSE in all renderers but with a much higher area light count. We increase the Tube Per Step to 200 (10 times more area lights) and re-time the results. 

Arnold3DelightRenderManSamples32 (314)64256Time with "20" Setting81s86.9s98sTime with "200" Setting186sNot available

Preliminary Conclusions (for self, re-wording needed)

...