Analysis of Super-Sampling Methods
Ray Tracing
The
Problem
a. the least amount of
samples per pixel
b. the least amount of
resources
while preserving the image quality without introduction of
artifacts such as jagged edges
and/or Moire patterns, etc..
References
Sampling
Methods/Patterns
Discussion
Removing aliases can be done using these sampling patterns. With higher number of samples per pixel the jagged edges are much smaller and therefore, almost invisible to the human eye. However, higher quality comes with a cost of longer runtime as each sample takes more samples per pixel. Hopefully, we can make the runtime shorter by applying adaptive strategy. Unfortunately, we ran out of time to implement correctly and were not able to see the expected results.
2 Samples - Multi-Threaded
Grid with 4 samples – 2.756 seconds (visible Moire Pattern)
Rotated Grid with 4 samples – 2.674 seconds (visible Moire Pattern)
Jitter with 4 samples – 2.583 seconds (visible Moire Pattern)
Random with 2 samples – 2.631 (jagged edges visible)
Random with 4 samples - 3.001 seconds (jagged edges visible)
NRooks with 4 samples – 2.660 seconds ( a bit of jagged edges)
Multi-Jitter with 8 pixels – 9.263 (great but expensive)
3 samples per pixel – Multi-Threaded – No Filter
GRID – 5.240 (visible Moire Pattern)
Rotated Grid – 5.141 (visible Moire Pattern)
Jitter – 5.311
NRooks – 5.272 (a bit of Jagged edges)
Multi-Jitter – 42.984 (good but expensive)
Single Thread |
|
|
|
|
|
|
Sample |
Random |
Grid |
Jittered |
RG |
Multi-Jettered |
Nrooks |
Box Filter |
|
|
|
|
|
|
2 |
5.112 |
7.436 |
7.253 |
7.413 |
22.093 |
7.368 |
3 |
6.197 |
13.069 |
13.377 |
13.377 |
98.456 |
13.391 |
4 |
7.36 |
21.416 |
21.603 |
21.452 |
296.823 |
21.449 |
Guassian |
|
|
|
|
|
|
2 |
4.65 |
7.017 |
7.452 |
7.32 |
24.299 |
7.22 |
3 |
5.834 |
14.184 |
14.354 |
14.283 |
115.238 |
14.563 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Multi-Thread (no filtrer) |
|
|
|
|
|
|
Sample |
Random |
Grid |
Jittered |
RG |
Multi-Jettered |
Nrooks |
2 |
2.631 |
2.756 |
2.583 |
2.675 |
9.263 |
2.66 |
3 |
2.422 |
5.29 |
9.311 |
5.142 |
42.984 |
5.272 |
9 |
5.564 |
|
|
|
|
|