Download presentation

Presentation is loading. Please wait.

Published byOswald O’Neal’ Modified about 1 year ago

1
Clipless Dual-Space Bounds for Faster Stochastic Rasterization Samuli Laine Timo Aila Tero Karras Jaakko Lehtinen NVIDIA Research

5
(x,y)

6
(x,y,u,v,t)

7
Motion Blur

8
t=0t=1

9
Motion Blur

10
t Accumulation Buffer [Haeberli ‘90]

11
t InterleaveUVT [Fatahalian ‘09]

12
t

13
t

14
t

15
4 samples/pixel (16 UVT triples)16 samples/pixel (64 UVT triples) Scene: Assassin’s Creed, courtesy of Ubisoft InterleaveUVT [Fatahalian ‘09]

16
Unique UVTs 4 samples/pixel, unique UVTs16 samples/pixel, unique UVTs Scene: Assassin’s Creed, courtesy of Ubisoft

17
5D Rasterization t=0t=1 t

18
5D Rasterization t=0t=1 t

19
5D Rasterization t=0t=1 t

20
5D Rasterization t

21
t=0t=1 t ?

22
5D Rasterization t=0t=1 t

23
5D Rasterization t=0t=1 t ?

24
5D Rasterization t=0t=1 t ?

25
Pixar Algorithm t=0t=1 t t=.5

26
Pixar Algorithm t=0t=1 t t=.5 XX

27
Pixar Algorithm t=0t=1 t t=.5 XX

28
Our Method t=0t=1 t

29
Our Method t=0t=1 t

30
Our Method t=0t=1 t XXX t range computed for pixel

31
Our Method Determine pixels potentially covered by triangle For each pixel Compute time bounds t min,t max Enumerate samples where t min ≤ t ≤ t max For each such sample, perform 5D coverage test Same for lens bounds u min,u max and v min,v max

32
Bound Computation Lens bounds computed in screen space Time bounds computed in dual space

33
Lens Bounds x y u=0u= –1u= +1 film lens focal plane u screen-space x is linear with u

34
Lens Bounds x y u= –1u= +1 screen-space x is linear with u

35
Lens Bounds x y u=au=b screen-space x is linear with u u a b

36
Time Bounds World-space affine motion Not affine in screen space, but affine in clip space Perspective causes singularities in screen space Operate in dual space

37
Clip Space and Dual Space x w t=0 t=1

38
Clip Space and Dual Space x w γ δ γ= –1 γ= +1

39
Clip Space and Dual Space x w γ δ δ { γ= –0.5

40
Clip Space and Dual Space x w γ δ

41
x w γ δ

42
x w γ δ

43
x w γ δ δ is linear in x and w δ = x – wγ x and w are linear in t δ is linear in t t=0 t=1 t=0 t=1

44
Clip Space and Dual Space x w γ δ t=0 t=1 t=0 t=1

45
Clip Space and Dual Space x w γ δ t=0 t=1 t=0 t=1

46
Clip Space and Dual Space x w γ δ t=0 t=1 t=0 t=1 t=a t

47
Clip Space and Dual Space x w γ δ t=0 t=1 t=0 t=1 t=a t a

48
Time Bounds Compute separately for x and y Intersect resulting spans If intersection is empty, skip pixel

49
Recap For each triangle For each pixel Compute t, u, v bounds Cull samples outside bounds Profit

50
Results Measure sample test efficiency (STE) Compare against methods that allow arbitrary sampling patterns # samples tested with full 5D test # samples hit STE =

51
Results Scene: Age of Conan PC MMO, courtesy of Funcom Bbox scan Pixar Our method 41664 static23.6 motion2.79.517.721.923.7 motion x 21.36.014.620.924.0 defocus1.74.48.813.923.1 defocus x 20.71.84.48.821.9 both0.71.22.64.35.6 both x 20.40.61.12.02.9 STE in %, scene Conan

52
Results Scene: Assassin’s Creed, courtesy of Ubisoft Bbox scan Pixar Our method 41664 static23.2 motion10.819.522.423.123.4 motion x 24.314.821.223.023.6 defocus9.515.119.021.123.2 defocus x 24.39.515.119.023.0 both4.56.87.210.114.1 both x 21.32.13.96.76.9 STE in %, scene Assassin

53
Results Bbox scan Pixar Our method 41664 static8.598.608.598.608.59 motion0.502.976.438.028.63 motion x 20.141.274.707.408.69 defocus0.190.591.533.138.57 defocus x 20.050.190.591.538.51 both0.120.250.491.084.51 both x 20.030.070.160.392.42 STE in %, scene Cars

54
Conclusions Each triangle processed once Arbitrary sampling patterns High STE Future work Combined motion + defocus case

55
Thank You

Similar presentations

© 2017 SlidePlayer.com Inc.

All rights reserved.

Ads by Google