Monte Carlo Methods for physically based Volume rendering ...
Transcript of Monte Carlo Methods for physically based Volume rendering ...
![Page 1: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/1.jpg)
Monte Carlo Methods for physically based Volume renderingMonte Carlo Methods for physically based Volume rendering SIGGRAPH 2018 CourseSIGGRAPH 2018 Course Advanced methods and acceleration data structuresAdvanced methods and acceleration data structures
Johannes Hanika Karlsruhe Institute of Technology Weta Digital
1
![Page 2: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/2.jpg)
contentcontentrainbows and heroesrainbows and heroes
spectral trackinghero wavelengths
emissive volumesemissive volumesfires, explosions, forward next event estimation
acceleration data structuresacceleration data structuresadaptive storage for heterogeneous media
2
![Page 3: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/3.jpg)
spectral trackingspectral trackingexample material: human skin, insanely complex layered structure
expensive to model and trace througho�en used in graphics: approximate by homogeneous chromatic medium
captures main look features: red blurlow path vertex count for more efficient simulation
3
![Page 4: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/4.jpg)
spectral trackingspectral trackinga problem o�en encountered in skin: chromatic media
collision coefficients depend on wavelength for instance free flight distance much longer for long wavelengths:
makes path invalid for different wavelength?can we still exploit coherence?
μ λ
4
![Page 5: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/5.jpg)
spectral trackingspectral trackinga problem o�en encountered in skin: chromatic media
collision coefficients depend on wavelength for instance free flight distance much longer for long wavelengths:
makes path invalid for different wavelength?can we still exploit coherence?
μ λ
5
![Page 6: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/6.jpg)
spectral tracking via MISspectral tracking via MIShero wavelength sampling [WND∗14]
450 500 550 600 650 700
450 500 550 600 650 700
sample perfectly for one single wavelength evaluate path for a stratified set of wavelengths at the same timeoptimally weighted combination via MIS (balance heuristic)
limited to regular tracking because it requires explicit evaluation of PDF
[WND∗14] Wilkie A., Nawaz S., Droske M., Weidlich A., Hanika J.: Hero wavelength spectral sampling. CGF (Proc. EGSR) 33, 4 (June 2014), 123–131.
λ0
λi
f( , )x̄ λi
p( , )∑j x̄ λj
6
![Page 7: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/7.jpg)
spectral tracking via MISspectral tracking via MISimage comparison 64sppimage comparison 64spp
skin material with 1 wavelength
7
![Page 8: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/8.jpg)
spectral tracking via MISspectral tracking via MISimage comparison 64sppimage comparison 64spp
skin material with 4 wavelengths (SSE)
8
![Page 9: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/9.jpg)
spectral tracking via MISspectral tracking via MISimage comparison 64sppimage comparison 64spp
skin material with 8 wavelengths (AVX)
note that all these images are using the exact same paths!
9
![Page 10: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/10.jpg)
spectral tracking without PDF [KHLN17]spectral tracking without PDF [KHLN17]sample fictitious event by common majorant how do decide for null collision, scattering, or absorption at next vertex ?probability according to
is a normalisation constant such that the sum to one
[KHLN17] Kutz P., Habel R., Li Y. K., Novák J.: Spectral and decomposition tracking for rendering heterogeneous volumes. ACM TOG (Proc. SIGGRAPH) 36, 4 (July 2017)
μ̄
x
P⋆
(x, λ) ∈ (x, λ), (x, λ), (x, λ)μ⋆ μn μs μa
(x) = reduce(| (x, λ)|)cP⋆ μ⋆
c P⋆
10
![Page 11: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/11.jpg)
spectral tracking without PDF [KHLN17]spectral tracking without PDF [KHLN17]probability according to pick by maximum over (always follow densest material)
pick by average weighted by spectral path throughput history
0.0 0.2 0.4 0.6 0.8 1.0
distance
2
3
4
5
6
extinction
µ.redµ.greenµ.blue
RGBmedium
0.0 0.2 0.4 0.6 0.8 1.0
distance0
1
2
3
4
5 throughput.redthroughput.greenthroughput.blue
Maximum
0.0 0.2 0.4 0.6 0.8 1.0
distance0
1
2
3
4
5
max throughput
throughput.redthroughput.greenthroughput.blue
History-aware avg.
[KHLN17] Kutz P., Habel R., Li Y. K., Novák J.: Spectral and decomposition tracking for rendering heterogeneous volumes. ACM TOG (Proc. SIGGRAPH) 36, 4 (July 2017)
P⋆ (x, λ) ∈ (x, λ), (x, λ), (x, λ)μ⋆ μn μs μa
λi
(x) = max(| (x, λ)|)cP⋆ μ⋆
( , λ)w̄ x̄
( ) = reduce(| ( , λ) (x, λ)|)cP⋆ xj w̄ x̄ μ⋆
11
![Page 12: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/12.jpg)
spectral tracking without PDF [KHLN17]spectral tracking without PDF [KHLN17]sample by common majorant how do decide for null collision, scattering, or absorption?probability according to , ,
pick by maximum over pick by average weighted by spectral path throughput historyresults in different noise patterns:
μ̄
(λ)μn (λ)μs (λ)μa
λi
12
![Page 13: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/13.jpg)
spectral trackingspectral trackinga few differences:a few differences:
sampling optimal for hero wavelengthsampling carfully balanced for all wavelengths
most important difference:most important difference:require PDFcannot provide PDFmeans considering one or the other you need to balance requirements of your system
do you need to mix in other importance sampling strategies?for instance equi-angular sampling?
13
![Page 14: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/14.jpg)
emissive mediaemissive mediathin/dense media make a differencethin/dense media make a difference
no event inside the medium means we cannot pick up emission:
density
14
![Page 15: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/15.jpg)
emissive mediaemissive mediathin/dense media make a differencethin/dense media make a difference
follow the idea of beams, collect emission along a ray
particularly well suited for regular tracking, touching all voxels anyways
15
![Page 16: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/16.jpg)
emissive mediaemissive mediathin/dense media make a differencethin/dense media make a difference
direct MIS combination with NEE [VH13] introduces noise: Point + NEE Line + NEE
reason: NEE cannot create paths with end point outside the mediumforward scattering PDF is poor, however, and now it picks up line emission!
[VH13] Villemin R., Hery C.: Practical illumination from flames. Journal of Computer Graphics Techniques 2, 2 (2013).
16
![Page 17: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/17.jpg)
emissive mediaemissive mediawhere do the additional firefly samples come from?where do the additional firefly samples come from?
example of a problematic pathNEE cannot create these pathsforward scattering is same bad as usual, but now has a contribution from the segment!
[SHZD17] Simon F., Hanika J., Zirr T., Dachsbacher C.: Line integration for rendering heterogeneous emissive volumes. CGF (Proc. EGSR) 36, 4 (June 2017).
17
![Page 18: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/18.jpg)
emissive mediaemissive mediaforward next event estimation (FNEE)forward next event estimation (FNEE)
next event estimation which also considers line emission [SHZD17]: Point + NEE Line + NEE Line + FNEE
NEE picks point in medium but only uses the direction to itline segment sampling completed by distance sampling
[SHZD17] Simon F., Hanika J., Zirr T., Dachsbacher C.: Line integration for rendering heterogeneous emissive volumes. CGF (Proc. EGSR) 36, 4 (June 2017).
18
![Page 19: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/19.jpg)
acceleration data structuresacceleration data structures
19
![Page 20: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/20.jpg)
acceleration data structuresacceleration data structuresnaive gridnaive grid
store coefficients per voxelonly single (half) float density and global cross sections ? If you can!motion blur?
(x, λ)μ⋆
ρ(x) σ⋆
20
![Page 21: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/21.jpg)
acceleration data structuresacceleration data structuresmotion blurmotion blur
store time resolved volume in 4D [Wre16]rasterise temporal output of simulation directly
[Wre16] Wrenninge M.: Efficient rendering of volumetric motion blur using temporally unstructured volumes. Journal of Computer Graphics Techniques (JCGT) 5, 1 (January 2016), 1–34.
21
![Page 22: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/22.jpg)
acceleration data structuresacceleration data structuresmotion blurmotion blur
store time resolved volume in 4D [Wre16]Ramer Douglas Peucker line compression per voxel supports non-linear motion
for instance up: density , right: time :
ρ(x, t) t
https://en.wikipedia.org/wiki/Ramer%E2%80%93Douglas%E2%80%93Peucker_algorithm
22
![Page 23: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/23.jpg)
acceleration data structuresacceleration data structuresmotion blurmotion blur
store time resolved volume in 4D [Wre16]supports non-linear motion better than Eulerian motion blur [KK07]
[Wre16] Wrenninge M.: Efficient rendering of volumetric motion blur using temporally unstructured volumes. Journal of Computer Graphics Techniques (JCGT) 5, 1 (January 2016), 1–34.
[KK07] Kim D., Ko H.-S.: Eulerian motion blur. In Eurographics Workshop on Natural Phenomena (2007).
23
![Page 24: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/24.jpg)
acceleration data structuresacceleration data structureshierarchical grid (aka super voxels [SKTM11])hierarchical grid (aka super voxels [SKTM11])
hierarchical traversal, local memory accesswell suited for null collision based trackers:
store majorants on coarse grid, fine data on fine level
[SKTM11] Szirmay-Kalos L., Tóth B., Magdics M.: Free path sampling in high resolution inhomogeneous participating media. CGF 30, 1 (2011), 85–97.
24
![Page 25: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/25.jpg)
acceleration data structuresacceleration data structuresregular trackingregular tracking
multi-level 3D DDA/Bresenham to enumerate all voxels pierced by rayexample: free flight distance sampling
25
![Page 26: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/26.jpg)
acceleration data structuresacceleration data structuresregular trackingregular tracking
multi-level 3D DDA/Bresenham to enumerate all voxels pierced by raywalk along ray direction
26
![Page 27: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/27.jpg)
acceleration data structuresacceleration data structuresregular trackingregular tracking
multi-level 3D DDA/Bresenham to enumerate all voxels pierced by raymemory access to super voxel block
27
![Page 28: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/28.jpg)
acceleration data structuresacceleration data structuresregular trackingregular tracking
memory access to voxels stored in brickall voxels are typically local in memory now
28
![Page 29: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/29.jpg)
acceleration data structuresacceleration data structuresregular trackingregular tracking
if no intersection is found in this blockcontinue to next block
29
![Page 30: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/30.jpg)
acceleration data structuresacceleration data structuresregular trackingregular tracking
accumulate optical thickness if found free flight distance, sub voxel accuracy by assuming homogeneous voxel
τ ← τ + ρ(x) ⋅ σt
τ >= − log (1 − ξ)
30
![Page 31: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/31.jpg)
acceleration data structuresacceleration data structuresnull collision trackingnull collision tracking
perform regular tracking on the coarse level [SKTM11]start of the algorithm very much the same as regular tracking
31
![Page 32: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/32.jpg)
acceleration data structuresacceleration data structuresnull collision trackingnull collision tracking
perform regular tracking on the coarse level [SKTM11]accumulate fictitious optical thickness τ ← τ + (x) ⋅ρ̄ σt
32
![Page 33: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/33.jpg)
acceleration data structuresacceleration data structuresnull collision trackingnull collision tracking
perform regular tracking on the coarse level [SKTM11]if found free flight distance inside this block τ >= − log (1 − ξ)
33
![Page 34: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/34.jpg)
acceleration data structuresacceleration data structuresnull collision trackingnull collision tracking
if found free flight distance inside this blockfind voxel position assuming homogeneous fictitious matter in this block
τ >= − log (1 − ξ)
x
34
![Page 35: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/35.jpg)
acceleration data structuresacceleration data structuresnull collision trackingnull collision tracking
find voxel position assuming homogeneous fictitious matter in this blockterminate if actual collision, i.e. (else restart tracking)
this last point is critical for performance, we'll get back to this!
x
ζ < (x)/ (x)μt μ̄t
35
![Page 36: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/36.jpg)
acceleration data structuresacceleration data structuresis a uniform grid the best we can do?is a uniform grid the best we can do?
regular tracking:
minimise visible integration error:larger voxels in uniform areas, out of frustum, and in depth
null collisions:
minimise amount of fictitious mattermeans minimise steps taken until a real event is found
end result:
fewer, larger voxels in uniform areas, empty space culled awaynull collisions only care about coarse level
36
![Page 37: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/37.jpg)
acceleration data structuresacceleration data structureskd-treeskd-trees
space partitioning scheme by finding largest empty rectangle [YIC*11]
[YIC∗11] Yue Y., Iwasaki K., Chen B., Dobashi Y., Nishita T.: Toward optimal space partitioning for unbiased, adaptive free path sampling of inhomogeneous participating media. CGF (Proc. Pacific Graphics) 30, 7 (2011), 1911–1919.
37
![Page 38: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/38.jpg)
acceleration data structuresacceleration data structures
combine super voxels [SKTM11] and kd-trees [YIC∗11] with adaptive blocks
adaptivity driven by
pixel footprint / camera tessellationheterogeneity / variation
just as super voxels: kd nodes store majorants in coarse blocksμ̄
38
![Page 39: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/39.jpg)
acceleration data structuresacceleration data structures
combine super voxels [SKTM11] and kd-trees [YIC∗11] with adaptive blocks
adaptivity driven by
pixel footprint / camera tessellationheterogeneity / variation
just as super voxels: kd nodes store majorants in coarse blocksperform regular tracking on coarse blocks [SKTM11]
μ̄
39
![Page 40: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/40.jpg)
acceleration data structuresacceleration data structures
combine super voxels [SKTM11] and kd-trees [YIC∗11] with adaptive blocks
adaptivity driven by
pixel footprint / camera tessellationheterogeneity / variation
just as super voxels: kd nodes store majorants in coarse blocksperform regular tracking on coarse blocks [SKTM11]access on fine levels to sample collision type
μ̄
(λ), (λ)μs μa
40
![Page 41: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/41.jpg)
acceleration data structuresacceleration data structuresregular trackingregular tracking
needs to step through every voxel, bad for fine tessellationswell chosen tessellation is a big advantage!
distance
dens
ity
distance
dens
ity
41
![Page 42: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/42.jpg)
acceleration data structuresacceleration data structuresnull collision-based trackingnull collision-based tracking
independent of tessellationefficient in tightly bounded, thin media (mean free path longer than voxel width)
distance
dens
ity
42
![Page 43: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/43.jpg)
acceleration data structuresacceleration data structuresnull collision-based trackingnull collision-based tracking
independent of tessellationinefficient for loose bounds, avoid during hierarchy construction!
distance
dens
ity
distance
dens
ity
43
![Page 44: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/44.jpg)
acceleration data structuresacceleration data structuresnull collision-based trackingnull collision-based tracking
independent of tessellationhigh number of events in dense media, regardless of tessellation!
distance
dens
ity
distance
dens
ityaccessing the memory within the same voxel is still expensivealleviated by decomposition tracking [KHLN17]
44
![Page 45: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/45.jpg)
acceleration data structuresacceleration data structuresdecomposition tracking [KHLN17]decomposition tracking [KHLN17]
separate coefficients into sum of components:coarse voxels for homogeneous partssparse fine details added on top
track through coarse part first and use for early out
also profits regular tracking
μ
45
![Page 46: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/46.jpg)
end of my partend of my partup next:up next:Jaroslav to present future work and outlook
46
![Page 47: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/47.jpg)
47
![Page 48: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/48.jpg)
summarysummaryfree flight distance samplingfree flight distance sampling
woodcock/delta trackingtransmittance estimationtransmittance estimation
track-lengthresidual ratiofree flight versions
path samplingpath samplingpath space formulationsummary of advanced methods
acceleration structuresacceleration structuresfor regular trackingfor null collisions (bottom-level)
48
![Page 49: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/49.jpg)
open research problemsopen research problemsnull collision algorithms and MISnull collision algorithms and MIS
missing link to integrate into powerful frameworkfor instance combine with equi-angular sampling
can we estimate the PDF?
expectation and division do not commute!
X =f( )x̄
p( )x̄
49
![Page 50: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/50.jpg)
open research problemsopen research problemsleverage recent advances in machine learningleverage recent advances in machine learning
special purpose denoisingincluding a volume prior?
path guiding for volumes?joint importance sampling for multiple vertices?
50
![Page 51: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/51.jpg)
open research problemsopen research problemsjoint handling of surfaces and geometryjoint handling of surfaces and geometry
still o�en surface transport is handled separatelymakes inclusion of all interreflections hardcustom-cut algorithms increase maintenance cost
represent surfaces as volumes, too?and then ideally jointly downscale for LOD!
51
![Page 52: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/52.jpg)
open research problemsopen research problemsgeneralisation to correlated scatterersgeneralisation to correlated scatterers
core assumption of exponential path length: uncorrelated particles!particle repulsion such as in cell growth is very correlatedreally, no collision can be found inside the current particle (min distance)some existing work
[d'Eon 2018, Jarabo et al. 2018, Bitterli et al. 2018]
52
![Page 53: Monte Carlo Methods for physically based Volume rendering ...](https://reader031.fdocuments.in/reader031/viewer/2022032215/622fbc066556f67e630b2bc4/html5/thumbnails/53.jpg)
thank you!thank you!any questions?any questions?
acknowledgements:acknowledgements:Peter Kutz
for tracing down many of early delta tracking papersreviewers
for feedback on the paper dra�Maurizio Nitti
for help w/ illustrations
53