SlideShare a Scribd company logo
1 of 59
Download to read offline
Universitat Polit`ecnica de Catalunya
Physically-based rendering of human skin
Master in Innovation and Research in Informatics
Roger Hernando
Advisors:
Antonio Chica
Pere-Pau V´azquez
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Outline
1 Motivation
2 Implemented skin rendering methods
3 Proposed extensions
4 Developed framework
5 Results
6 Conclusions & future work
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Motivation
Motivation
Enhance the rendering of human scanned characters:
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Introduction
Skin rendering
Human perception is highly
specialized.
Very sensitive to the
appearance of human
skin.
Skin is a multilayered
translucent material:
Light scatters through
skin.
Light interacts with each
layer.
Described in terms of the
BSSRDF not in terms of the
BRDF.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Introduction
BRDF vs BSSRDF
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Introduction
Skin rendering
Skin simulation:
Subsurface scattering.
Forward scattering.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Subsurface scattering
Subsurface scattering
BSSRDF S relates the outgoing radiance L0(x0, −→ω0) at a point
x0 with the incoming radiant flux Φi (xi , −→ωi ) at a point xi :
dL0(x0, −→ω0) = S(xi , −→ωi ; x0, −→ω0)dΦi (xi , −→ωi ) (1)
8D function:
S(xi , −→ωi ; x0, −→ω0) = S(xi , yi , σi , φi ; x0, y0, σ0, φ0) (2)
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Subsurface scattering
Subsurface scattering
Simplified using the diffusion approximation R(r):
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Subsurface scattering
Subsurface scattering
Radiant exitance M at a point (x, y):
M(x, y) = E(x , y )R(r )dx dy (3)
Expressed as 2D convolution:
M(x, y) = E(x, y) ∗ R(r) (4)
For real time rendering, R(r) is approximated by a set of 1D
separable convolutions.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Objective
Objective
Implement state-of-the-art methods to render the skin
(subsurface scattering, forward scattering).
Propose some extensions to the methods.
Implement a test-bed application and also other PBR
techniques.
Test and compare the implemented methods.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Outline
1 Motivation
2 Implemented skin rendering methods
3 Proposed extensions
4 Developed framework
5 Results
6 Conclusions & future work
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Implemented Methods
Methods
Subsurface scattering:
Screen space subsurface scattering.
Separable pre-integrated subsurface scattering.
Separable artistic subsurface scattering.
Forward scattering:
Real-Time realistic skin translucency.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
Subsurface scattering
Subsurface scattering:
Screen space subsurface scattering.
Separable pre-integrated subsurface scattering.
Separable artistic subsurface scattering.
Screen-space methods.
Mimicking the skin diffusion profile.
Subsurface scattering should be applied only to the diffuse
lighting.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
First method: Screen space subsurface scattering
Jimenez et al. used d’Eon and Lubeke approximation R(r) as
a sum of gaussian 6 functions:
Rd (r) =
k
i=1
wi G(vi , r) (5)
Convolution performed in screen space.
12 1D convolutions (expensive).
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
Screen space subsurface scattering Overview
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
Second method: Separable pre-integrated subsurface
scattering
Approximate R(r) with just one separable convolution.
Assuming the irradiance E is additively separable, the
separable kernel is defined as follows:
A(r) =
1
||ap||1
ap(r) (6)
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
Second method: Separable pre-integrated subsurface
scattering
Stages of the algorithm:
Compute the diffusion kernels ap using a Monte Carlo
simulation.
Kernel computation is slow.
Precompute the kernels for later use.
Uneven energy distribution, more energy near the origin.
More samples are taken near the origin.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
Third method: Separable artistic subsurface scattering
Known limitations:
Separable pre-integrated subsurface scattering is not an artistic
friendly method.
Kernel controlled with various parameters.
Based on the sum of gaussians diffusion.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
Third method: Separable artistic subsurface scattering
Parameters:
Weight (w): filter width.
Strength (s): amount of light which penetrates the skin.
Falloff (f ): amount of light travelling through the skin.
A(r) = p
r ∗ w
0.001 + f
∗ s + δ(r) + (1 − s) (7)
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
Third method: Separable artistic subsurface scattering
Highly configurable:
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
Forward scattering
Forward scattering:
Real-Time realistic skin translucency.
Obtain the distance travelled by the light inside an object.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
Fourth method: Real-Time realistic skin translucency
Computing the distance traveled through the object.
light
eye
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
Fourth method: Real-Time realistic skin translucency
Use the path length with the transmittance function T(s):
T(s) =
k
i=1
wi e−s2/vi
(8)
Using d’Eon and Lubeke weights.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Skin rendering methods
Fourth method: Real-Time realistic skin translucency
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Outline
1 Motivation
2 Implemented skin rendering methods
3 Proposed extensions
4 Developed framework
5 Results
6 Conclusions & future work
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Extensions
Extensions
Problems with current methods:
Halos.
Incorrect diffusion.
Based in non-physically-based previous work:
Modulate the subsurface scattering effect with mesh local
curvature.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Halos
Halos
Close points in screen space may be far away in the geometry.
Problem partially tackled by correction factors (but not
enough):
//correction
float depth = texture(depthTex, offset).r;
float s = min(correction * abs(depthM - depth), 1.0);
colorS.rgb = mix(colorS.rgb, colorM.rgb, s);
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Halos
Halos
Mikkelsen uses a Cross Bilateral Filter to weight the diffusion
profile.
CBF[I, E]p =
q∈S Gσs e−||p−q||Gσr e−(Ep−Eq)Iq
q∈S Gσs e−||p−q||Gσr e−(Ep−Eq)
(9)
It works like a bilateral filter but uses an auxiliary image for
weighting.
I(p) = I(x(p)) ∗ cos3
(φi )
||x(p)||2
cos(φj)
(10)
φi is the angle between the z-axis and the view.
φj is the angle between the point normal and the -view.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Halos
Halos
The original implementation depends on the distance between
the eye and the object.
Assume x(p) lies on the zNear plane.
This technique produces strange artifacts when used directly
with our subsurface scattering techniques.
Slightly modify the subsurface scattering algorithms to take
this artifacts into account.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Halos
Halos
Results:
The Halos artifacts are eliminated using this extension.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Incorrect diffusion
Incorrect diffusion
Limitations:
Our meshes do not differentiate between different scanned
elements (skin, hair, cloths).
In video games artists provide this information (texture or
meshes).
We do not have this information.
Blurring between different zones.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Incorrect diffusion
Incorrect diffusion
Proposed Solution:
Weight the filter using the lab-color distance to differentiate
between skin and non-skin zones.
Bilateral filter:
BF[I]p =
q∈S Gσs e−||p−q||Gσs e−(Ip−Iq)Ip
q∈S Gσs e−||p−q||Gσs e−(cp−cq)
(11)
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Incorrect diffusion
Incorrect diffusion
Results:
The diffusion between skin and non-skin zones is eliminated
using this startegy:
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Scattering modulation
Scattering modulation
Non-physically based subsurface scattering: the subsurface
scattering effect is more noticeable in high curvature zones.
Modulate the scattering strength with the screen space
curvature.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Scattering modulation
Scattering modulation
Various strategies to modulate the subsurface scattering.
Increase the effect according to its local curvature.
Decrease the effect in zones with lower curvature and increase
it otherwise.
Reduce the effect up to a minimum at zones with low
curvature and increase it at zones with high curvature.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Scattering modulation
Scattering modulation
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Outline
1 Motivation
2 Implemented skin rendering methods
3 Proposed extensions
4 Developed framework
5 Results
6 Conclusions & future work
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
PBR
Scene Lighting
Features:
Subsurface scattering.
Specular Reflections.
Environment lighting.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Speculars
BRDF Specular
BRDF specular:
Cspec(l, v) =
F(l, h)G(l, v, h)D(h)
4(n · l)(n · v)
Lc(n · l) (12)
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Speculars
BRDF Specular
Fresnel: defines the fraction of light reflected from an
optically flat surface.
Fslick(F0, l, h) = F0 + (1 − F0)(1 − (l · h))5
(13)
Shadow-masking function: defines the percentage of
microfacets with h as their normal vector that are not
shadowed or masked.
Gimplicit(l, v, h) = (n · l)(n · v) (14)
Distribution of normals function: concentration of microfacets
that are oriented such that they could reflect light from l into
v.
DPhong (h) = π
αr + 2
2π
(n · h)αr
(15)
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Speculars
BRDF Specular
Finally, our physically-based specular BRDF is defined as
follows:
Cspec(l, v) =
αr + 2
8
(n · h)αr
Fslick(F0, l, h)Lc(n · l) (16)
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Speculars
BRDF Specular
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Global Ilumination
Irradiance maps
Ambient occlusion.
Environment lighting using an irradiance map.
EMdiff (n) = k∈Ω max(0, lk · n)L(lk)
k∈Ω max(0, lk · n)
(17)
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Global Ilumination
Irradiance maps
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Global Ilumination
Irradiance maps
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Extras
Color linearity
Important to take into account that the color captured by a
sensor is not stored in a linear way.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Extras
Color linearity
When a non-linear color is used, it produces an incorrect
rendering.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Application
Snapshot
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Application
Pipeline
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Outline
1 Motivation
2 Implemented skin rendering methods
3 Proposed extensions
4 Developed framework
5 Results
6 Conclusions & future work
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Performance
Results
Application performance:
Application
Shadow map MainRender AddSpecular Tonemap
0.447 ms 1.262 ms 0.148ms 0.977 ms
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Performance
Results
Subsurface scattering methods:
View Gausian sum Artistic Pre-int Kernel
Close 9.428 ms 1.731 ms 1.799 ms
Mid 2.01 ms 0.492 ms 0.487 ms
Far 0.676 ms 0.312ms 0.36 ms
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Performance
Results
Performance vs #samples:
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Performance
Results
Without subsurface scattering:
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Performance
Results
With subsurface scattering:
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Outline
1 Motivation
2 Implemented skin rendering methods
3 Proposed extensions
4 Developed framework
5 Results
6 Conclusions & future work
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Conclusions
Conclusions
Explored different algorithms to render the skin.
Proposed and explained some extensions to improve the
rendering quality.
Implemented a testbed application integrating the subsurface
scattering methods plus other techniques in order to produce
high quality renders.
Methods performance analysis.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Future work
Future work
Explore non-physically based methods to render the skin.
Improve the re-usability of the forward scattering method.
Explore segmentation methods to distinguish between skin
and non-skin zones.
Roger Hernando Physically-based rendering of human skin
Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w
Questions
Questions
Questions?
Roger Hernando Physically-based rendering of human skin

More Related Content

Similar to Physically based skin rendering

Image enhancement using_piecewise_linear_contrast_
Image enhancement using_piecewise_linear_contrast_Image enhancement using_piecewise_linear_contrast_
Image enhancement using_piecewise_linear_contrast_
Satish Injeti
 
ROBUST STATISTICAL APPROACH FOR EXTRACTION OF MOVING HUMAN SILHOUETTES FROM V...
ROBUST STATISTICAL APPROACH FOR EXTRACTION OF MOVING HUMAN SILHOUETTES FROM V...ROBUST STATISTICAL APPROACH FOR EXTRACTION OF MOVING HUMAN SILHOUETTES FROM V...
ROBUST STATISTICAL APPROACH FOR EXTRACTION OF MOVING HUMAN SILHOUETTES FROM V...
ijitjournal
 
one shot15729752 Deep Learning for AI and DS
one shot15729752 Deep Learning for AI and DSone shot15729752 Deep Learning for AI and DS
one shot15729752 Deep Learning for AI and DS
ManiMaran230751
 

Similar to Physically based skin rendering (20)

Digital Image Processing - Image Restoration
Digital Image Processing - Image RestorationDigital Image Processing - Image Restoration
Digital Image Processing - Image Restoration
 
Skin Color Detection Using Region-Based Approach
Skin Color Detection Using Region-Based ApproachSkin Color Detection Using Region-Based Approach
Skin Color Detection Using Region-Based Approach
 
F0164348
F0164348F0164348
F0164348
 
HSV Brightness Factor Matching for Gesture Recognition System
HSV Brightness Factor Matching for Gesture Recognition SystemHSV Brightness Factor Matching for Gesture Recognition System
HSV Brightness Factor Matching for Gesture Recognition System
 
4d and 4d visualization
4d and 4d visualization 4d and 4d visualization
4d and 4d visualization
 
Image processing
Image processingImage processing
Image processing
 
Recognition of Facial Emotions Based on Sparse Coding
Recognition of Facial Emotions Based on Sparse CodingRecognition of Facial Emotions Based on Sparse Coding
Recognition of Facial Emotions Based on Sparse Coding
 
Region Based Skin Color Detection
Region Based Skin Color DetectionRegion Based Skin Color Detection
Region Based Skin Color Detection
 
Face and Eye Detection Varying Scenarios With Haar Classifier_2015
Face and Eye Detection Varying Scenarios With Haar Classifier_2015Face and Eye Detection Varying Scenarios With Haar Classifier_2015
Face and Eye Detection Varying Scenarios With Haar Classifier_2015
 
Image processing and compression techniques
Image processing and compression techniquesImage processing and compression techniques
Image processing and compression techniques
 
I045034956
I045034956I045034956
I045034956
 
Fast Simulation of Skin Sliding
Fast Simulation of Skin SlidingFast Simulation of Skin Sliding
Fast Simulation of Skin Sliding
 
NIR imaging
NIR imaging NIR imaging
NIR imaging
 
[IJET-V1I5P9] Author: Prutha Gandhi, Dhanashri Dalvi, Pallavi Gaikwad, Shubha...
[IJET-V1I5P9] Author: Prutha Gandhi, Dhanashri Dalvi, Pallavi Gaikwad, Shubha...[IJET-V1I5P9] Author: Prutha Gandhi, Dhanashri Dalvi, Pallavi Gaikwad, Shubha...
[IJET-V1I5P9] Author: Prutha Gandhi, Dhanashri Dalvi, Pallavi Gaikwad, Shubha...
 
Image enhancement using_piecewise_linear_contrast_
Image enhancement using_piecewise_linear_contrast_Image enhancement using_piecewise_linear_contrast_
Image enhancement using_piecewise_linear_contrast_
 
ROBUST STATISTICAL APPROACH FOR EXTRACTION OF MOVING HUMAN SILHOUETTES FROM V...
ROBUST STATISTICAL APPROACH FOR EXTRACTION OF MOVING HUMAN SILHOUETTES FROM V...ROBUST STATISTICAL APPROACH FOR EXTRACTION OF MOVING HUMAN SILHOUETTES FROM V...
ROBUST STATISTICAL APPROACH FOR EXTRACTION OF MOVING HUMAN SILHOUETTES FROM V...
 
one shot15729752 Deep Learning for AI and DS
one shot15729752 Deep Learning for AI and DSone shot15729752 Deep Learning for AI and DS
one shot15729752 Deep Learning for AI and DS
 
Illumination Invariant Hand Gesture Classification against Complex Background...
Illumination Invariant Hand Gesture Classification against Complex Background...Illumination Invariant Hand Gesture Classification against Complex Background...
Illumination Invariant Hand Gesture Classification against Complex Background...
 
Report face recognition : ArganRecogn
Report face recognition :  ArganRecognReport face recognition :  ArganRecogn
Report face recognition : ArganRecogn
 
20230213_ComputerVision_연구.pptx
20230213_ComputerVision_연구.pptx20230213_ComputerVision_연구.pptx
20230213_ComputerVision_연구.pptx
 

Recently uploaded

Why Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businessWhy Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire business
panagenda
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Safe Software
 

Recently uploaded (20)

Why Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businessWhy Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire business
 
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
 
AXA XL - Insurer Innovation Award Americas 2024
AXA XL - Insurer Innovation Award Americas 2024AXA XL - Insurer Innovation Award Americas 2024
AXA XL - Insurer Innovation Award Americas 2024
 
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingRepurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
 
Corporate and higher education May webinar.pptx
Corporate and higher education May webinar.pptxCorporate and higher education May webinar.pptx
Corporate and higher education May webinar.pptx
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
 
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, AdobeApidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
 
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdfRising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
 
ICT role in 21st century education and its challenges
ICT role in 21st century education and its challengesICT role in 21st century education and its challenges
ICT role in 21st century education and its challenges
 
MS Copilot expands with MS Graph connectors
MS Copilot expands with MS Graph connectorsMS Copilot expands with MS Graph connectors
MS Copilot expands with MS Graph connectors
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processors
 
DBX First Quarter 2024 Investor Presentation
DBX First Quarter 2024 Investor PresentationDBX First Quarter 2024 Investor Presentation
DBX First Quarter 2024 Investor Presentation
 
Apidays New York 2024 - APIs in 2030: The Risk of Technological Sleepwalk by ...
Apidays New York 2024 - APIs in 2030: The Risk of Technological Sleepwalk by ...Apidays New York 2024 - APIs in 2030: The Risk of Technological Sleepwalk by ...
Apidays New York 2024 - APIs in 2030: The Risk of Technological Sleepwalk by ...
 
Ransomware_Q4_2023. The report. [EN].pdf
Ransomware_Q4_2023. The report. [EN].pdfRansomware_Q4_2023. The report. [EN].pdf
Ransomware_Q4_2023. The report. [EN].pdf
 
Manulife - Insurer Transformation Award 2024
Manulife - Insurer Transformation Award 2024Manulife - Insurer Transformation Award 2024
Manulife - Insurer Transformation Award 2024
 
Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...
Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...
Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...
 
Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...
 
[BuildWithAI] Introduction to Gemini.pdf
[BuildWithAI] Introduction to Gemini.pdf[BuildWithAI] Introduction to Gemini.pdf
[BuildWithAI] Introduction to Gemini.pdf
 
Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024
 
"I see eyes in my soup": How Delivery Hero implemented the safety system for ...
"I see eyes in my soup": How Delivery Hero implemented the safety system for ..."I see eyes in my soup": How Delivery Hero implemented the safety system for ...
"I see eyes in my soup": How Delivery Hero implemented the safety system for ...
 

Physically based skin rendering

  • 1. Universitat Polit`ecnica de Catalunya Physically-based rendering of human skin Master in Innovation and Research in Informatics Roger Hernando Advisors: Antonio Chica Pere-Pau V´azquez
  • 2. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Outline 1 Motivation 2 Implemented skin rendering methods 3 Proposed extensions 4 Developed framework 5 Results 6 Conclusions & future work Roger Hernando Physically-based rendering of human skin
  • 3. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Motivation Motivation Enhance the rendering of human scanned characters: Roger Hernando Physically-based rendering of human skin
  • 4. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Introduction Skin rendering Human perception is highly specialized. Very sensitive to the appearance of human skin. Skin is a multilayered translucent material: Light scatters through skin. Light interacts with each layer. Described in terms of the BSSRDF not in terms of the BRDF. Roger Hernando Physically-based rendering of human skin
  • 5. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Introduction BRDF vs BSSRDF Roger Hernando Physically-based rendering of human skin
  • 6. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Introduction Skin rendering Skin simulation: Subsurface scattering. Forward scattering. Roger Hernando Physically-based rendering of human skin
  • 7. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Subsurface scattering Subsurface scattering BSSRDF S relates the outgoing radiance L0(x0, −→ω0) at a point x0 with the incoming radiant flux Φi (xi , −→ωi ) at a point xi : dL0(x0, −→ω0) = S(xi , −→ωi ; x0, −→ω0)dΦi (xi , −→ωi ) (1) 8D function: S(xi , −→ωi ; x0, −→ω0) = S(xi , yi , σi , φi ; x0, y0, σ0, φ0) (2) Roger Hernando Physically-based rendering of human skin
  • 8. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Subsurface scattering Subsurface scattering Simplified using the diffusion approximation R(r): Roger Hernando Physically-based rendering of human skin
  • 9. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Subsurface scattering Subsurface scattering Radiant exitance M at a point (x, y): M(x, y) = E(x , y )R(r )dx dy (3) Expressed as 2D convolution: M(x, y) = E(x, y) ∗ R(r) (4) For real time rendering, R(r) is approximated by a set of 1D separable convolutions. Roger Hernando Physically-based rendering of human skin
  • 10. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Objective Objective Implement state-of-the-art methods to render the skin (subsurface scattering, forward scattering). Propose some extensions to the methods. Implement a test-bed application and also other PBR techniques. Test and compare the implemented methods. Roger Hernando Physically-based rendering of human skin
  • 11. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Outline 1 Motivation 2 Implemented skin rendering methods 3 Proposed extensions 4 Developed framework 5 Results 6 Conclusions & future work Roger Hernando Physically-based rendering of human skin
  • 12. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Implemented Methods Methods Subsurface scattering: Screen space subsurface scattering. Separable pre-integrated subsurface scattering. Separable artistic subsurface scattering. Forward scattering: Real-Time realistic skin translucency. Roger Hernando Physically-based rendering of human skin
  • 13. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods Subsurface scattering Subsurface scattering: Screen space subsurface scattering. Separable pre-integrated subsurface scattering. Separable artistic subsurface scattering. Screen-space methods. Mimicking the skin diffusion profile. Subsurface scattering should be applied only to the diffuse lighting. Roger Hernando Physically-based rendering of human skin
  • 14. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods First method: Screen space subsurface scattering Jimenez et al. used d’Eon and Lubeke approximation R(r) as a sum of gaussian 6 functions: Rd (r) = k i=1 wi G(vi , r) (5) Convolution performed in screen space. 12 1D convolutions (expensive). Roger Hernando Physically-based rendering of human skin
  • 15. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods Screen space subsurface scattering Overview Roger Hernando Physically-based rendering of human skin
  • 16. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods Second method: Separable pre-integrated subsurface scattering Approximate R(r) with just one separable convolution. Assuming the irradiance E is additively separable, the separable kernel is defined as follows: A(r) = 1 ||ap||1 ap(r) (6) Roger Hernando Physically-based rendering of human skin
  • 17. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods Second method: Separable pre-integrated subsurface scattering Stages of the algorithm: Compute the diffusion kernels ap using a Monte Carlo simulation. Kernel computation is slow. Precompute the kernels for later use. Uneven energy distribution, more energy near the origin. More samples are taken near the origin. Roger Hernando Physically-based rendering of human skin
  • 18. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods Third method: Separable artistic subsurface scattering Known limitations: Separable pre-integrated subsurface scattering is not an artistic friendly method. Kernel controlled with various parameters. Based on the sum of gaussians diffusion. Roger Hernando Physically-based rendering of human skin
  • 19. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods Third method: Separable artistic subsurface scattering Parameters: Weight (w): filter width. Strength (s): amount of light which penetrates the skin. Falloff (f ): amount of light travelling through the skin. A(r) = p r ∗ w 0.001 + f ∗ s + δ(r) + (1 − s) (7) Roger Hernando Physically-based rendering of human skin
  • 20. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods Third method: Separable artistic subsurface scattering Highly configurable: Roger Hernando Physically-based rendering of human skin
  • 21. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods Forward scattering Forward scattering: Real-Time realistic skin translucency. Obtain the distance travelled by the light inside an object. Roger Hernando Physically-based rendering of human skin
  • 22. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods Fourth method: Real-Time realistic skin translucency Computing the distance traveled through the object. light eye Roger Hernando Physically-based rendering of human skin
  • 23. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods Fourth method: Real-Time realistic skin translucency Use the path length with the transmittance function T(s): T(s) = k i=1 wi e−s2/vi (8) Using d’Eon and Lubeke weights. Roger Hernando Physically-based rendering of human skin
  • 24. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Skin rendering methods Fourth method: Real-Time realistic skin translucency Roger Hernando Physically-based rendering of human skin
  • 25. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Outline 1 Motivation 2 Implemented skin rendering methods 3 Proposed extensions 4 Developed framework 5 Results 6 Conclusions & future work Roger Hernando Physically-based rendering of human skin
  • 26. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Extensions Extensions Problems with current methods: Halos. Incorrect diffusion. Based in non-physically-based previous work: Modulate the subsurface scattering effect with mesh local curvature. Roger Hernando Physically-based rendering of human skin
  • 27. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Halos Halos Close points in screen space may be far away in the geometry. Problem partially tackled by correction factors (but not enough): //correction float depth = texture(depthTex, offset).r; float s = min(correction * abs(depthM - depth), 1.0); colorS.rgb = mix(colorS.rgb, colorM.rgb, s); Roger Hernando Physically-based rendering of human skin
  • 28. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Halos Halos Mikkelsen uses a Cross Bilateral Filter to weight the diffusion profile. CBF[I, E]p = q∈S Gσs e−||p−q||Gσr e−(Ep−Eq)Iq q∈S Gσs e−||p−q||Gσr e−(Ep−Eq) (9) It works like a bilateral filter but uses an auxiliary image for weighting. I(p) = I(x(p)) ∗ cos3 (φi ) ||x(p)||2 cos(φj) (10) φi is the angle between the z-axis and the view. φj is the angle between the point normal and the -view. Roger Hernando Physically-based rendering of human skin
  • 29. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Halos Halos The original implementation depends on the distance between the eye and the object. Assume x(p) lies on the zNear plane. This technique produces strange artifacts when used directly with our subsurface scattering techniques. Slightly modify the subsurface scattering algorithms to take this artifacts into account. Roger Hernando Physically-based rendering of human skin
  • 30. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Halos Halos Results: The Halos artifacts are eliminated using this extension. Roger Hernando Physically-based rendering of human skin
  • 31. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Incorrect diffusion Incorrect diffusion Limitations: Our meshes do not differentiate between different scanned elements (skin, hair, cloths). In video games artists provide this information (texture or meshes). We do not have this information. Blurring between different zones. Roger Hernando Physically-based rendering of human skin
  • 32. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Incorrect diffusion Incorrect diffusion Proposed Solution: Weight the filter using the lab-color distance to differentiate between skin and non-skin zones. Bilateral filter: BF[I]p = q∈S Gσs e−||p−q||Gσs e−(Ip−Iq)Ip q∈S Gσs e−||p−q||Gσs e−(cp−cq) (11) Roger Hernando Physically-based rendering of human skin
  • 33. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Incorrect diffusion Incorrect diffusion Results: The diffusion between skin and non-skin zones is eliminated using this startegy: Roger Hernando Physically-based rendering of human skin
  • 34. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Scattering modulation Scattering modulation Non-physically based subsurface scattering: the subsurface scattering effect is more noticeable in high curvature zones. Modulate the scattering strength with the screen space curvature. Roger Hernando Physically-based rendering of human skin
  • 35. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Scattering modulation Scattering modulation Various strategies to modulate the subsurface scattering. Increase the effect according to its local curvature. Decrease the effect in zones with lower curvature and increase it otherwise. Reduce the effect up to a minimum at zones with low curvature and increase it at zones with high curvature. Roger Hernando Physically-based rendering of human skin
  • 36. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Scattering modulation Scattering modulation Roger Hernando Physically-based rendering of human skin
  • 37. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Outline 1 Motivation 2 Implemented skin rendering methods 3 Proposed extensions 4 Developed framework 5 Results 6 Conclusions & future work Roger Hernando Physically-based rendering of human skin
  • 38. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w PBR Scene Lighting Features: Subsurface scattering. Specular Reflections. Environment lighting. Roger Hernando Physically-based rendering of human skin
  • 39. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Speculars BRDF Specular BRDF specular: Cspec(l, v) = F(l, h)G(l, v, h)D(h) 4(n · l)(n · v) Lc(n · l) (12) Roger Hernando Physically-based rendering of human skin
  • 40. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Speculars BRDF Specular Fresnel: defines the fraction of light reflected from an optically flat surface. Fslick(F0, l, h) = F0 + (1 − F0)(1 − (l · h))5 (13) Shadow-masking function: defines the percentage of microfacets with h as their normal vector that are not shadowed or masked. Gimplicit(l, v, h) = (n · l)(n · v) (14) Distribution of normals function: concentration of microfacets that are oriented such that they could reflect light from l into v. DPhong (h) = π αr + 2 2π (n · h)αr (15) Roger Hernando Physically-based rendering of human skin
  • 41. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Speculars BRDF Specular Finally, our physically-based specular BRDF is defined as follows: Cspec(l, v) = αr + 2 8 (n · h)αr Fslick(F0, l, h)Lc(n · l) (16) Roger Hernando Physically-based rendering of human skin
  • 42. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Speculars BRDF Specular Roger Hernando Physically-based rendering of human skin
  • 43. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Global Ilumination Irradiance maps Ambient occlusion. Environment lighting using an irradiance map. EMdiff (n) = k∈Ω max(0, lk · n)L(lk) k∈Ω max(0, lk · n) (17) Roger Hernando Physically-based rendering of human skin
  • 44. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Global Ilumination Irradiance maps Roger Hernando Physically-based rendering of human skin
  • 45. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Global Ilumination Irradiance maps Roger Hernando Physically-based rendering of human skin
  • 46. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Extras Color linearity Important to take into account that the color captured by a sensor is not stored in a linear way. Roger Hernando Physically-based rendering of human skin
  • 47. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Extras Color linearity When a non-linear color is used, it produces an incorrect rendering. Roger Hernando Physically-based rendering of human skin
  • 48. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Application Snapshot Roger Hernando Physically-based rendering of human skin
  • 49. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Application Pipeline Roger Hernando Physically-based rendering of human skin
  • 50. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Outline 1 Motivation 2 Implemented skin rendering methods 3 Proposed extensions 4 Developed framework 5 Results 6 Conclusions & future work Roger Hernando Physically-based rendering of human skin
  • 51. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Performance Results Application performance: Application Shadow map MainRender AddSpecular Tonemap 0.447 ms 1.262 ms 0.148ms 0.977 ms Roger Hernando Physically-based rendering of human skin
  • 52. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Performance Results Subsurface scattering methods: View Gausian sum Artistic Pre-int Kernel Close 9.428 ms 1.731 ms 1.799 ms Mid 2.01 ms 0.492 ms 0.487 ms Far 0.676 ms 0.312ms 0.36 ms Roger Hernando Physically-based rendering of human skin
  • 53. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Performance Results Performance vs #samples: Roger Hernando Physically-based rendering of human skin
  • 54. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Performance Results Without subsurface scattering: Roger Hernando Physically-based rendering of human skin
  • 55. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Performance Results With subsurface scattering: Roger Hernando Physically-based rendering of human skin
  • 56. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Outline 1 Motivation 2 Implemented skin rendering methods 3 Proposed extensions 4 Developed framework 5 Results 6 Conclusions & future work Roger Hernando Physically-based rendering of human skin
  • 57. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Conclusions Conclusions Explored different algorithms to render the skin. Proposed and explained some extensions to improve the rendering quality. Implemented a testbed application integrating the subsurface scattering methods plus other techniques in order to produce high quality renders. Methods performance analysis. Roger Hernando Physically-based rendering of human skin
  • 58. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Future work Future work Explore non-physically based methods to render the skin. Improve the re-usability of the forward scattering method. Explore segmentation methods to distinguish between skin and non-skin zones. Roger Hernando Physically-based rendering of human skin
  • 59. Motivation Implemented skin rendering methods Proposed extensions Developed framework Results Conclusions & future w Questions Questions Questions? Roger Hernando Physically-based rendering of human skin