ICTGraphicsLab – Skin Microstructure Deformation With Displacement Map Convolution
The Accurate Source To Find Transcript To ICTGraphicsLab – Skin Microstructure Deformation With Displacement Map Convolution.”
[ICTGraphicsLab – Skin Microstructure Deformation With Displacement Map Convolution]
SIGGRAPH 2015 Technical Papers
SKIN MICROSTRUCTURE DEFORMATION WITH DISPLACEMENT MAP CONVOLUTION
Koki Nagano Graham Fyffe Oleg Alexander
Jarnej Barbic* Hao Li* Abhijeet Ghosh** Paul Debevec
USC Institute for Creative Technologies
University of Southern California* Imperial College London**
Human skin is a complex multilayer material, which exhibits a range of effects under deformation at both the meso-and micro-sales.
ALEXANDER ET AL 2009 BEELER ET AL 2011
BEELER ET AL 2010 ALEXANDER ET AL 2013
[ICT Graphics Laboratory:] Source: LYBIO.net
While, modern scanning techniques record mesostructure dynamics at a sub-millimeter resolution. Microstructure dynamics below a tenth of a millimeter are generally not recorded despite their effect on facial reflectance and appearance.
RECORDING MICROSTRUCTURE UNDER DEFORMATION
LED SPEHRE MACRO LENS CAMERA
In this work, we acquire the shape of deforming skin microstructure using polarized gradient illumination, macro photography, and a skin measurement gantry able to stretch and compress skin at measuring intervals.
MATERIAL STRETCH SCANNING GANTRY
RECORDED SKIN MICROSTRUCTURE DATA UNDER MULTIPLE LIGHTING CONDITIONS
Each scan performed in under half a second records the skin surface shape at 10 micron resolution.
DYNAMIC DISPLACEMENT MAPS
From the normal maps derived from the lighting conditions, we integrate a displacement map for each amount of deformation.
RECORDED SKIN PATCH SEQUENCE
FOREHEAD SCAN UNDER CHANGING STRESS
The surface roughness changes under deformation has seen in the surface orientation histograms over the sample area. We also measured the effects of deformation at several skin orientations. Graphing this data, we see that skin becomes shinier when its stretched and rougher when its compressed.
CONVOLVING DISPLACEMENT MAP
Using the measured microstructure normal distributions as a guide, we blur the microstructure in the direction of stretching and sharpen the microstructure in the direction of compression.
SIMULATING STRETCHING AND COMPRESSION
Thus the surface flattens when stretched and appears to bunch up when compressed. This yield similar directional texture and changing surface roughness as the surface deforms, even though it doesn’t simulate the full complexity of dynamic microstructure.
DYNAMIC MICROGEOMETRY STATIC MICROGEOMETRY
Here, we use this skin microstructure deformation model on a deforming sphere, which becomes shinier when it expands and rougher when it shrinks. When squashed and stretched, its microstructure develops an isotropic texture. The dynamic microstructure gives a stronger sense of surface strain especially in this twisting example. We apply our model to a slab of skin, where the skin mesostructure is simulated using a low resolution finite element simulation. And the micro structure deformation is simulated using displacement map convolution.
Here we apply the technique while blending between two facial scans where the mesostructure is linearly interpolated between the scans and the dynamic microstructure is created through displacement map convolution. Visualizing the local surface strain gives a sense of which parts of the displacement map are being blurred and sharpened and in which direction.
Seen without microstructure, the mesostructure scans on their own miss significant aspects of realistic skin appearance.
[ICT Graphics Laboratory:] Source: LYBIO.net
Adding a static microstructure displacement map improves the skin like quality, but doesn’t look natural where the skin compresses and stretches significantly.
Adding the dynamic microstructure map provides a more visceral sense of surface tension and emphasizes the expression. The effect can also be seen more clearly by visualizing the specular surface reflection on its own.
We can also apply this technique to a captured facial performance. This animation transitions between different 4K mesostructure maps and uses dynamic microgeometry computed on the fly from a 16K microstructure displacement map.
Again, the dynamic microgeometry produces a noticeable effect on the skin reflectance and is an additional indication of the deformation and strain of the skin.
PHOTO DYNAMIC MICROGEOMETRY
We validate the technique by showing side by side photos of the face with a compressed expression lit by a point light and the corresponding simulated expression. As desired, the specular reflections shows similar texture orientation and roughness variation in the photos and renderings.
RENDERED WITH VRAY
PROJECT DIRECTOR KOKI NAGANO
PERFORMANCE CAPTURE GRAHAM FYFFE
RENDING ARTIST CHIHYUAN (JASON) HUANG
DIGITAL ARTIST OLEG ALEXANDER
DIGITAL ARTIST JAY BUSCH
RENDING SUPERVISOR CHRISTOPHER NICHOLS
SENIOR RENDING ENGINEER VLADIMIR KOYLAZOV
DIGITAL ARTIST RUSKO RUSKOV
MODELER MATHIEU AERNI
HAIR ARTIST DANNY YOUNG
VIDEO NARRATION TODD RICHMOND
SENIOR SUPERVISOR PAUL DEBEVEC
THE DIGITAL HUMAN LEAGUE WILL RETURN…
Learn more and visit ICT Graphics Lab:
ICTGraphicsLab – Skin Microstructure Deformation With Displacement Map Convolution. Each scan performed in under half a second records the skin surface shape at 10 micron resolution. Science & Technology Transcript, Dialogue, Remarks, Saying, Quotes, Words And Text.