Efficient Acquisition and Realistic Rendering of Car Paint
- Slides: 24
Efficient Acquisition and Realistic Rendering of Car Paint Johannes Günther, Tongbo Chen, Michael Goesele, Ingo Wald, and Hans-Peter Seidel MPI Informatik Saarbrücken, Germany
Motivation § Virtual prototyping, car design by computer § Mainly two materials – Glass: ok, physical properties well known – Car paint: not so easy § Goal: Realistic appearance of virtual cars, close to reality November 18, 2005 Phong BRDF: “plastic” look VMV, Erlangen, Germany 2
Outline § Introduction & Previous Work § Efficient Acquisition – Measurement Setup – BRDF Representation and Modelling § Realistic Rendering – BRDF Evaluation – Illumination – Simulation of Sparkling § Results § Conclusion & Future Work November 18, 2005 VMV, Erlangen, Germany 3
Previous Work § BRDF Acquisition [Marschner ‘ 98, Matusik ‘ 03] – Image based, automatic fast § Car paint [Ershov ‘ 01, ‘ 04] – Complex models, many effects – Not designed for animation context § Illumination by Environment Maps [Debevec ‘ 98] § Realtime Ray Tracing [Wald ’ 01, ‘ 04] November 18, 2005 VMV, Erlangen, Germany 4
Outline § § § Introduction & Previous Work Efficient Acquisition Realistic Rendering Results Conclusion & Future Work November 18, 2005 VMV, Erlangen, Germany 5
Measurement Setup CCD camera white LED painted sphere turn table November 18, 2005 VMV, Erlangen, Germany 6
Measurement Process § Turn table: rotate light source 180° every 1° § At each position: take HDR image – One view direction, one light direction – Sphere: each pixel different normal many BRDF sample at once § Time: ca. 30 minutes per target November 18, 2005 VMV, Erlangen, Germany 7
Targets November 18, 2005 VMV, Erlangen, Germany 8
Modeling § Use Cook-Torrance BRDF – physically derived (micro facets) – showed to perform well [Ngan EGSR ‘ 05] § Non-linear fitting § Multiple lobes to account for nature of car paints November 18, 2005 VMV, Erlangen, Germany 9
Modeling § Use Cook-Torrance BRDF – physically derived (micro facets) – showed to perform well [Ngan EGSR ‘ 05] § Non-linear fitting § Multiple lobes to account for nature of car paints highlight glitter reflectance (clear coat) (flakes) φ base color November 18, 2005 VMV, Erlangen, Germany 10
Outline § § § Introduction & Previous Work Efficient Acquisition Realistic Rendering Results Conclusion & Future Work November 18, 2005 VMV, Erlangen, Germany 11
Complex Illumination § HDR Environment Maps for direct illumination § Options for BRDF evaluation: a) Sample Environment Map – Good for specular BRDFs Decompose car paint BRDF into diffuse part and highly specular part November 18, 2005 reflectance b) Sample BRDF VMV, Erlangen, Germany highly specular – Discretize into directional lights [Kollig ‘ 03, Agarwal ‘ 03, …] – Works well for diffuse BRDFs car paint BRDF t li p s mostly diffuse φ 12
Sparkles § Prominent feature of metallic paints § Tiny bright spots when viewed from close distance clear coat § Caused by mirror-like flakes § Reflect light directly to eye November 18, 2005 flakes base color VMV, Erlangen, Germany 13
Modeling Flakes § Coherent sparkles during animation Model flakes explicitly (the normal) § (Integrated) sparkles appear as glitter in BRDF Derive statistical flake distribution from fitted glitter lobe § Use procedural normal map § Flakes are very small anti-aliasing by over sampling November 18, 2005 VMV, Erlangen, Germany 14
Outline § § § Introduction & Previous Work Efficient Acquisition Realistic Rendering Results Conclusion & Future Work November 18, 2005 VMV, Erlangen, Germany 15
Model Comparison Phong BRDF table November 18, 2005 VMV, Erlangen, Germany Clear. Coat™ fitted BRDF 16
Video November 18, 2005 VMV, Erlangen, Germany 17
Conclusion Easy-to-build and fast acquisition system Measured car paint and measured lighting environment for convincing car renderings Frame-to-frame coherent sparkling simulation § Future Work – Extend car paint database – Multi-level methods for sparkles (avoid aliasing) November 18, 2005 VMV, Erlangen, Germany 18
Data sets available § Project homepage: http: //www. mpi-inf. mpg. de/~guenther/carpaint/ November 18, 2005 VMV, Erlangen, Germany 19
Questions? § Project homepage: http: //www. mpi-inf. mpg. de/~guenther/carpaint/ Thank You November 18, 2005 VMV, Erlangen, Germany 20
November 18, 2005 VMV, Erlangen, Germany 21
Performance vs. Quality § Cluster of 20 dual Opteron 2. 5 GHz PCs § Vary parameter to tune rendering speed or quality 640 × 480 16 lights no over sampling 12. 1 fps November 18, 2005 640 × 480 128 lights 16 spp 1. 3 fps VMV, Erlangen, Germany 1280 × 960 1024 lights 64 spp 97 sec 22
Offline Rendering November 18, 2005 VMV, Erlangen, Germany 23
The Different Car Paints November 18, 2005 VMV, Erlangen, Germany 24
- Not be en presente simple
- Car paint rendering
- If car a passes car b, then car a must be ____.
- Productively efficient vs allocatively efficient
- Productively efficient vs allocatively efficient
- Productively efficient vs allocatively efficient
- Allocative efficiency vs productive efficiency
- Productive inefficiency and allocative inefficiency
- Second language vs foreign language
- A 1000 kg car and a 2000 kg car is hoisted the same height
- Car 2 car communication consortium
- Image-based modeling and rendering
- Realistic job preview advantages and disadvantages
- Benefits of hr forecasting
- Michalangelo
- Arsir
- Hanspeter pfister
- Rendering pipeline
- High dynamic range rendering
- Rendering pipeline in computer graphics
- Jerry tessendorf
- Chris buehler
- Reyes rendering
- Photorealistic rendering carlsbad
- Clustered forward rendering