High Dynamic Range Lighting Paul Debevec, USC Institute for Creative Technologies. March 24, Game Developer s Conference 1

Transcription

1 March 4, 004 High Dynamic Range Lighting Paul Debevec University of Southern California Institute for Creative Technologies March 4, 004 5:30 6:30 pm Scenes lit with point light sources lack realism Real-World HDR Lighting Environments Funston Beach Eucalyptus Grove Illuminating Objects using Measurements of Real Light Uffizi Gallery Grace Cathedral Lighting Environments from the Light Probe Image Gallery: Object Light Environment assigned glow material property in Greg Ward s RADIANCE system. Elements of HDRI and IBL High Dynamic Range (HDR) Images Pixels beyond 0-55 Pixel proportional to light levels Light Probe Images Omnidirectional HDR images, or HDR environment maps Lighting with real illumination environments yields greater realism Global Illumination Illuminating CG objects with images of incident illumination 004 Game Developer s Conference 1

2 March 4, 004 IBL Tutorial Dynamic Range in the Real World In Jan/Feb Computer Graphics and Applications and the SIGGRAPH 00 IBL Course Notes Sony VX000 video camera Office interior Indirect light from window 1/60 th sec shutter f/5.6 aperture 0 ND filters 0dB gain Dynamic Range in the Real World Outside in the shade 1/1000 th sec shutter f/5.6 aperture 0 ND filters 0dB gain Dynamic Range in the Real World Outside in the sun 1/1000 th sec shutter f/11 aperture 0 ND filters 0dB gain 16 times the light as inside 64 times the light as inside Dynamic Range in the Real World Straight at the sun 1/10,000 th sec shutter f/11 aperture 13 stops ND filters 0dB gain Dynamic Range in the Real World Very dim room 1/4 th sec shutter f/1.6 aperture 0 stops ND filters 18dB gain 5,000,000 times the light as inside 1/1500 th the light than inside 004 Game Developer s Conference

3 March 4, 004,000,000, ,000 Dynamic Range in the Real World High-Dynamic Range Photography 5, Debevec and Malik, Recovering High Dynamic Range Radiance Maps from Photographs, SIGGRAPH ,000 : 1 Visualization: Greg Ward Chris Tchou and Paul Debevec. HDR Shop. SIGGRAPH 001 Technical Sketch A m o u n t o f L i g h t Gamma. graph Pixel Value 55 Implications: 18 is less than ¼ as bright as is more than 4 times as bright as is twice as bright as is half as bright as = / = 93 See also Charles Poynton s Gamma FAQ: DirectX 9 HDR Data Formats 3-bit floating point textures D3DFMT_A3B3G3R3F / D3DFMT_R3F IEEE compatible 16-bit floating point textures D3DFMT_A16B16G16R16F saves memory bandwidth often sufficient dynamic range and precision HDR Image File Formats 004 Game Developer s Conference 3

4 March 4, 004 HDR Formats: RADIANCE Format (.pic,.hdr) Greg Ward s Real Pixels format 3 bits / pixel HDR Formats: Portable FloatMap (.pfm) 1 bytes per pixel, 4 for each channel Red Green Blue Exponent (145, 15, 87, 149) = (145, 15, 87, 103) = (145, 15, 87) * ^(149-18) = (145, 15, 87) * ^(103-18) = ( , , ) ( , , ) Ward, Greg. "Real Pixels," in Graphics Gems IV, edited by James Arvo, Academic Press, 1994 sign exponent mantissa Text header similar to Jeff Poskanzer s.ppm image format: Floating Point TIFF similar PF <binary image data> HDR Formats: ILM s OpenEXR (.exr) 6 bytes per pixel, for each channel, compressed sign exponent mantissa Several lossless compression options, :1 typical Compatible with the half datatype in NVidia's Cg Supported natively on GeForce FX and Quadro FX Available at: HDR Formats: Ward s LogLuv TIFF based on human color perception 4 bits:10 for log luminance 14 for chromaticity index 3 bits:15 log luminance 8 u chrominance 8 v chrominance 1 sign Larson, G.W., Overcoming Gamut and Dynamic Range Limitations in Digital Images, Proceedings of the Sixth Color Imaging Conference, November Panoramic (Omnidirectional) Photography Light Probe Images: Capturing Real-World Illumination Other techniques: - Panoramic Stitching (Realviz Stitcher) -Fisheye Images - Scanning Panoramic Cameras (Panoscan, Spheron) 004 Game Developer s Conference 4

5 March 4, 004 HDR Image of a Mirrored Ball Comparison: HDRI versus single image lighting (60,40,35) (18,17,19) (60,890,1300) Assembled from ten digital images, t = 1/4 to 1/10000 sec (5700,8400,11800) (11700,7300,600) Image-Based Lighting: Illuminating Synthetic Objects with Real Light Acquiring the Light Probe Rendering with Natural Light, SIGGRAPH 98 Assembling the Light Probe RNL Environment mapped onto interior of large cube 004 Game Developer s Conference 5

6 March 4, 004 RNL Example RNL Example Renderer Output Defocus & Glare Added RNL Example RNL Example Soft Focus Added Light Falloff (Vignetting) Added Real-Time RNL Jason Mitchell, John Isidoro, Alex Vlachos Rendered in Real Time on ATI RADEON Game Developer s Conference 6

7 March 4, 004 HDR Rendering Process Building the Scene Scene Geometry lit with HDR Light Probes HDR Scene Bloom Filter + Tone Map Displayable Image Render reflected scene into HDR planar reflection map for table top HDR light probe for distant environment HDR environment maps for local reflections from balls on pedestals Postprocess to get glows Tone map to displayable image Local Reflection Distant HDR Light probe is always sampled with reflection vector in pixel shader Local environment map is sampled with a blend of the surface normal (N) and the reflection vector (R) 100% R Blend between R and N 100% N Up Frame Postprocessing HDR Scene ¼ Size Frame Horizontal 3-Gaussian Filters Vertical 3- Gaussian Filters Filter 50x50 pixel region with sum of three Gaussians Gaussians are σ=, σ=6 and σ=14 + Tone Map Displayable Image Real Time Tone Mapping Masaki Kawase DirectX 9 Demo Very Underexposed Underexposed Good exposure Overexposed Game Developer s Conference 7

8 March 4, 004 Clipped Light Probe (sun too bright) Capturing Light Probes in the Sun How bright is the sun? Radius = 695,000 km Distance = 149,600,000 km => degrees in diameter seen from earth = radians radius 1/ = 46,334 times brighter than white Can we recover the sun? + α + α Shoot Diffuse Sphere Solve for Sun Scaling Factor + α Diffuse Ball α α = (1.166, 0.973, 0.701) 004 Game Developer s Conference 8

9 March 4, 004 Verify composite probe matches diffuse ball + = Lit with Sun Lit with Probe Real Diffuse = Real Diffuse Rendered Diffuse Avg. Error (0.5%, 0.3%, 0.%) RMS Error = (.%, 1.8%, 1.3%) Background plate Synthetic objects added Lighting Entire Environments with Outdoor Light Probes Rendered in Arnold by Marcos Fajardo. 004 Game Developer s Conference 9

10 March 4, Samples per Pixel 6h, min Rendered in Arnold by Marcos Fajardo. Rendered in Arnold by Marcos Fajardo. 4 Samples per Pixel 6h, min Making High Dynamic Range Lighting Efficient Rendering Light Probes as Light Sources LightGen by Jon Cohen et al. at Supports Maya, RADIANCE, Mental Ray, Lightwave Game Developer s Conference 10

11 March 4, 004 Structured Importance Sampling of Environment Maps Importance Sampling 3000 samples Noisy and slow! Sameer Agarwal Henrik Wann Jensen Serge Belongie Ravi Ramamoorthi Structured Importance Sampling of Environment Maps Sameer Agarwal Serge Belongie Henrik Wann Jensen Ravi Ramamoorthi Structured Importance Sampling 300 samples Yay! Approximating Environments Step 1) Partition into regions of increasing brightness Step ) Use Hochbaum-Schmoys Algorithm to place samples in the brightest region(s) Step 3) Repeat for the next brightest region, but make sure you consider the samples you added above first Step 4) Repeat until you ve covered the whole environment. Light Stage 1.0 The Light Stage: 60-second exposure Debevec, Hawkins, Tchou, Duiker, Sarokin, and Sagar. Acquiring the Reflectance Field of a Human Face. SIGGRAPH Game Developer s Conference 11

12 High Dynamic Range Lighting Light Stage 4D Reflectance Field Light Stage Results March 4, 004 Modulated Images Lighting Reflectance Functions 1 normalized light map reflectance function lighting product Environments from the Light Probe Image Gallery rendered pixel 1 DCT Basis Smith Smith and and Rowe. Rowe. Compressed Compressed domain domain processing processing of of JPEG-encoded JPEG-encoded images. images Real-Time IBL with Spherical Harmonics Frequency Space Environment Map Rendering Ravi Ramamoorthi, Pat Hanrahan, SIGGRAPH00 Interactive Interactive Lighting Lighting Demo Demo Chris Chris Tchou, Tchou, Dan Dan Maas Maas SIGGRAPH SIGGRAPH Creative Creative Applications Applications Laboratory Laboratory Game Developer s Conference Precomputed Radiance Transfer for Real-Time Rendering in Dynamic, Low-Frequency Lighting Environments Peter-Pike Sloan, Jan Kautz, John Snyder, SIGGRAPH00 1

13 High Dynamic Range Lighting Real-time IBL Techniques for Complex BRDFs Ramamoorthi and Hanrahan, March 4, 004 Real-time IBL Techniques P.-P. Sloan, J. Kautz, J. Snyder, Precomputed Radiance Transfer for Real-Time Rendering in Dynamic, Low-Frequency Lighting Environments, SIGGRAPH 00 An Efficient Representation for Irradiance Environment Maps, Siggraph 001. Ramamoorthi and Hanrahan, Frequency Space Environment Map Rendering, Siggraph 00. Excellent Overview: Robin Green, Spherical Harmonic Lighting: The Gritty Details, GDC 003. All-Frequency Shadows Using Non-linear Wavelet Lighting Approximation Ren Ng Stanford University Ravi Ramamoorthi Columbia University Pat Hanrahan Stanford University Non-linear Lighting Approximation All frequencies! D Harr transform - orthonormal basis Approx. lighting L (EM) in (non-linear) wavelet basis Weighting (error minimization) Unweighted Transport weighted Area weighted Light transport T as sparse matrix B = TL (sparse matrix-vector mult.) Better than spherical harmonics! - blurred lighting - soft shadows High energy lights (> 104) SH (100) Wavelets (100) Low energy lights (< 10) Further investigation required! Weighting scheme Spherical wavelets Image-Relighting Comparison IMAGE-BASED LIGHTING IN FIAT LUX Paul Debevec, Tim Hawkins, Westley Sarokin, H. P. Duiker, Christine Cheng, Tal Garfinkel, Jenny Huang SIGGRAPH 99 Electronic Theater 004 Game Developer s Conference 13

14 March 4, 004 Assembled Panorama Light Probe Images Lighting Calculation Impostor light sources Interior of St. Peter s reconstructed from one viewpoint Debevec, Taylor, and Malik. Modeling and Rendering Architecture from Photographs. SIGGRAPH 96 Renderings made with Radiance: HDR Lighting Real-World Reflectance Properties Recovered Ward Model Reflectance Parameters ρ d ρs α Gardner, Tchou, Hawkins, and Debevec SIGGRAPH 003 ρ t N D 004 Game Developer s Conference 14

15 March 4, 004 The Ward Model with Translucency (Single Light Source) δ exp tan I cosθ + ( + ) + α i ρ d cosθi ρt cos θi π ρs π cosθ 4α r Note: all cosines are clamped to be non-negative In terms of vectors I ρd π Where: δ = cos ρ ρ exp tan ρ ρ ρ ρ L N + ρ ρ ρ t s V N 4α ( L N ) + ρ ( L N ) ( L + V N ) = cos ( H N ) 1 1 = half angle δ α Interpolated / Constant Values I ρd π ρ ρ exp tan ρ ρ ρ ρ L N + ρ ρ ρ t s V N 4α ( L N ) + ρ ( L N ) I = light intensity L = light direction vector V = view (camera) direction vector H = half angle vector δ α HDR Texture Maps I ρd π ρ ρ exp tan ρ ρ ρ ρ L N + ρ ρ ρ t s V N 4α ( L N ) + ρ ( L N ) ρ d ρ t ρ s α N = Diffuse reflectance (RGB) = Translucent transmission (RGB) = Specular reflectance (RGB) = Specular roughness (A) = Surface normal (XYZ) δ α Gaussian Specular Lobe Table LLS Real-Time Demo I ρd π ρ ρ exp tan ρ ρ ρ ρ L N + ρ ρ ρ t s V N 4α ( L N ) + ρ ( L N ) δ α Stored in a texture map indexed by ( α, H N ) 004 Game Developer s Conference 15

16 March 4, 004 High Dynamic Range Display System High Dynamic Range Display System Emerging Technologies Helge Seetzen Lorne Whitehead Dept. of Physics of Astronomy University of British Columbia Wolfgang Stuerzlinger Andrejs Vorozcovs Dept. of Computer Science York University Greg Ward AnyHere Consulting Thanks! Jason Mitchell, Chris Brennan, Masaki Kawase Chris Tchou, Andrew Gardner, Tim Hawkins, H.P. Duiker, Westley Sarokin Sponsors: National Science Foundation, Interactive Pictures Corporation, Interval Research Corporation, the US Army, TOPPAN Printing Co. Inc., and the University of Southern California Game Developer s Conference 16