Recovering High Dynamic Range Radiance Maps from Photographs

1
Recovering High Dynamic Range Radiance Maps from
Photographs

• Paul E. Debevec, Jitendra Malik. In SIGGRAPH 97,
  August 1997

2
Introduction

• Dynamic Range of a scene is the contrast ratio
  (brightest / darkest parts)
• Recovering High Dynamic Range Radiance Maps
• Multiple Photographs
• Recover Response function ( film response in
  exposure )
• Differently Exposed Photographs
• Image-Based Modeling Rendering
• Image same ( exposure setting , film response
  function )
• Recovering reflection models ( BRDF )
• require absolute Radiance Values

3
Image Acquisition Pipeline
4
Film Response Recovery

• The response of a film
• Optical Density of film against the Exposure
• digital number Z ( development , scanning ,
  digitization process )
• Input Z ,

5
Film Response Recovery

• To complete description
• - ( unit exposure)
• - weighting function
• - pixel location ( Zmin Zmax , image )

6
(No Transcript)
7
Constructing the HDR Radiance Maps

• Curve - Radiance values ( associated with g )

Combining the multiple exposures - reduce Noise,
artifacts ( such as film grain )
8
Constructing the HDR Radiance Map

• Storage
• Map ? image format

9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
Rendering Synthetic Objects into Real Scene
Bridging Traditional and Image-Based Graphics
with Global Illumination and High Dynamic Range
Photography

• Paul E. Debevec,. In SIGGRAPH 98, July 1998

13
Introduction

• Scene Radiance Global Illumination
• High Dynamic Range Image-based model
• Illuminate the NEW objects
• 3 components
• Distant Scene
• Local Scene
• Synthetic Objects

14
(No Transcript)
15
Introduction

• Light Probe
• HDR Panoramic Radiance Map ( near rendered
  location )
• Light-Based Model

16
Illuminating Synthetic Objects with Real Light

• Accurately Recording Light in a scene
• Large areas of Indirect Light
• Concentrated areas of Direct Light
• Recovering HDR
• Measure of Scene Radiance

17
(No Transcript)
18
(No Transcript)
19
The General Method

• Distant Scene ( a Light-Based Model )
• Light-Based Model
• Accurate measure of Incident Illumination
• ( in vicinity of objects, desired viewpoint )
• Local Scene ( approximate Material-Based Model )
• Interact with the Synthetic Objects
• Geometry, Reflectance characteristic
• Synthetic Object
• variety of Shapes , Materials

20
Compositing using a Light Probe

• Constructing a Light-Based Model of real scene
• Fully Dynamic Range Omnidirectional Radiance Map
• Radiance measurement ? mapped ( geometry of
  Distant Scene )
• Mapping from the Probe to the scene model
• Mapping coordinates ( ball , world )
• Position, Size, camera Parameter ( location in
  scene, focal length )
• Assume Small , Orthograph ? good approximation
• Creating Rendering

21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)