Efficient Rendering Method for
Atmospheric Model Objects

Sungkye Lee
Department of Computer Science
Texas A&M University
terry@cs.tamu.edu

---

Project Proposal (Feb. 9, 2005)
Summary : Modeling and rendering complex volumetric natural phenomena is a difficult task. Rendering of atmospheric bodies involves modeling the complex interaction of light throughout the highly scattering medium of water and air particles. Scattering by these particles creates many well-known atmospheric optical phenomena including rainbows, halos, the corona, and clouds. Unfortunately, most systems are difficult to use, require adjustment of numerous, complex parameters, and are non-interactive. Moreover, many radiative transport approximations in computer graphics are ill-suited to render complex angularly dependent effects in the presence of multiple scattering at reasonable frame rates.
Proposal : There are several papers about these solutions showing its techniques of efficient algorithms focusing the effective communication of information through graphics rendering, modeling, abstraction, animation, and perceptualization. I'll research these recent works, compare their strength and weakness, and make some simulator. Moreover I'll try to find which method is efficiency and ideal, and, if I could, I'd like to propose an ideal technique and algorithm.
List of Goals : Project Update #1 (March 9) - Research about 10 papers and select 2~4 methods - Analyze algorithms Project Update #2 (April 6) - Analyze and compare algorithms - List strength and weakness - Make simulator Final Report (May 9) - Make a conclusion peculiarities and characteristics - Suggest a better algorithm - Propose future work Project Proposal MS-Word File

---

Project Update #1 (March 9, 2005)

Visually accurate method for rendering volumetric multi-field weather data

• 1. Summary
  They have developed a new visually accurate multi-field weather visualization system that effectively conveys finely varying atmospheric data detail, improves the assessment of weather models, aids the training of weather observers, and presents more complete information in an intuitive style. Volumetric rendering systems are useful for weather data because of their capability to show varying degrees of opacity in inhomogeneous cloud systems. Translucent lighting based on the forward dominance of cloud particles allows the user to simultaneously gain insight to inner structures, while observing the overall structure of the cloud.

  In order to render accurately the multiple particle field densities, they first translate the liquid water content of the various fields into particle densities, and scattering coef-ficients. The particles interact with light differently, and thus they utilize the multiple particle field scattering function. They use volume rendering with a volumetric light transport approximation to translate these optical properties into an image.

• 2. Visualization Methodology
  - Meteorological Data
  - Volumetric Representation of Hydrometeor Particles
  - An Optical Model for Multiple Hydrometeor Fields
  - A Physically Inspired Illumination Model for Cloud Rendering
  - Procedural Detail for Visual Accuracy
  

Hardware-accelerated method of rendering the sky

• 1. Summary
  Hardware-accelerated techniques in atmospheric modeling and rendering include imposter systems,cloud formation modeling systems, and intuitive modeling systems. The small angle approximation, which has been applied to atmospheric rendering, is effcient but discards the phase effects after the first scattering event. These hardwareaccelerated systems have increased the efficiency of atmospheric rendering systems but have not achieved the capability to render fine angularly dependent effects.

  They have developed a new system for the efficient rendering of many atmospheric optical phenomena. By creating a model based on multiple scattering phase functions for the first few scatters, they have developed a rendering system optimized for rendering the many complex optical phenomena found in the phase functions of cloud volumes. This system is capable of capturing tne angularly dependent chromatic effects, that can trouble other approaches to global illumination. The scattering behavior of water and air particles is vastly different,and, thus, they handle them separately. Aerial perspective is approximated by considering atmospheric inscattering up to halfway through the cloud as in front of the cloud, and the remaining atmospheric inscattering as behind it.

• 2. Atmospheric Data
  - Particle Properties
  - Rendering Properties
  

• 3. Rendering Atmospheric Data
  - Aerial Perspective
  • Scattering Coefficient
  • Atmospheric Transparency
  • Sample Lighting
  - Multiple Scattering Phase Function Lighting Approximation
  • Estimating Angular Distribution
  • Intensity Calculation
  • Rendering with Angular Distribution
  • Direct sunlight

Reference

• 1. Efficient Rendering of Atmospheric Phenomena
    [Kirk Rileyy, David S. Eberty, Martin Krausy, Jerry Tessendorfz, and Charles Hansenx]
  
• 2. A SYSTEM FOR REALISTIC WEATHER RENDERING
    [Kirk Rileyy, David S. Eberty, Martin Krausy, Charles Hansenx, and Jason Levit]
  
• 3. Visually Accurate Multi-Field Weather Visualization
    [Kirk Rileyy, David S. Eberty, Martin Krausy, Charles Hansenx, and Jason Levit]
  
• 4. A Real-Time Cloud Modeling, Rendering, and Animation System
    [Joshua Schpok, Joseph Simons, David S. Ebert, and Charles Hansen]
  
• 5. Procedural Techniques and Real-Time Graphics
    [David S. Ebert]
  

---

Project Update #2 (April 6)

Real-Time Cloud Rendering and Animation

• 1. Summary
  Modeling and animating complex volumetric natural phenomena, such as clouds, is a difficult task. Most systems are difficult to use, require adjustment of numerous, complex parameters, and are non-interactive. Therefore, Rendering & Perceptualization Lab in Purdue University have developed an intuitive, interactive system to artistically model, animate, and render visually convincing volumetric clouds using modern consumer graphics hardware. Their natural, high-level interface models volumetric clouds through the use of qualitative cloud attributes. The animation of the implicit skeletal structures and independent transformation of octaves of noise emulate various environmental conditions. The resulting interactive design, rendering, and animation system produces perceptually convincing volumetric cloud models that can be used in interactive systems or exported for higher quality offline rendering.

• 2. Atmospheric Data
  The atmosphere is an inhomogeneous mixture of a wide variety of particles. In order to develop a system for the accurate rendering of these particles, it is essential to first specify the particles' properties. We then translate these properties into the spatially varying scattering probabilities and phase functions required to volumetrically render the particle fields.

               Table 1. Variable and parameter descriptions
  

• 3. Rendering Properties
  One quantity of interest for calculating light transport is the extinction coefcient, bex. This is the probability, per meter, that light intersects a particle. As described by Blinn [Bli82], this can be calculated by multiplying the particle density with the extinction cross-section.

  The Rayleigh phase function for air molecules is given in Equation

  

  At a point in space, s, this is calculated as the following:

  

  Here it is the particle concentration in particles per cubic meter. For multiple particle fields, the extinction coefcients linearly combine.

  

  The optical depth, along a segment between s and s'is the integral of the extinction coeficient along the ray.

  

  The transparency, T, of the medium is then:

  

  The total phase function at a point in space is calculated by weighting each of the individual particle phase functions by their contribution to the total scattering coeficient.

  

  Now that we have a system for calculating the spatially varying properties of the particle fields, we shift our attention to a system for rendering these particles that can capture realistic illumination effects.

Rendering Simulator - SWELL

• 1. About this Program
  Swell is an interactive cloud modeling tool as outlined in the 2003 publication. By out-sourcing the evaluation of procedural techniques to latest programmable graphics hardware, they create photorealistic images at interactive rates. They continue to simplify cloud creation by devising a set of qualitatively independent rendering attributes to more intuitively adjust the final image.

  Schematic diagram of operations
  

• 2. User Interface
  They organize variables hierarchically through a tree of GLUI roll-up groups. At the top-most groups, we expose the most common and general controls, and more detailed, specific controls under successive groups. Novice users can design complete clouds using the most basic controls, while advanced users can customize properties deeper in the tree.

  Schematic diagram of operations
  

  The interface hierarchically organizes the controls to expose the most common first, and more specific customizations in successive levels.

• 3. Simulator Download
  

  Win32 Binary Download

Reference

• 1. Efficient Rendering of Atmospheric Phenomena
    [Kirk Rileyy, David S. Eberty, Martin Krausy, Jerry Tessendorfz, and Charles Hansenx]
  
• 2. A SYSTEM FOR REALISTIC WEATHER RENDERING
    [Kirk Rileyy, David S. Eberty, Martin Krausy, Charles Hansenx, and Jason Levit]
  
• 3. Visually Accurate Multi-Field Weather Visualization
    [Kirk Rileyy, David S. Eberty, Martin Krausy, Charles Hansenx, and Jason Levit]
  
• 4. A Real-Time Cloud Modeling, Rendering, and Animation System
    [Joshua Schpok, Joseph Simons, David S. Ebert, and Charles Hansen]
  
• 5. Procedural Techniques and Real-Time Graphics
    [David S. Ebert]
  

---

Final Report (May 9)

Real-Time Cloud Modeling, Rendering and Animation System

• 1. Summary

  Modeling and animating complex volumetric natural phenomena, such as clouds, is a difficult task. Most systems are difficult to use, require adjustment of numerous, complex parameters, and are non-interactive. Therefore, many scientists have developed an intuitive, interactive system to artistically model, animate, and render visually convincing volumetric clouds using modern consumer graphics hardware. Their natural, high-level interface models volumetric clouds through the use of qualitative cloud attributes. The animation of the implicit skeletal structures and independent transformation of octaves of noise emulate various environmental conditions. The resulting interactive design, rendering, and animation system produces perceptually convincing volumetric cloud models that can be used in interactive systems or exported for higher quality offline rendering.

• 2. Previous work
  • (1) Vis5d (Hibbard & Sante, 1986)
    They provided popular tool for multi-variable visualization of weather.

  • (2) IBM (Treinish, 1997), Georgia Tech (Tianyue et al. , 2001)
    They have done significant work in weather visualization. While these are all very useful tools, they do not create a visually accurate representation.

  • (3)(Blinn, 1982), (Kajiya & Von Herzen, 1984), (Max, 1995), (Klassen, 1987), (Nishita et al. , 1996), (Preetham et al. , 1999), (Dobashi et al. , 2002), (Harris & Lastra, 2001), (Stam, 1995)
    These systems have described the rendering of atmospheric bodies, but do not handle the need for multi-field, accurate renderings at interactive frame rates.

  • (4) (Drebin et al. , 1988)
    They researched earlier in volume rendering.

  • (5) (Ebert & Parent, 1990), (Engel et al. , 2001), (Kniss et al. , 2003), (Jensen & Christensen, 1998)
    From early work in (Drebin et al. , 1988) to more recent extensions in gaseous and hardware accelerated rendering in them.

  • (6)(Perlin & Hoffert, 1989)
    They provides a useful framework for the addition of simulated detail.

  • (7)(Stam & Fiume,1995), (Ebert et al. , 2003)
    They use warped blobs, and cover many different systems for adding simulated detail.

• 3. Description of work
  • (1) Researched this area
    - Research about 10 papers and select 2~4 methods
    - Read and understand thier process

  • (2)Analyze algorithm, and its implemenation
    - Analyze and compare algorithms
    - List strength and weakness

  • (3)Run simulator and get some results
    - Make a conclusion peculiarities and characteristics
    - Suggest a better algorithm

• 4. Simulating Results
  

  (1) Dirtiness (5.00625% -> 79.5125%)

  

  (2) Sharpness (3.0% -> 30.394%)

  

  (3) Detail (0.0% -> 44.0%)

  

• 5. Analysis of work
  
  • Meeting goals
    I was planning to research these recent works, compare their strength and weakness, and make some simulator. Moreover I'd like to propose an ideal technique and algorithm. However, this topic was obscure, and it involves so many parts or details. Also I tried to make or upgrade a simulator which renders particles based on its optical properties, but I couldn't, because it's very huge system and they doesn't provice its source code. I was just satisfied with being understand how it works, with running simulator, and with getting some results.

  • Future work
    
    • By implementing better characterizations of the particle sizes and improved particle scattering data, the system will produce more accurate images.
      
    • Extending the system to interact with extra variables in the simulation will also increase its usability.
      
    • Improving the light transport approximation will also make more realistic representations of the data.
      
    • Allowing the cloud models to interact with the sky produces spatial visual cues and improve sense of scale.
      
    • Improved map projections allow application of the system to a wider range of user data.
      

Reference

• 1. Efficient Rendering of Atmospheric Phenomena
    [Kirk Rileyy, David S. Eberty, Martin Krausy, Jerry Tessendorfz, and Charles Hansenx]
  
• 2. A SYSTEM FOR REALISTIC WEATHER RENDERING
    [Kirk Rileyy, David S. Eberty, Martin Krausy, Charles Hansenx, and Jason Levit]
  
• 3. Visually Accurate Multi-Field Weather Visualization
    [Kirk Rileyy, David S. Eberty, Martin Krausy, Charles Hansenx, and Jason Levit]
  
• 4. A Real-Time Cloud Modeling, Rendering, and Animation System
    [Joshua Schpok, Joseph Simons, David S. Ebert, and Charles Hansen]
  
• 5. Procedural Techniques and Real-Time Graphics
    [David S. Ebert]
  

---

Coprywright ©  2005 Sungkye Lee