**Chloe Son Hanlin Wang Final Project Report**
(#) Chloe Son (F00361v) & Hanlin Wang (F003615))
This scene from Hugo captures the cinematic, steampunk quality we want to convey with our scene environment.
Under the theme of *"It's what's inside that counts"*, our team has designed a concept within a grand clock tower, where a mustached robot ardently operates the giant hands of the clock. In doing so, he counts the passage of time.
To bring this scene to life, we augmented our renderer to showcase sunlight pouring into the tower through the glass clock face as well as brushed metal materials to fortify our steampunk aesthetic.
*Below is our concept sketch, utilized as reference for our self-made 3D models and feature design:*
(#) New Features of Our Renderer
(##) Directional Light Source *(4)*
*Description:*
* We desired to create a scene that was wholly dependent on directional light rather than background light for a dramatic lighting effect. We did so by creating an enclosed space (the clock tower interior) for our scene, cutting out a window into the outside where the background was set as 0, and placed a diffuse light outside the window to cast light through the window and into the room so as to cast sun beams and shadows into the clock tower.
*Validation Tests:*
* **NormalsIntegrator Test:** To test the directional light source, we first rendered out our scene and the light using the NormalsIntegrator, which allowed us to very clearly see how the light was shading the surfaces.
* **Custom Scene for Visual Testing:** Another test we conducted was modeling a custom scene that had gaps in the ceiling to observe how our Integrator allowed light to shine into the room from outside.
*Encountered Problems:*
* We struggled with rendering the wall onto which our clock face was attached. When the wall was present in the scene, it would not render correctly and behave as if it were not there, but the light would also be occluded by it, as well. We solved this issue by bringing the wall into Autodesk Maya and realizing that our renderer could not interpret the geometry due to the non-manifold nature of the mesh because we had cut out a hole in the wall so light could go through that part of the wall and the clock face. We fixed the issue by using the "Quad draw" tool in Maya to re-topologize the wall to consist of quad polygons.
(##) Anisotropic Microfacet BRDF *(4)*
*Description:*
* We implemented a rough metal material based on a microfacet BRDF. To statistically stimulate the apperance of reflective surfaces
with microscopic ridges and facets, we defined a Beckmann distribution function which gives the differential area of microfacets w.r.t the surface normal w_h
(the halfway vector between incident and reflected rays). The distribution supports anisotropy by varying the roughness_x and roughness_y, which becomes isotropic when the two terms are equal.
*Validation Tests:*
Comparison between Microfacet material and regular Metal material (same roughness = 0.2). Note the stronger highlights and darker shadows of the microfacet material.
*Encountered Problems:*
* Understanding the BRDFs involved proved to be difficult due to PBRT's involved BRDF functions which attempt to map real world microfacet behavior.
Attempting to understand the calculations and their geometric meanings proved to be challenging, but simplifying our implementation to borrow Metal'
sampling method (perfect reflection randomly offset according to a roughness parameter) proved to give us a very convincing result.
(##) God Rays via Homogeneous Participating Media *(8 | Advanced Technique)*
*Description:*
* Our homogenerous participating media permeates throughout our spaces because it is meant to simulate interactions with particles in atmosphere.
*Validation Tests:*
* **Adjusting Absorption and Scattering Coefficients** - *visually test the efficacy of our homogeneous medium through setting the absorption sigma to 0 and scattering sigma to 1 in order to attain a Lambertian effect and absorption to 0 and scattering to 0 to attain black.*
*Encountered Problems:*
* In testing our homogeneous medium with various test scenes, we observed that certain scenes would render well and others wouldn't. In trying to discern why this variation in results occurred, we examined jsons and realized through trial and error that we had to accordingly **adjust our sigma values to accommodate for different scales of geometry.** For example, with scenes involving larger scaled rooms, we greatly decreased our values for coefficients to reduce our sigma_t. We also decreased our number of bounces in our scene to limit the amount of times the light bounces before reaching the eye. Doing so produced a cleaner lighting effect.
In terms of external libraries, we developed our code in credit to the learnings provided by Professor Jarosz's slides and the PBRT textbook.
**Bonus Work as Digital Artists**
In order to accurately achieve our concept vision, we utilized our 3D modeling skills to model the clock interior and robot and rig the robot in order to pose it. In order to optimize the performance of our renderer, we sought a balance between minimizing polygon count, yet maximizing detail in our 3D geometry.
(#) Final Image Showcase
*Thank you Prof. Jarosz and Shaojie for instructing and helping us in our ray-tracing education!*
This scene from Hugo captures the cinematic, steampunk quality we want to convey with our scene environment.
Under the theme of *"It's what's inside that counts"*, our team has designed a concept within a grand clock tower, where a mustached robot ardently operates the giant hands of the clock. In doing so, he counts the passage of time.
To bring this scene to life, we augmented our renderer to showcase sunlight pouring into the tower through the glass clock face as well as brushed metal materials to fortify our steampunk aesthetic.
*Below is our concept sketch, utilized as reference for our self-made 3D models and feature design:*
(#) New Features of Our Renderer
(##) Directional Light Source *(4)*
*Description:*
* We desired to create a scene that was wholly dependent on directional light rather than background light for a dramatic lighting effect. We did so by creating an enclosed space (the clock tower interior) for our scene, cutting out a window into the outside where the background was set as 0, and placed a diffuse light outside the window to cast light through the window and into the room so as to cast sun beams and shadows into the clock tower.
*Validation Tests:*
* **NormalsIntegrator Test:** To test the directional light source, we first rendered out our scene and the light using the NormalsIntegrator, which allowed us to very clearly see how the light was shading the surfaces.
* **Custom Scene for Visual Testing:** Another test we conducted was modeling a custom scene that had gaps in the ceiling to observe how our Integrator allowed light to shine into the room from outside.
*Encountered Problems:*
* We struggled with rendering the wall onto which our clock face was attached. When the wall was present in the scene, it would not render correctly and behave as if it were not there, but the light would also be occluded by it, as well. We solved this issue by bringing the wall into Autodesk Maya and realizing that our renderer could not interpret the geometry due to the non-manifold nature of the mesh because we had cut out a hole in the wall so light could go through that part of the wall and the clock face. We fixed the issue by using the "Quad draw" tool in Maya to re-topologize the wall to consist of quad polygons.
(##) Anisotropic Microfacet BRDF *(4)*
*Description:*
* We implemented a rough metal material based on a microfacet BRDF. To statistically stimulate the apperance of reflective surfaces
with microscopic ridges and facets, we defined a Beckmann distribution function which gives the differential area of microfacets w.r.t the surface normal w_h
(the halfway vector between incident and reflected rays). The distribution supports anisotropy by varying the roughness_x and roughness_y, which becomes isotropic when the two terms are equal.
*Validation Tests:*
*Description:*
* Our homogenerous participating media permeates throughout our spaces because it is meant to simulate interactions with particles in atmosphere.
*Validation Tests:*
* **Adjusting Absorption and Scattering Coefficients** - *visually test the efficacy of our homogeneous medium through setting the absorption sigma to 0 and scattering sigma to 1 in order to attain a Lambertian effect and absorption to 0 and scattering to 0 to attain black.*
*Encountered Problems:*
* In testing our homogeneous medium with various test scenes, we observed that certain scenes would render well and others wouldn't. In trying to discern why this variation in results occurred, we examined jsons and realized through trial and error that we had to accordingly **adjust our sigma values to accommodate for different scales of geometry.** For example, with scenes involving larger scaled rooms, we greatly decreased our values for coefficients to reduce our sigma_t. We also decreased our number of bounces in our scene to limit the amount of times the light bounces before reaching the eye. Doing so produced a cleaner lighting effect.
In terms of external libraries, we developed our code in credit to the learnings provided by Professor Jarosz's slides and the PBRT textbook.
**Bonus Work as Digital Artists**
In order to accurately achieve our concept vision, we utilized our 3D modeling skills to model the clock interior and robot and rig the robot in order to pose it. In order to optimize the performance of our renderer, we sought a balance between minimizing polygon count, yet maximizing detail in our 3D geometry.
(#) Final Image Showcase
*Thank you Prof. Jarosz and Shaojie for instructing and helping us in our ray-tracing education!*