SEM Lung Cancer | Haritha Uppalapati

Final Renders

Render

Reference

For this project, I chose a Scanning Electron Microscope image of lung cancer cells. The image attracted me because of its strong lighting, colors, and complicated geometry. I used Nuke to darken the shadows and add a blue tint to some areas.

Basic Geometry - Visualizer Integrator

Basic Geometry

Displacement

Hair Geo Script

Basic Colors

Initially, I could not figure out how to create the hairlike geometry from the reference. I made a python script that created random, bent cylinders on the selected vertices, but this method did achieve my desired look. A C language script written by Professor Malcolm Kesson (2020) was used to create randomly rotated and sized toruses located at and aligned to each vertex normal. This added a cilia or hair like quality to the surface that did not require procedural modeling. I modeled the larger, more prominent particles.

Hair Geo Script

OSL Shaders & Shading Network

Edge OSL Shader

Edge + Directional + Alligator OSL Shaders

Edge OSL

Directional OSL

Alligator OSL

Main Surface Shading Network

Hair Shading Network

I used three main OSL shaders: the Edge, Directional, and Alligator Noise shaders. The Edge shader allowed the look that microscopic images portray similar to self luminance.

Since the occlusion integrator does not render Rman lights, the Directional shader was needed to imitate lighting from a chosen direction.

The Alligator Noise Shader was created by SideFX and can be used as a displacement to created a randomly bumpy surface which allowed the surface to look more naturalistic and blend in with the hairs.

Floating Particles Shading Network

To make the hairs look more random and natural, I added a PxrFractal to their surfaces so that they would look less smooth and have more dimension.

Edge Color Process

Directional Lighting Process

Draft Render

I learned a lot about procedural texturing, implementing osl shaders, and using python scripts to create geometry throughout this process. There are still many aspects that could be improved, but I feel that this is a good start being my first time working on this type of project. I think I was successful in portraying this SEM image and am eager to find ways to make it look more accurate.