---
title: "Toy Car Rendering"
canonical: "https://rmanwiki-27.pixar.com/space/RU/654671935/Toy%20Car%20Rendering"
format: markdown
---
### Matching Footage and Rendering

Before starting the rendering process in Maya, I first needed to match the object’s rotation to the real turntable footage. Using motion tracking in Nuke, I applied a bandpass filter to isolate the platform's movement and manually tracked key points to identify its rotation axis, though the data was limited to 2D space.

After importing the data into Maya, I manually adjusted the scale and orientation to align the physical turntable with the digital scene. By projecting the 2D tracks onto a 3D plane, I was able to accurately synchronize the object’s rotation in CG with the filmed footage.

![image9.jpg](media://75f83376-62ea-4fdf-8a9d-64eacc1a925a)


After aligning the CG object with the footage, I prepared the scene for the final render. To capture the character’s shadow on the ground without affecting the alpha, I used a Holdout plane. This made it easier to integrate the render into the background plate.

Before rendering, I ran several test frames to ensure that depth of field (DOF) and motion blur were behaving correctly, by comparing the sharpness with the reference footage. The final render was done using RenderMan RIS.

For compositing flexibility, I exported a set of AOVs using Light Path Expressions (LPE), including diffuse direct, diffuse indirect, specular direct, and specular indirect. I also used Cryptomatte for fast and accurate masking during compositing.

To reduce sampling noise, I enabled the RenderMan Machine Learning Denoiser, which noticeably improved the output's clarity while keeping render times low.


![image14.jpg](media://773df260-5668-4483-9e3e-a64d91c7eaf9)

![image12.jpg](media://bf2048ea-ade3-49b0-bdf2-73f20cb89dce)

---