# Roblox Solved The Physics Problem That Stumped Everyone!

Source: https://www.youtube.com/watch?v=TzIKbjuSy2A
Recap page: https://rapidrecap.app/video/TzIKbjuSy2A
Generated: 2025-07-13T20:48:39.384+00:00

---
## Quick Overview

The new Augmented Vertex Block Descent (AVBD) physics simulation method, developed by Roblox and the University of Utah, solves long-standing problems in physics simulations by accurately handling high mass ratios, friction, and stiffness, delivering stable and correct results faster than real-time, even in scenarios where previous methods fail.

**Key Points:**
- The new Augmented Vertex Block Descent (AVBD) method simulates complex physics scenes faster than real-time, achieving 100 frames per second on a consumer GPU.
- AVBD correctly simulates scenarios with high mass ratios, such as a heavy pendulum on a light chain or a Tesla Model S hanging from Lego bricks, where previous methods like VBD and XPBD fail due to excessive stretching or instability.
- The method accurately handles friction, allowing objects to behave realistically on inclined planes and preventing incorrect sticking or sliding seen in older techniques.
- AVBD maintains stability and accuracy even with fewer iterations, making it highly efficient for real-time applications in games and movies.
- It resolves issues with alternating stiffness in connected bodies, producing accurate sagging behavior in spring simulations where previous methods overstretch.
- The new technique correctly simulates collisions and material deformation, as demonstrated by a ball breaking through a brick wall and a chainmail holding a heavy ball, unlike prior methods that show incorrect penetration or disintegration.
- The AVBD method is open-source and freely available, allowing anyone to experiment with its capabilities.

**Context:** Physics simulations are crucial for realistic graphics in games and movies, but accurately modeling complex interactions like collisions, friction, and material deformation, especially with varying object properties, has been a long-standing challenge. Previous methods, such as Vertex Block Descent (VBD) and XPBD, often struggled with stability and accuracy in scenarios involving high mass ratios or extreme stiffness differences, leading to unrealistic visual artifacts.

## Detailed Analysis

The video introduces Augmented Vertex Block Descent (AVBD), a groundbreaking physics simulation method developed through a collaboration between Roblox and the University of Utah. This new technique addresses critical limitations of previous methods like Vertex Block Descent (VBD) and XPBD, which often produced incorrect or unstable results in complex scenarios. AVBD achieves remarkable speed, simulating scenes with over 110,000 blocks at 100 frames per second on a single consumer GPU, significantly faster than real-time. Its core innovation lies in an augmented Lagrangian formulation that gradually adjusts the "strictness" of constraint enforcement, akin to a bouncer adapting their force to maintain order. This adaptive approach allows AVBD to accurately handle challenging situations such as high mass ratios (e.g., a heavy pendulum on a light chain or a car suspended by Lego bricks), where older methods exhibit excessive stretching or complete breakdown. Furthermore, AVBD correctly simulates friction and stiffness, preventing issues like objects incorrectly sticking or overstretching in spring simulations. The method's robustness and accuracy, even with fewer iterations, make it ideal for real-time applications in games and movies, offering a significant leap forward in realistic physical interactions. The source code is open-source and freely available, encouraging widespread adoption and experimentation.

### Introduction to AVBD

- A new, free, and open-source physics simulation method developed by Roblox and the University of Utah, capable of solving previously impossible problems in real-time.

### Performance and Scale

- AVBD simulates complex scenes with 110,000 blocks at 100 frames per second on a consumer GPU, demonstrating performance faster than real-time
- The previous VBD method could simulate incredible scenes but had limitations with millions of points.

### Limitations of Previous Methods

- Older methods like VBD and XPBD fail on seemingly simple cases, such as a house of cards collapsing incorrectly due to excessive friction
- They also fail to accurately simulate a heavy pendulum on a light chain, causing the chain to stretch unrealistically
- Previous methods incorrectly simulate a ball breaking through a chainmail, failing to hold it up as expected.

### AVBD's Superior Accuracy

- AVBD correctly simulates the house of cards collapsing and the heavy pendulum swinging without stretching, even with fewer iterations
- It accurately simulates the chainmail holding the ball, demonstrating correct collision constraints
- AVBD also correctly simulates a stack of blocks on a ball and a ball breaking through a brick wall, unlike previous methods that show instability or incorrect behavior.

### Technical Innovation

- AVBD uses an augmented Lagrangian formulation that gradually adjusts the "strictness" of enforcing physical rules during simulation, preventing initial rigidity and improving stability
- This adaptive enforcement is crucial for handling high mass ratios and stiffness differences, which were problematic for prior techniques.

### Accessibility and Impact

- The AVBD method is open-source and freely available, allowing developers and researchers to utilize and experiment with this advanced physics engine
- This represents a significant contribution to humanity, enabling more realistic and stable physics in games, movies, and other virtual environments.

![Screenshot at 0:00: A large pile of colorful blocks on a white sheet, demonstrating a complex simulation scene](https://ss.rapidrecap.app/screens/TzIKbjuSy2A/00-00-00.png)
![Screenshot at 0:09: A pink ball impacting a wall made of colorful blocks, showing the destruction simulated by the new method](https://ss.rapidrecap.app/screens/TzIKbjuSy2A/00-00-09.png)
![Screenshot at 0:24: A large circular structure made of colorful blocks being impacted by a pink ball, demonstrating the scale and detail of the simulation](https://ss.rapidrecap.app/screens/TzIKbjuSy2A/00-00-24.png)
![Screenshot at 0:44: A green armadillo-like creature with a long nose, demonstrating soft body dynamics simulation](https://ss.rapidrecap.app/screens/TzIKbjuSy2A/00-00-44.png)
![Screenshot at 0:53: Pink spiky balls filling a glass teapot, illustrating the simulation of many interacting soft bodies](https://ss.rapidrecap.app/screens/TzIKbjuSy2A/00-00-53.png)
![Screenshot at 1:10: Four purple octopus-like figures in boxes, showing experiments with different friction coefficients](https://ss.rapidrecap.app/screens/TzIKbjuSy2A/00-01-10.png)
![Screenshot at 1:34: Comparison of old and new methods for simulating a spiky ball, highlighting improved accuracy and stability](https://ss.rapidrecap.app/screens/TzIKbjuSy2A/00-01-34.png)
![Screenshot at 1:46: Side-by-side comparison of previous (VBD) and new (AVBD) methods showing a house of cards collapsing correctly with AVBD](https://ss.rapidrecap.app/screens/TzIKbjuSy2A/00-01-46.png)
![Screenshot at 2:15: Comparison of pendulum simulations, showing previous methods failing (stretching) while AVBD remains stable](https://ss.rapidrecap.app/screens/TzIKbjuSy2A/00-02-15.png)
![Screenshot at 2:34: Four-panel comparison of different methods simulating a ball hitting a suspended bar, showing AVBD's superior stability](https://ss.rapidrecap.app/screens/TzIKbjuSy2A/00-02-34.png)