clintonbrennan.com

What It Is

This article on clintonbrennan.com is a personal technical blog post about “Simple 2D Deterministic Physics Simulation.” It is not a complete commercial product, but rather a write-up of the author’s experience implementing custom deterministic physics for a Unity3D RTS lockstep networking scenario. The article notes that Unity’s PhysX can produce non-deterministic results across platforms due to differences in floating-point handling and compiler optimizations, so the author chose to implement a simple 2D physics simulation using integer math only.

Core Features and Technical Approach

The main purpose of the article is to provide predictable collision and movement results for lockstep synchronization. The implementation is limited to the x-z plane, while rendering can still be 3D. Colliders are restricted to Circle and Rectangle types to keep complexity low. Collision handling uses Impulse Resolution and is divided into Broad Phase and Narrow Phase. In the Broad Phase, a QuadTree recursively subdivides the area, then AABB checks are used to quickly filter potential collisions. To avoid duplicate collision pairs, the author uses a CollisionBroadPhase class with custom GetHashCode and Equals implementations, storing results in a HashSet. The Narrow Phase handles three collision types: rectangle-vs-rectangle, circle-vs-circle, and rectangle-vs-circle.

Languages, Frameworks, and APIs

The article explicitly covers C#, DotNet, and Unity3D. Example code shows designs such as PhysicsEngine.GameWorldScale, PhysicsObject.Attach<T>(), GetAttached<T>(), and Update(int fixedFramesPerSecond). GameWorldScale is set to 100, using integers to represent two decimal places of precision. Overall, it is more of a reference code architecture than a formal SDK; there are no installation instructions, package management details, or complete API documentation.

Open Source, Pricing, and Documentation

The article itself is publicly available for free. The page mentions a Bitbucket source code link, but multiple commenters report that the repository has no access permission or that the source code is unavailable. The author once replied that it had been made public, but later comments still say the link is broken, so the current open-source status cannot be confirmed. As a blog post, the documentation quality is fairly good: it clearly explains the motivation, data structures, and main loop. However, it lacks a license, versioning, tests, performance metrics, and a complete sample project.

Pros, Cons, and Suitable Users

Its strengths are a clear problem definition and a direct focus on the common determinism challenges in Unity lockstep synchronization, offering practical ideas such as integer-based math, QuadTree, AABB, and impulse resolution. Its weaknesses are its very narrow scope, suitability only for simple 2D collisions, unstable source availability, no maintenance commitment, and the fact that it should not be treated as a production-grade physics library. It is suitable as a reference for Unity RTS developers, multiplayer synchronization projects, and learners studying deterministic physics. It is not suitable for teams looking for a plug-and-play solution, long-term maintenance, or full ecosystem support.

Access in China and Alternatives

The article does not provide information about access from mainland China, payment, or mirrors, so its accessibility status is unknown. Since this is a personal blog with a Bitbucket source link, actual availability may depend on the network environment. For mature alternatives, consider evaluating Unity PhysX, Box2D, or building a custom deterministic physics solution based on fixed-point or integer math.