How fast is the hair/cloth simulation demonstrated in the video?

The demonstration shows up to nearly 1.5 million simulation elements running in about 7 milliseconds per frame, enabling real-time or near-real-time performance in some scenarios. Timestamp reference: 124.

What is caserat rods and why is it important?

Caserat rods is a technique that tracks every point along a branch, recording how stretched, bent, and twisted each point is to achieve stable and accurate deformations. Timestamp reference: 293.

What is the 'instant drying' brick-wall analogy, and what are its limits?

The analogy describes laying bricks at once and using a drying 'foam' that reduces waiting, enabling fast assembly with good stability for general use, but it can be too fast in complex cases where mortar-brick interactions matter. Timestamp reference: 351.

When should one prefer the old vs new methods according to the video?

For high-precision scientific engineering or surgeries, the older, slower method may be necessary. For computer games and movies, the newer approach offers strong stability and speed. Timestamp reference: 465.

Where can I learn more or access the code?

The video mentions that source code is available and points to description links for more details and access. Timestamp reference: 473.

This New Physics Engine Is 45x Faster!

Two Minute Papers

Science & Technology3 min read10 min video

Jan 7, 2026|141,879 views|7,828|359

Save to Pod

Key Moments

TL;DR

Caserat rods: 45x faster, stable real-time hair/cloth sim.

Key Insights

The Caserat Rods approach enables stable simulations with large time steps, delivering real-time-like performance for complex materials.

It handles hair, cloth, ropes, trees, and multi-material structures with high fidelity even under extreme deformations.

Core idea combines split position/rotation optimization with a closed-form orientation update to remove bottlenecks in prior methods.

Compared to the previous discrete elastic rods technique, it achieves ~45x speedups while maintaining robust stability in tough scenarios.

Trade-offs exist: in some highly precise micro-interactions, older slower methods can be more accurate; the technique is ideal for games/films; source code is openly available.

INTRODUCTION: WHY A NEW PHYSICS ENGINE MATTERS

The video opens by framing a long-standing problem in animation and simulation: making complex materials like fur, hair, cloth, or flexible structures behave realistically without manual frame-by-frame tweaking. Artists waste weeks correcting frames when the physics engines stall or drift. The new Caserat Rods approach promises stability large enough to let the artist press play and rely on physics to drive the motion. Demonstrations span hair, cloth, ropes, and bridges under extreme forces, highlighting the potential to cut production time while preserving realism.

THE CORE IDEA: CASERAT RODS AND ITS STABILITY

At the heart of the approach is a split position and rotation optimization scheme with a closed-form Gaussian quasi-static orientation update, designed to stay numerically stable even with large time steps. The analogy of building a wall shows the radical shift: laying all bricks at once and applying instant-drying foam instead of waiting for mortar to cure. This bypasses traditional sequential constraints, enabling robust, fast updates across many points along a structure, whether hair strands or branch segments.

REAL-TIME DEMONSTRATIONS: HAIR, CLOTH, AND MULTI-MATERIAL SCENARIOS

The demonstrations push the method to extremes: nearly 1.5 million vertices in a character’s hair, updated at frames around 7 milliseconds, pushing toward real-time playback. A quarter-million-vertex knitted text is simulated with convincing hang – not perfectly static, but impressively responsive. Clothing with tens of thousands of strands moves interactively under user input, and even wildly elastic multi-material assemblies (rubber, handles) deform plausibly. The shader-like results show broad coverage across textures, fibers, and dynamic interactions.

PERFORMANCE COMPARISON: 45x FASTER AND STABLE

Comparisons with the older discrete elastic rods technique show a dramatic speedup: roughly 45 times faster in the worst cases shown, with the new method maintaining similar levels of stability in challenging deformations. Even in extreme scenarios such as a category-5 hurricane on a bridge, the new approach remains numerically stable while reproducing realistic damage. The visuals suggest the bulk of the computation is done simultaneously rather than via incremental, frame-by-frame steps, enabling much higher throughput.

LIMITATIONS, TRADE-OFFS, AND USE CASES

The instant-drying foam metaphor highlights both strength and a limitation: in very specific, complex interactions like knot tightening or rods being crushed from multiple directions, the method may sacrifice some precision. For cinema and games, the difference is usually imperceptible, but for high-precision engineering or surgical simulations, the older slower methods can be more accurate. The creators emphasize choosing the right tool for the job, with the new Caserat Rods approach serving as a fast default and the legacy methods reserved for when micro-detail matters.

APPLICATIONS, ACCESSIBILITY, AND FUTURE IMPLICATIONS

The method is pitched as broadly applicable to real-time or near-real-time simulation across hair, cloth, foliage, and complex rods. Importantly, the researchers publish the source code, making it freely available for the community, which could accelerate adoption in industry and academia alike. The video also includes a plug for Lambda GPU Cloud for running large AI models, illustrating the broader trend toward affordable access to powerful hardware. Together, these pieces suggest a future where robust physics engines are commonplace in consumer workloads.

Mentioned in This Episode

●Tools & Products

●People Referenced

Cheat Sheet: practical takeaways from the video (high-level)

Practical takeaways from this episode

Do This

Highlight the main analogy: instant drying vs. traditional mortar as a way to think about time-step stability.

Use the old slow method for high-precision engineering or surgery use cases.

When possible, leverage robust new techniques for real-time media like hair and cloth simulations.

Avoid This

Rely on instant foam analogies for high-precision engineering where accuracy of interaction matters.

Ignore stability concerns when using large time steps in complex, multi-material scenarios.

Performance comparison: previous method vs new method

Data extracted from this episode

Metric	Previous method (8 years ago)	New method
Speed/efficiency	Not specified in detail	45x faster (as shown in example)
Per-frame time (new method)	Not stated for previous	Approximately 7 ms per frame for large-scale hair/strands
Notes	Typical elastic-rod approach with stability limitations	Maintains stability with large time steps in tested scenarios

Common Questions

The video presents a robust physics method capable of handling hair, cloth, and large structural deformations with large time steps, reducing the need for frame-by-frame manual fixes in movies and games. Timestamp reference: 32.