NCSOFT has revealed details of Project G, an ambitious large-scale strategy game developed with Unreal Engine 5. Its main technical challenge is rendering thousands of units on screen while maintaining high individual fidelity, an achievement that redefines the limits of the RTS genre. To achieve this, the studio has implemented a hybrid pipeline that combines digital sculpting tools with massive effects systems.
Technical Pipeline: From ZBrush to Houdini for Digital Warfare 🛠️
The production flow begins in ZBrush, where artists sculpt high-resolution models for each unit, guaranteeing minute details even on the smallest soldiers. These assets go to Maya for retopology and rigging creation, optimizing geometry for real-time. The real trick lies in Houdini, which generates massive VFX systems (explosions, smoke, and destruction) that integrate directly into the engine. To manage the graphical load, the team implements a dynamic LOD system and GPU instancing, where each unit shares reduced vertex buffers. Additionally, frustum culling and hierarchical occlusion allow UE5 to instantly discard objects outside the player's field of view, maintaining a stable framerate even with 10,000 active entities.
The Fidelity Dilemma at Massive Scale ⚔️
The paradox of Project G is that each soldier must look unique from a close camera, yet be indistinguishable at a distance. The solution involves using atlas textures and parameterized materials, which allow color and wear variations without multiplying draw calls. Although the engine handles the load with Nanite and Lumen, the real challenge is not technical but design-oriented: convincing the player that each unit has individual value amidst digital chaos. Project G demonstrates that, in the next generation of strategy, scale is not at odds with detail.
As a developer, what specific technical challenges does the implementation of thousands of simultaneous units in Unreal Engine 5 present, and what solutions has NCSOFT adopted to maintain stable performance in Project G?
(PS: shaders are like mayonnaise: if they break, you start all over again)