100 Heads 2ms 👩🦱, Figma WebGPU 🖥️, Godot 4.6 dev1 🎮, Japan devs AI 🤖
⚙️ Engines & Plugins
Godot 4.6 Dev1 Lands: Bugfix Wave, XR Boost, Editor QoL
Godot 4.6 dev1 kicks off with 220 fixes and a few powerful new features, laying the groundwork for both 4.6 and the upcoming 4.5.1. Highlights include drag-and-drop @export creation in scripts, OpenXR Spatial Entities Extensions support, decoupled mouse/keyboard focus behavior, and hot-swappable editor themes. Rendering adds ellipse drawing and Forward+ ubershader variants, alongside performance and workflow tweaks across the engine. It’s a pre-release—back up your projects—and help test to stabilize both branches.
Raylib Meets Unreal: Overlay Rendering and Blueprints in One Plugin
RaylibUE is a new open source plugin that brings Raylib’s approachable C API into Unreal Engine. It exposes 50+ Raylib draw calls as Blueprint nodes and renders via an independent OpenGL overlay, remaining compatible with DX11, DX12, and Vulkan projects. Built for education and prototyping, it preserves UE input/performance and includes a command buffer for real-world pipeline practice. Check the GitHub and a starter video to dive in fast.
🎨 VFX & 3D Rendering
Mobile-Ready Venom Reveal VFX in Unity by Sakura Rabbit
Sakura Rabbit debuts a slick Venom-style reveal effect built in Unity, demoed with a Spider-Verse Gwen model and optimized for mobile. The asset is available now on the artist’s Fanbox for easy integration. The feature also highlights more tech art gems—adaptive heels, outfit transformations, a cyberpunk subway, and rendering studies—perfect references for stylized VFX and character shaders.
Ultra-Fast Strand Hair: 100 Heads in 2 ms on RTX 4090
Cem Yuksel’s team unveils a strand-based hair renderer that hits 2 ms for 100 unique 100,000-strand hairstyles on an RTX 4090 with 8× MSAA. The trick: mesh shaders, on-the-fly geometry, and LOD mixing—shrinking all 100 models to just 1.7 MB. Procedural styling and a stable coordinate system keep it animation-ready. It’s a major leap toward dense, crowd-scale hair in real-time games.
📈 Industry & Monetisation Insights
The Ad Mediation Trap: Empty Tests, Weak Networks, Weird Winners
Ad monetization can surprise you: “empty tests” on leading mediation platforms show ~4% variance, banner refresh ABs can overstate gains, and adding bidders can sometimes lower ARPDAU. Case studies also show weak networks may still boost auction competitiveness, while removing tiny-share instances can hurt performance. Even nonexistent waterfall IDs can generate “ghost” impressions. Learn how to validate results and avoid costly rollouts.
Over Half of Japanese Studios Now Use AI in Development
AI has gone mainstream in Japan’s game industry: 51% of companies use AI or genAI in development, according to CESA via Nikkei. Studios most often apply it to produce videos/images, generate stories and text, and assist programming—and 32% even leverage AI to help build game engines. The sector employs ~200,000 people with an average age of 37 and salary of ¥7.23M (~$49k), up 3% YoY. A snapshot of how rapidly AI is reshaping game production in Japan.
🚀 Performance & Scale
From 30 Minutes to 2.5: Supercharging Clustered LOD for Zorah
NVIDIA’s Zorah demo is colossal—36 GB and 1.64B triangles—and this post shows how meshoptimizer slashed clustered LOD processing from ~30 minutes to ~2.5. The gains came from sparse data paths, mesh-level load balancing, SIMD AABB math, Morton sorting, and a thread-local arena allocator. The approach now powers vk_lod_clusters, generating a 62 GB cache in ~3m20s. Bonus: Zorah renders at 26 ms (RT) on an RTX 3050 with a ~2 GB geometry pool.
Shipping WebGPU at Scale: Figma’s Rendering Migration and Performance Wins
Figma migrated from WebGL to WebGPU by redesigning its draw API, batching uniforms, and auto-translating GLSL to WGSL to keep one shader codebase. They integrated WebGPU via Emscripten and Dawn, adapted to async readbacks, and built a dynamic WebGPU→WebGL fallback for stability—especially on Windows. Performance testing across devices drove optimizations like bindGroup reuse and renderPass batching. With rollout complete, Figma is targeting compute shaders, MSAA, and RenderBundles for further speedups.