Quady3D — Treasure-room Blender and Unity study Open treasure-room.blend in Blender 4.5.10 LTS or newer. The GLB is a separate export of the scene geometry and materials. This is a manually assembled static scene, using an existing Quady3D cat-chest model generated from text with Hunyuan3D 3.0. The original source model is unchanged. The working scene uses meters. Floor and wall modules use a 1 m grid. Chest width was chosen as 1.15 m for this layout; it is not a measured reconstruction. Two chest objects share one mesh. The working chest texture is 1024 x 1024; the source texture was 4096 x 4096. Walls, floor and layout were made in Blender. Saved Quady3D rusted-iron PBR maps were manually connected to wall trim. Height was not used for displacement. The Blender file includes the presentation camera, lights and studio backdrop. The GLB contains 44 mesh objects and 15,422 triangles including repeated objects. It does not contain gameplay, colliders or LODs. The prepared room GLB was imported with glTFast 6.14.0 and run in Unity 2022.3.62f3c1, using the Built-in pipeline, Linear color space and Metal. Six materials loaded, the 1 m floor spacing and 1.15 m chest width were preserved, and five separately added box colliders passed ray checks and a 2 m capsule drop. These colliders are not included in this GLB or Blender download. A native ARM64 Development build on an Apple M5 with 16 GB RAM at 1280 x 800 was sampled for 15 seconds after a 3-second warm-up, with no focus loss and VSync off. Average frame time was 1.93 ms, P95 was 4.49 ms, average draw calls were 180 and average SetPass calls were 66. Peak Unity allocated memory was 46.87 MiB; this is not total process memory or VRAM. GPU time was not collected. The result applies only to this small static room with its prepared 1K chest texture. It does not measure the original 4K source, other hardware, long-session stability, gameplay, animation or URP/HDRP. No Unreal, Godot or Roblox result is provided. Prepare collision, LODs and gameplay components and test a representative scene on your target device before use.