Memory Model & Dirty-Frame Gating

Details on memory allocation, double-buffered Bitmap swaps, and dirty-frame gating.

---

1. Double-Buffered Front/Back Bitmaps

To prevent screen tearing when 60 FPS video is rendering simultaneously while Unreal's Render Thread copies pixels:

• Back Buffer Bitmap: Asynchronously written to by the ImageReader hardware surface listener.
• Front Buffer Bitmap: Locked and read by Unreal's Render Thread via AndroidBitmap_lockPixels.
• Atomic Swap: When a new frame arrives, the front and back buffer pointers are atomically swapped.

---

2. Memory Footprint by Resolution

Memory allocated per active QuestWebViewComponent instance:

Resolution	Single Frame Buffer	Double-Buffer Total	GPU Texture	Total Instance RAM
720p (1280×720)	3.68 MB	7.37 MB	3.68 MB	~11.0 MB
1080p (1920×1080)	8.29 MB	16.58 MB	8.29 MB	~24.8 MB
1440p (2560×1440)	14.74 MB	29.49 MB	14.74 MB	~44.2 MB
4K (3840×2160)	33.17 MB	66.35 MB	33.17 MB	~99.5 MB

---

3. Dirty-Frame Gating Architecture

When web content is static:

1. mIsFrameDirty remains false.
2. Unreal's TickComponent immediately skips texture upload logic in 0.001 ms.
3. Result: Running 5 idle web tabs costs virtually 0% extra GPU and CPU overhead!