Rendering Pipeline

Gondwana’s render path is intentionally explicit. Bitmap backbuffers use dirty-region redraw to avoid unnecessary CPU-side pixel work. GPU backbuffers, by contrast, render through a GPU-backed SKSurface, where the engine can favor full-frame redraw and let the GPU/Skia pipeline handle the heavier lifting.

---

High-level flow

Each engine cycle does two broad things:

Background work

• pre-cycle timers
• input polling
• animation advancement
• sprite movement
• collision resolution
• camera updates

Foreground work

• direct drawing updates
• render visible views into backbuffers
• present backbuffers to adapters
• post-cycle timers

---

Render path on bitmap backbuffers

1. If the scene needs a full refresh, visible world regions are enqueued for each layer
2. Each view collects dirty world rectangles and converts them to screen-space rectangles
3. Dirty screen areas are pre-cleared
4. Each visible layer redraws only drawables intersecting those dirty world regions
5. View-level direct drawings are rendered on top
6. Backbuffer dirty screen area is presented to the platform adapter

---

Render path on GPU backbuffers

GPU surfaces currently take a different path: they redraw the full viewport each GL paint. This avoids cross-thread refresh-queue races and fits the GL-thread ownership model better.

---

Why this design works

The engine separates:

• what changed (RefreshQueue)
• what is visible (View)
• what gets drawn (Backbuffer)
• how it reaches the screen (RenderSurfaceAdapterBase)

That keeps the core engine predictable and platform-agnostic.

---

Where to read next

• Gondwana/Engine.cs
• Gondwana/Rendering/RenderSurfaceHost.cs
• Gondwana/Rendering/Backbuffers/