SUPCOM RTS

Dev Journal Entry

Tree LOD and Distance Pass

Active Iteration

Trees / Performance

LOD visibility tuning outcomes

This pass focused on preserving forest silhouette continuity while avoiding frame time spikes from heavy near field geometry.

We separated high and mid tree distance behavior so costly meshes drop earlier while mid LOD carries long range readability.

  • Tree max distance and fade windows were tuned for smoother transitions.
  • Mid LOD handoff was aligned to reduce visible pop and overlap artifacts.
  • Distance presets remain overrideable through runtime settings for profiling.

GPU safety checks at higher densities

Raising tree counts stressed placement and culling grids. We validated storage limits and added adaptive guards so map scale and device constraints do not trigger unsafe allocations.

The grid path now scales cell sizing when limits are approached, trading precision for stability.

  • Storage buffer limits are read from the device and respected dynamically.
  • Grid resolution is clamped to keep allocation sizes safe on weaker adapters.
  • Fallback behavior avoids hard failure when density settings are extreme.

Next tuning focus

The next objective is to recover additional far field readability without regressing heavy scenes. We will target culling precision and draw submission efficiency before pushing visibility distances again.

This keeps visual gains tied to measurable performance headroom.

  • Improve far field tree culling selectivity.
  • Reduce redundant draw work in dense canopy overlap zones.
  • Expand benchmark scenarios for low and mid tier GPUs.

Key metrics snapshot

This pass is tracked with camera-sweep scenarios so pop behavior and frame variance can be compared directly before and after LOD tuning.

Performance checks run alongside GPU safety checks to avoid trading stability for visual continuity.

  • Near/high and mid LOD overlap-window transition quality.
  • GPU storage-buffer headroom and placement-grid safety margin.
  • Frame-time variance under high-density forest scenarios.
  • Pop-in and overlap artifact count during scripted flythroughs.

Next Steps