Resolution-aware visible nodes << total nodes in large dataset per-frame work is O(visible), not O(total) explicitly adapt to screen resolution O(visible) depends on pixel count work well on standard monitors fully exploit high-res displays scale to large datasets and large screens