Water shaped by the landscape
I am developing water as a system that can be authored with the landscape: spline-defined rivers, carved beds and banks, local flow, rock contact and foam. Surface shape, simulation and optics need to work together, but they also need separate controls. A plausible riverbed will not fix an opaque water shader, and a good reflection will not hide folded geometry at an obstacle. Much of the recent work has been about finding that separation. Foam accumulated too broadly, clear streams inherited settings from rapids, and a minimum diffuse contribution left a milky veil between particles. The process images preserve those stages. They show how the problem developed and why the fixes involved material response and preset state as well as the water simulation.
The River Spline workflow separates terrain carving from enabling water. Width, depth and bank dimensions can be adjusted along the route, with adaptive sizing where the landscape varies. Terrain changes remain undoable, so I can compare the bed and bank fit before committing to the water's appearance.
The clear-water revision removed the minimum diffuse foam opacity and separated roughness from wave amplitude, absorption and scattering. Foam source settings now belong to each preset, preventing a clear-stream preset from retaining the previous rapids configuration. This fixes the source of the unwanted veil rather than trying to conceal it with colour or reflections.
Earlier iterations, material tests and diagnostic captures. Select a topic to follow the work behind the selected results.
Authored river domains feed local water computation and distance-dependent detail. I inspect obstacle contact, normals and foam continuity independently, then evaluate transmission, Fresnel reflection and volume absorption together. Foam should describe energetic surface behaviour while preserving the water material underneath it.
The latest clear-surface changes passed shader compilation and focused GPU and preset-transition checks. A fresh visual review was still pending in that development record. The older gallery images therefore remain process evidence, including excessive foam and diagnostic views, rather than illustrations of a finished clear-water result.
The October captures predate the latest clear-water revision. They document development states; the silent stills cannot show flow or foam transport over time.
NextTerrain