Repository navigation
Install: download hyprw-1.4.0.zip, then in Blender Edit → Preferences → Get Extensions → Install from Disk (replaces 1.3.0). Requirements: Blender 5.2 or newer (tested on 5.3.0), Linux x86-64. Restart Blender after updating. Aerated Falling Water needs a new bake (the bake now stores the surface velocities).
New — materials from measurements
- Aerated Falling Water (Water › Fine-tune, on): falling water takes up air and turns white the further it falls, fully white at its breakup length. The breakup length comes from the largest inflow (Ervine et al. 1997, with the turbulence at the edge measured by Carrillo et al. 2020); Breakup overrides it. A short fall with much water stays nearly clear, as in nature: the learning scene turns white from 1.6 m and fully at 11.4 m. The bake stores the surface velocities for it, which also give Cycles motion blur on the water.
- Ripples From Wind (on): wind from the scene's force fields roughens the water as Cox & Munk (1954) measured it: 0.32 at 2 m/s, 0.40 at 5 m/s, 0.48 at 10 m/s. Without wind nothing changes.
- Wet Obstacles (off): rocks and banks turn darker and get a water film from the frame the water reaches them (Lekner & Dorf 1988: an albedo of 0.3 becomes 0.18). Works on coarse obstacles too: a modifier subdivides them for display and render; the mesh itself stays as it is. Switched off, the materials are as before.
- Foam Decay (3 s): foam thins out as it ages instead of vanishing at once (Callaghan et al. 2012, 552 breaking waves: 1.4–4.8 s).
- Traced Caustics (off): Cycles' Shadow Caustics trace the light the simulated waves throw onto the ground. Sharper network than the pattern, but measured 31 % darker on average and slower — the method is approximate.
Checked
- Foam brightness: a foam particle reflects 31 %, within the measured 22–55 % (Whitlock 1982, Koepke 1984).
Not added, and why
- Rainbows in the spray: built (each colour refracted with its own index) but not switched on. A rainbow is a caustic Cycles finds only by chance; 300,000 drops at 256 samples gave coloured noise instead of a bow.
Fixed
- Blender 5.3 changed how geometry nodes modifier inputs are set; HyprW uses the new way and the old one on 5.2.
Tests: 300 checks, 0 failures.
