Conversation
Cull tiles that dont overlap. Push verts down using a hack. I think we need a sea level offset. The height needs to be pushed down per-vert, not per tile. A tile could overlap multiple lakes.
Works "ok" but destroys precision for any large delta :/. Also foam could not separate displacement from sea level.
# Conflicts: # crest/Assets/Crest/Crest/Scripts/LodData/LodDataMgrAnimWaves.cs
# Conflicts: # crest/Assets/Crest/Crest/Scripts/LodData/LodDataMgrSeaFloorDepth.cs # crest/Assets/Crest/Crest/Scripts/LodData/OceanDepthCache.cs # crest/Assets/Crest/Crest/Scripts/OceanRenderer.cs # crest/Assets/Crest/Crest/Shaders/OceanInputs/AnimWavesAddHeightFromGeometry.shader
# Conflicts: # crest/Assets/Crest/Crest/Scripts/Shapes/ShapeGerstnerBatched.cs
# Conflicts: # crest/Assets/Crest/Crest-Examples/Main/Materials/GerstnerPatchFeathered.mat.meta
Any ideas yourself? |
Not sure, its always tricky. Wondering if we should try make it textured or just make it grayscale or a grid pattern or etc. Textured will always be more professional/impressive etc but can be tricky to make look good. For the asset store versions we have some pbr textures used in the PirateCove scene which we have permission to use, maybe they are useful. Maybe a good approach is to take pirate cove and resculpt it into a different scene with a lake and a river, and could sprinkle in the rocks etc, and maybe can look good that way. The textures and assets would not be usable in BIRP (would not be pushable to MIT licensed git i dont think) but thats prob a secondary problem at this point. Other ideas.. Not sure if theres a free asset we could get away with using: https://www.turbosquid.com/Search/3D-Models/river Thought maybe this had a river but does not look like it: https://www.youtube.com/watch?v=iV-224nMwN8&ab_channel=Unity I guess a last option would be to put some funds towards contracting an artist to look at it. This would be good but i dont have one in mind so we'd have to ask on the discord or find one online, i think theres websites for this. I'll keep thinking about it. |
|
We can do both. Having basic scenes like boats.unity which have the flexibility to be purely instructive are useful. And it is good to have a scene for built-in since it is used for evaluation. Then we can keep fancy scenes like Pirate Cove for UAS only. |
|
Thanks for the notes. Shadows fixed now Data reverted Attempt at docs done |
Agreed |
|
@daleeidd i think i've addressed everything, re-requested review |
Mainly to be consistent with other functions.
magenta404
left a comment
There was a problem hiding this comment.
Nice work. Just a few review items. Mostly minor ones.
I've enabled docs for this branch in case you wish to review your changes: https://crest.readthedocs.io/en/feature-local-water-bodies/
crest/Assets/Crest/Crest/Scripts/LodData/LodDataMgrSeaFloorDepth.cs
Outdated
Show resolved
Hide resolved
|
I think that may be everything addressed, rerequested review. Thanks for those |
magenta404
left a comment
There was a problem hiding this comment.
Great work on achieving an important milestone! Works well and no issues that I experienced.
|
Awesome thanks for all your help! |
Status: Ready to merge
What & why:
An often-requested feature is to use Crest for local bodies of water (limited in area), rather than filling the whole world. I refer to this as 'local' as opposed to 'global'.
It turns out Crest is able to do local water well \o/ by adding the following mechanisms.
How:
Authoring all this is a bit of a brainful / juggling act, so i've also experimented with adding a wizard (Window/Crest/Create Water Body) which allows the user to place a water body, configure some settings, and it will create the setup. The resulting setup can be tweaked, moved, duplicated. - #561 (merged) . UPDATE: The authoring has been made much easier and I have now removed this wizard.
Testing:
There's a Lakes scene added. It's not representative of what would be pushed, its just for testing.
Issues:
Note on materials: There isnt a good mechanism for varying water materials/appearance. We could probably add a mechansim to assign an ocean material to each WaterBody, but this wouldn't help in the case of rivers which would need smoothly varying & connected attributes across water bodies. I think this is a future fish to fry. UPDATE: the simple approach merged in #902 . A more fancy approach would need to create a separate lod data for material blends, or such.