Experimenting coupling a CPU-based rigid body simulation with a GPU-based fluid simulation in OpenGL ES.
###Demo
(Rigid body integration is still in the works)
####Todo
- Switch to MAC (staggerd) grid
- Establish efficient pipeline for reading forces to CPU (via glReadPixels)
####Requires:
####Building Install haxe, then install the build tool aether from haxelib*
haxelib install aether
Then setup the aether command with
haxelib run aether setup
This creates an command alias so that aether executes haxelib run aether.
Next, make sure you've got the dependancies installed:
haxelib install lime
haxelib git gltoolbox https://github.com/haxiomic/GLToolbox.git
haxelib git shaderblox https://github.com/haxiomic/shaderblox.git
Now you should be good to go, cd into the gpu fluid directory and run
aether test html5
A browser should open with the result (test is aether's build and run command.)
* haxe just produces the javascript but aether handles the html template as well as the iOS and android project files when targeting mobile. Currently, lime is undergoing an overhaul so mobile targets are not working.
####Can I build for Desktop and mobile targets?
Desktop yes, mobile not yet. The lime library isn't quite finished, so it's not a super smooth process, follow the instructions on lime's git page to build lime's native binaries and then run
aether test mac or aether test windows or aether test linux. I've only tested the mac c++ build.
####Why haxe?
I'd recommend reading this article on the topic io.pellucid.com/blog/the-benefits-of-transpiling-to-javascript. In this project, haxe is used to create a tight binding between the glsl and the host. This makes setting a glsl uniform straightforward, something like shader.coordinate.x = 0.5; instead of juggling uniform locations and whatnot.