Vortex2D¶
2D real-time fluid engine¶
Introduction¶
This is a 2D engine with the goal of being used in real-time scenarios, e.g. video games, in the same manner as a Box2D.
It is a hybrid engine that uses particles for the advection part (moving the particles given their velocities) and a grid to resolve the incompressible constrains. However, the particles are not visible to the user and you only work with the grid directly.
The engine runs directly on the GPU using Vulkan compute shaders. The rendering is then also done with Vulkan. The grids are represented by textures and operations by the user are all done by drawing shapes on the textures. The engine itself is written in C++ and it provides a simple wrapper around Vulkan and the basic rendering functionalities: shapes, textures, sprites, blending, render textures, etc.
Table of content¶
Setup¶
Vortex2D is multi-platform and currently supports the following:
- Windows
- Linux
- macOS
- iOS
CMake is used to generate the appropriate build scripts for each platform. The dependencies, which are fetched when calling cmake, are glm and SPIRV-cross. The tests use gtest and the examples use glfw.
The only dependency required is python. There a several variables that can be used to configure:
CMake | Builds |
---|---|
VORTEX2D_ENABLE_TESTS | builds the tests |
VORTEX2D_ENABLE_EXAMPLES | builds the examples |
VORTEX2D_ENABLE_DOCS | builds the documentation |
The main library is built as a dll on windows, shared library on linux and (dynamic) framework on macOS/iOS.
Prerequisite¶
Following dependencies are necessary:
- Python
- glslangValidator (comes with Vulkan SDK)
Following minimum compilers are necessary:
- GCC 5.4 or later
- MSVC 2015 or later
- Clang 3.4 or later
Windows¶
To build on windows, cmake-gui is the easiest to use. Only the variables specified above should be changed.
Linux¶
The package xorg-dev might need to first be installed. Again, regular cmake commands should be use to configure cmake:
cmake ..
macOS¶
In addition to the normal variables, we need to specify the location of MoltenVK and the glslang compiler. The glslang compiler can be downloaded from its project on github: https://github.com/KhronosGroup/glslang/releases
cmake .. -DMOLTENVK_DIR=path_to/MoltenVK/Package/Latest/MoltenVK/ -DGLSL_VALIDATOR=path_to/bin/glslangValidator
iOS¶
The framework needs to signed on iOS, so the following variables need to be defined:
Variable | Value |
CODE_SIGN_IDENTITY | “iPhone Developer” |
DEVELOPMENT_TEAM_ID | set to the team id, can be found on the apple developer portal |
In addition, the MoltenVK location has to be specified, and the toolchain:
cmake .. -DCMAKE_TOOLCHAIN_FILE=../cmake/ios.toolchain.cmake -DIOS_PLATFORM=OS -DIOS_ARCH=arm64 -DENABLE_VISIBILITY=true -DGLSL_VALIDATOR=path_to/bin/glslangValidator -DMOLTENVK_DIR=path_to_sdk/MoltenVK/ -DCODE_SIGN_IDENTITY="iPhone Developer" -DDEVELOPMENT_TEAM_ID=XXXXXX
Documentation¶
To build the documentation the following is required:
- doxygen
- sphinx
- sphinx_rtd_theme
- sphinx breathe
Rendering¶
Initialization¶
The rendering API is very basic and supports only the most basic functionality.
Create an instance of Vortex2D::Renderer::Instance
which is then used to create an instance of Vortex2D::Renderer::Device
.
The device is then used to create any other object. The main one is the Vortex2D::Renderer::RenderWindow
which is a window where to render sprites and polygons.
The function Vortex2D::Fluid::RenderWindow::Display()
is then used to present the result to the screen.
Vortex2D::Renderer::Instance instance("Application name", extensions); // pass list of required extensions
Vortex2D::Renderer::Device device(instance.GetPhysicalDevice(), surface);
Vortex2D::Renderer::RenderWindow window(device, surface, width, height);
Note that the instance requires a list of extensions necessary to create a window. With GLFW they can be retrived as:
std::vector<const char*> GetGLFWExtensions()
{
std::vector<const char*> extensions;
unsigned int glfwExtensionCount = 0;
const char** glfwExtensions;
// get the required extensions from GLFW
glfwExtensions = glfwGetRequiredInstanceExtensions(&glfwExtensionCount);
for (unsigned int i = 0; i < glfwExtensionCount; i++)
{
extensions.push_back(glfwExtensions[i]);
}
return extensions;
}
In addition, you also need to create a surface which can be also done with the help of GLFW:
vk::UniqueSurfaceKHR GetGLFWSurface(GLFWwindow* window, vk::Instance instance)
{
// create surface
VkSurfaceKHR surface;
if (glfwCreateWindowSurface(static_cast<VkInstance>(instance), window, nullptr, &surface) != VK_SUCCESS)
{
throw std::runtime_error("failed to create window surface!");
}
return vk::UniqueSurfaceKHR(surface, vk::SurfaceKHRDeleter{instance});
}
Render Targets¶
To be able to render, we need to record Vortex2D::Renderer::RenderCommand
on a Vortex2D::Renderer::RenderTarget
. There are two implementations of it:
You can render implementations of the abstract class Vortex2D::Renderer::Drawable
, which get recorder in the render command. To actually render it on the render target, the submit function needs to be called. Note, it can be called repeatedly (e.g. over several frames).
In addition, the blend state needs to be passed in, see Vortex2D::Renderer::ColorBlendState
.
Shapes¶
We are now ready to draw things on the screen. Let’s start with some shapes like rectangles and circles:
Vortex2D::Renderer::Rectangle rectangle(device, {100.0f, 100.0f});
Vortex2D::Renderer::Ellipse circle(device, {50.0f, 50.0f});
auto blendMode = vk::PipelineColorBlendAttachmentState()
.setBlendEnable(true)
.setAlphaBlendOp(vk::BlendOp::eAdd)
.setColorBlendOp(vk::BlendOp::eAdd)
.setSrcColorBlendFactor(vk::BlendFactor::eSrcAlpha)
.setSrcAlphaBlendFactor(vk::BlendFactor::eOne)
.setDstColorBlendFactor(vk::BlendFactor::eOneMinusSrcAlpha)
.setDstAlphaBlendFactor(vk::BlendFactor::eZero);
// note that rectangle, circle and render need to be alive for the duration of the rendering
auto render = renderTarget.Record({rectangle, circle}, blendMode);
render.Submit();
Textures¶
Of course we can also render textures, using sprites.
Vortex2D::Renderer::Texture texture(device, 100, 100, vk::Format::eR8G8B8A8Unorm);
Vortex2D::Renderer::Sprite sprite(device, texture);
Transformations¶
The shapes and textures can be positioned, i.e. are transformable. You can set the following properties on them:
- Position
- Scale
- Rotation
- Anchor
As an example:
Vortex2D::Renderer::Ellipse circle(device, {50.0f, 50.0f});
circle.Colour = {0.0f, 0.0f, 1.0f, 1.0f};
circle.Position = {500.0f, 400.0f};
Level sets¶
A level set is a signed distance field. It’s a field containing positive or negative value, where the values are 0 represent a contour, or border. This is used to represent shapes, the numbers give you the distance to the shape border. It’s the fundamental way that we represent the area of a fluid and the area of the obstacles, i.e. the boundaries.
The level set is represented simply as a float texture. To set the level set, we simply render on that texture. This means that the class Vortex2D::Fluid::LevelSet
inherits Vortex2D::Renderer::RenderTexture
.
Basic shapes¶
There is a list of basic shapes that can be used to render on a level set:
They are used the same way as regular drawable shapes, i.e.
Vortex2D::Fluid::Rectangle rectangle(device, {100.0f, 100.0f});
rectangle.Position = {40.0f, 60.0f};
Vortex2D::Fluid::LevelSet levelSet(device, {400, 400});
auto renderCmd = levelSet.Record({rectangle});
renderCmd.Submit(); // note that renderCmd and rectangle have to be alive untill the rendering is done
Combining shapes¶
Multiple shapes can be combined together to build the level set. You can either take the union or the intersection when rendering. This happens by using certain blend states which are:
Vortex2D::Renderer::IntersectionBlend
Vortex2D::Renderer::UnionBlend
After combining several shapes, the resulting float texture is not a signed distance field. It needs to be reinitialised which is simply done by calling
Vortex2D::Fluid::LevelSet::Reinitialise()
.
World¶
The world classes are the centre of the engine, where the fluid gets animated. They contain essentially three fields:
- The velocity field
- The liquid phi field
- The solid phi field
The first one contain the velocity of the fluid at every point, the second one defines where the fluid is. This is a signed distance field where a negative value indicates this is a fluid location. Finally the last one contains the location of solid obstacles, again as a signed distance field where the negative values indicate the solid’s location.
Each can be visualised as a texture with the getters:
Renderer::RenderTexture& GetVelocity();
DistanceField LiquidDistanceField();
DistanceField SolidDistanceField();
Of course, to get interesting fluid simulations, we need to set values on them. Setting the signed distance fields is straightword (see Level sets):
Renderer::RenderCommand RecordLiquidPhi(Renderer::RenderTarget::DrawableList drawables);
Renderer::RenderCommand RecordStaticSolidPhi(Renderer::RenderTarget::DrawableList drawables);
Note that this only has to be done once.
For velocities however, the simulation needs to set the velocities at a specific time during the simulation, so instead of ourselves calling Vortex2D::Renderer::RenderCommand::Submit()
we pass the Vortex2D::Renderer::RenderCommand()
to the World::Fluid::World()
class:
Renderer::RenderCommand RecordVelocity(Renderer::RenderTarget::DrawableList drawables);
void SubmitVelocity(Renderer::RenderCommand& renderCommand);
Stepping through the simulation is done with the Vortex2D::Fluid::World::Step()
function, which takes as parameter the number of iterations used in the linear solver.
This can either be a fixed number of steps, or until the error reaches a certain threshhold.
auto iterations = Fluid::FixedParams(12);
world.Step(iterations);
Smoke World¶
This is a type of fluid simulation where the fluid area doesn’t move. This is used to simulate smoke type effects by having a colored texture be advected by the velocity field.
The class Vortex2D::Fluid::Density
is used for this, it is simply a texture that can be rendered (i.e. a sprite).
The simulation is setup as so:
Fluid::Density density(device, size, vk::Format::eR8G8B8A8);
Fluid::SmokeWorld world(device, size, 0.033);
world.FieldBind(density);
Water World¶
This is a classical water type of fluid simulation. This has a fluid area which evoles over time, i.e. a area of water moving. The area of water and non-water can be specified by rendering onto the word, where each pixel indicates the number of particles to add/substract.
Renderer::RenderCommand RecordParticleCount(Renderer::RenderTarget::DrawableList drawables);
The constraint is that the drawable needs to render integer values, which is provided for example by Vortec2D::Renderer::IntRectangle
and used:
Renderer::IntRectangle fluid(device, {150.0f, 50.0f});
fluid.Position = {50.0f, 25.0f};
fluid.Colour = glm::vec4(4); // can also be -4
world.RecordParticleCount({fluid}).Submit().Wait();
Rigid body¶
Rigid bodies are the way to have dynamic interations with the fluid (other then changing the velocity field directly). Vortex2D only provides a way to get current forces applied to the rigidbodies, and applying velocities from the rigidbody to the fluid. Updating the rigidbody’s forces, velocities and position needs to be done by a seperate engine, such as Box2D.
Rigidbodies have three types:
- Static
- Weak
- Strong
Static bodies¶
Static bodies act on the fluid, but the fluid doesn’t act on the fluid. They have a velocity that is imparted on the fluid. Think of motorized objects pushing through the fluid.
Weak/Strong bodies¶
Weak rigidbodies are affected by the fluid. They can also in turn, affect the fluid, which is called a strong coupling with the fluid.
Rigid body updates¶
Mass and velocity is set using simple setter functions:
Rigidbody rigidbody(device, size, drawable, type);
rigidbody.SetMassData(mass, inertia);
rigidbody.SetVelocities(velocity, angle);
Position and orientation is updated the same as with shapes:
rigidbody.Position = {100.0f, 100.0f}
rigidbody.Rotation = 43.0f;
Rigid body coupling¶
To have the fluid influence the rigid bodies and vice versa, two functions need to be implemented by deriving:
The first one has forces from the fluid applied to the rigidbody. The second has velocities from the rigidbody applied to the fluid.
An example implementation with Box2D is as follow:
void Box2DRigidbody::ApplyForces()
{
if (GetType() & Vortex2D::Fluid::RigidBody::Type::eWeak)
{
auto force = GetForces();
b2Vec2 b2Force = {force.velocity.x, force.velocity.y};
mBody->ApplyForceToCenter(b2Force, true);
mBody->ApplyTorque(force.angular_velocity, true);
}
}
void Box2DRigidbody::ApplyVelocities()
{
auto pos = mBody->GetPosition();
Position = {pos.x, pos.y};
Rotation = glm::degrees(mBody->GetAngle());
if (GetType() & Vortex2D::Fluid::RigidBody::Type::eStatic)
{
glm::vec2 vel = {mBody->GetLinearVelocity().x, mBody->GetLinearVelocity().y};
float angularVelocity = mBody->GetAngularVelocity();
SetVelocities(vel, angularVelocity);
}
}
Note that any rigidbody physics can be used: chipmonk, bullet, etc.
Engine updates¶
Finally the rigidbody also needs to be updates, in lock-step with the fluid simulation.
Again, this is done by implementing Vortex2D::Fluid::RigidBody::Step()
.
An example implementation with Box2D:
void Box2DSolver::Step(float delta)
{
const int velocityStep = 8;
const int positionStep = 3;
mWorld.Step(delta, velocityStep, positionStep);
}
The delta is the same used to create the world object.
Renderer API reference¶
Classes¶
Vortex2D::Renderer::Clear
Vortex2D::Renderer::Drawable
Vortex2D::Renderer::Ellipse
Vortex2D::Renderer::GenericBuffer
Vortex2D::Renderer::IndirectBuffer
Vortex2D::Renderer::Instance
Vortex2D::Renderer::IntRectangle
Vortex2D::Renderer::Rectangle
Vortex2D::Renderer::RenderState
Vortex2D::Renderer::RenderTarget
Vortex2D::Renderer::RenderTexture
Vortex2D::Renderer::RenderWindow
Vortex2D::Renderer::Sprite
Vortex2D::Renderer::Timer
Vortex2D::Renderer::Transformable
Vortex2D::Renderer::UniformBuffer
Vortex2D::Renderer::VertexBuffer
Vortex2D::Renderer::Work
API Reference¶
-
namespace
Renderer
¶ Typedefs
-
using
DescriptorTypeBindings
= std::map<uint32_t, vk::DescriptorType>¶
-
typedef std::vector<glm::vec2>
Path
¶
Functions
-
template<template<typename> class
BufferType
, typenameT
>
voidCopyTo
(BufferType<T> &buffer, T &t)¶ Copy the content of a buffer in an object.
-
template<template<typename> class
BufferType
, typenameT
>
voidCopyTo
(BufferType<T> &buffer, std::vector<T> &t)¶ Copy the content of a buffer to a vector. Vector needs to have the correct size already.
-
template<template<typename> class
BufferType
, typenameT
>
voidCopyFrom
(BufferType<T> &buffer, const T &t)¶ Copy the content of an object to the buffer.
-
template<template<typename> class
BufferType
, typenameT
>
voidCopyFrom
(BufferType<T> &buffer, const std::vector<T> &t)¶ Copy the content of a vector to the buffer.
-
void
BufferBarrier
(vk::Buffer buffer, vk::CommandBuffer commandBuffer, vk::AccessFlags oldAccess, vk::AccessFlags newAccess)¶ Inserts a barrier for the given buffer, command buffer and access.
- Parameters
buffer
: the vulkan buffer handlecommandBuffer
: the command buffer to inserts the barrieroldAccess
: old accessnewAccess
: new access
-
bool
operator==
(const ShaderLayout &left, const ShaderLayout &right)¶
-
bool
operator==
(const PipelineLayout &left, const PipelineLayout &right)¶
-
void
Bind
(const Device &device, DescriptorSet &dstSet, const PipelineLayout &layout, const std::vector<BindingInput> &bindingInputs)¶ Bind the resources (buffer or texture/sampler) to a DescriptorSet.
- Parameters
device
: vulkan devicedstSet
: vulkan descriptor setlayout
: pipeline layoutbindingInputs
: list of resources (buffer or texture/sampler)
-
bool
HasLayer
(const char *extension, const std::vector<vk::LayerProperties> &availableExtensions)¶
-
bool
HasExtension
(const char *extension, const std::vector<vk::ExtensionProperties> &availableExtensions)¶
-
bool
operator==
(const GraphicsPipeline &left, const GraphicsPipeline &right)¶
-
bool
operator==
(const SpecConstInfo &left, const SpecConstInfo &right)¶
-
template<typename
Type
>
SpecConstInfo::Value<Type>SpecConstValue
(uint32_t id, Type value)¶ Constructs a specialization constant value.
-
template<typename ...
Args
>
SpecConstInfoSpecConst
(Args&&... args)¶ Constructs a SpecConstInfo with given values of specialisation constants.
-
bool
operator==
(const RenderState &left, const RenderState right)¶
-
vk::DeviceSize
GetBytesPerPixel
(vk::Format format)¶ Gets the number of bytes per pixel given the format.
- Return
- bytes per pixel
- Parameters
format
: of texture
-
void
TextureBarrier
(vk::Image image, vk::CommandBuffer commandBuffer, vk::ImageLayout oldLayout, vk::AccessFlags srcMask, vk::ImageLayout newLayout, vk::AccessFlags dstMask)¶ Inserts a barrier for the given texture, command buffer and access.
- Parameters
image
: the vulkan image handlecommandBuffer
: the vulkan command bufferoldLayout
: old layoutsrcMask
: old accessnewLayout
: new layoutdstMask
: new access
-
ComputeSize
MakeStencilComputeSize
(const glm::ivec2 &size, int radius)¶ Create a ComputeSize for a stencil type shader.
- Return
- calculate ComputeSize
- Parameters
size
: the domain sizeradius
: the stencil size
-
ComputeSize
MakeCheckerboardComputeSize
(const glm::ivec2 &size)¶ Create a ComputeSize for a checkerboard type shader.
- Return
- calculate ComputeSize
- Parameters
size
: the domain size
-
class
AbstractShape
: public Vortex2D::Renderer::Shape¶ - #include <Shapes.h>
An polygonal shape where the fragment shader can be specified for customisation.
Subclassed by Vortex2D::Renderer::IntRectangle, Vortex2D::Renderer::Rectangle
Public Functions
-
void
Initialize
(const RenderState &renderState)¶ Initialize the drawable for a particular state. This might include creating the correct pipeline. If it was already initialized, it will do nothing.
- Parameters
renderState
: the state to initialize with.
-
void
Update
(const glm::mat4 &projection, const glm::mat4 &view)¶ Update the MVP matrix of the drawable.
- Parameters
projection
: the projection matrixview
: the view matrix
-
void
Draw
(vk::CommandBuffer commandBuffer, const RenderState &renderState)¶ Draw for the given render state. This has to be initialized before.
- Parameters
commandBuffer
: the command buffer to record into.renderState
: the render state to use.
-
void
-
class
AbstractSprite
: public Vortex2D::Renderer::Drawable, public Vortex2D::Renderer::Transformable¶ - #include <Sprite.h>
a Sprite, i.e. a drawable that can render a texture. The fragment shader can be specified for customisation.
Subclassed by Vortex2D::Fluid::DistanceField, Vortex2D::Renderer::Sprite
Public Functions
-
void
Initialize
(const RenderState &renderState)¶ Initialize the drawable for a particular state. This might include creating the correct pipeline. If it was already initialized, it will do nothing.
- Parameters
renderState
: the state to initialize with.
-
void
Update
(const glm::mat4 &projection, const glm::mat4 &view)¶ Update the MVP matrix of the drawable.
- Parameters
projection
: the projection matrixview
: the view matrix
-
void
Draw
(vk::CommandBuffer commandBuffer, const RenderState &renderState)¶ Draw for the given render state. This has to be initialized before.
- Parameters
commandBuffer
: the command buffer to record into.renderState
: the render state to use.
-
void
-
struct
BindingInput
¶ - #include <DescriptorSet.h>
The texture/sampler or buffer that can be binded to a shader.
-
template<typename
T
>
classBuffer
: public Vortex2D::Renderer::GenericBuffer¶ - #include <Buffer.h>
a storage buffer type of buffer
-
class
Clear
: public Vortex2D::Renderer::Drawable¶ - #include <Shapes.h>
A drawable that simply clears the target.
Public Functions
-
void
Initialize
(const RenderState &renderState)¶ Initialize the drawable for a particular state. This might include creating the correct pipeline. If it was already initialized, it will do nothing.
- Parameters
renderState
: the state to initialize with.
-
void
Update
(const glm::mat4 &projection, const glm::mat4 &view)¶ Update the MVP matrix of the drawable.
- Parameters
projection
: the projection matrixview
: the view matrix
-
void
Draw
(vk::CommandBuffer commandBuffer, const RenderState &renderState)¶ Draw for the given render state. This has to be initialized before.
- Parameters
commandBuffer
: the command buffer to record into.renderState
: the render state to use.
-
void
-
struct
ColorBlendState
¶ - #include <RenderState.h>
The blend state and blend constant.
-
class
CommandBuffer
¶ - #include <CommandBuffer.h>
Can record commands, then submit them (multiple times). A fence can used to wait on the completion of the commands.
Public Functions
-
CommandBuffer
(const Device &device, bool synchronise = true)¶ Creates a command buffer which can be synchronized.
- Parameters
device
: vulkan devicesynchronise
: flag to determine if the command buffer can be waited on.
-
CommandBuffer &
Record
(CommandFn commandFn)¶ Record some commands. The commads are recorded in the lambda which is immediately executed.
- Parameters
commandFn
: a functor, or simply a lambda, where commands are recorded.
-
CommandBuffer &
Record
(const RenderTarget &renderTarget, vk::Framebuffer framebuffer, CommandFn commandFn)¶ Record some commands inside a render pass. The commads are recorded in the lambda which is immediately executed.
- Parameters
renderTarget
: the render target which contains the render pass to record intoframebuffer
: the frame buffer where the render pass will render.commandFn
: a functor, or simply a lambda, where commands are recorded.
-
CommandBuffer &
Wait
()¶ Wait for the command submit to finish. Does nothing if the synchronise flag was false.
-
CommandBuffer &
Reset
()¶ Reset the command buffer so it can be recorded again.
-
CommandBuffer &
Submit
(const std::initializer_list<vk::Semaphore> &waitSemaphores = {}, const std::initializer_list<vk::Semaphore> &signalSemaphores = {})¶ submit the command buffer
-
operator bool
() const¶ explicit conversion operator to bool, indicates if the command was properly recorded and can be sumitted.
-
-
struct
ComputeSize
¶ - #include <Work.h>
Used for a compute shader, and defines the group size, local size and domain size.
Public Functions
-
ComputeSize
(const glm::ivec2 &size, const glm::ivec2 &localSize = GetLocalSize2D())¶ Creates a ComputeSize using a 2D domain size and the default 2D local size.
- Parameters
size
: the domain sizelocalSize
: the local size of the shader
-
ComputeSize
(int size, int localSize = GetLocalSize1D())¶ Creates a ComputeSize using a 1D domain size and the default 1D local size.
- Parameters
size
: the domain sizelocalSize
: the local size of the shader
Public Static Functions
-
static glm::ivec2
GetLocalSize2D
()¶ The default local size for 2D compute shaders.
- Return
- a 2d vector
-
static int
GetLocalSize1D
()¶ The default local size for 1D compute shaders.
- Return
- a integer value
-
static glm::ivec2
GetWorkSize
(const glm::ivec2 &size, const glm::ivec2 &localSize = GetLocalSize2D())¶ Computes the 2D group size given a domain size.
- Return
- the group size
- Parameters
size
: the domain size of the shaderlocalSize
: the local size of the shader
-
static glm::ivec2
GetWorkSize
(int size, int localSize = GetLocalSize1D())¶ Computes the 1D group size given a domain size.
- Return
- the group size
- Parameters
size
: the domain size of the shaderlocalSize
: the local size of the shader
-
static ComputeSize
Default2D
()¶ A default ComputeSize using the default 2D local size. The domain size is (1,1)
- Return
- a default compute size
-
static ComputeSize
Default1D
()¶ A default ComputeSize using the default 1D local size. The domain size is (1,1)
- Return
- a default compute size
-
-
struct
DescriptorImage
¶ - #include <DescriptorSet.h>
The texture or sampler that can be binded to a shader.
-
struct
DescriptorSet
¶ - #include <DescriptorSet.h>
The binding of an object for a shader.
-
class
Device
¶ - #include <Device.h>
Encapsulation around the vulkan device. Allows to create command buffers, layout, bindings, memory and shaders.
-
struct
DispatchParams
¶ - #include <Work.h>
Parameters for indirect compute: group size, local size, etc.
-
struct
Drawable
¶ - #include <Drawable.h>
Interface of a drawable object.
Subclassed by Vortex2D::Fluid::Circle, Vortex2D::Fluid::Polygon, Vortex2D::Renderer::AbstractSprite, Vortex2D::Renderer::Clear, Vortex2D::Renderer::Shape
Public Functions
-
virtual void
Initialize
(const RenderState &renderState) = 0¶ Initialize the drawable for a particular state. This might include creating the correct pipeline. If it was already initialized, it will do nothing.
- Parameters
renderState
: the state to initialize with.
-
virtual void
Update
(const glm::mat4 &projection, const glm::mat4 &view) = 0¶ Update the MVP matrix of the drawable.
- Parameters
projection
: the projection matrixview
: the view matrix
-
virtual void
Draw
(vk::CommandBuffer commandBuffer, const RenderState &renderState) = 0¶ Draw for the given render state. This has to be initialized before.
- Parameters
commandBuffer
: the command buffer to record into.renderState
: the render state to use.
-
virtual void
-
struct
DynamicDispatcher
¶ - #include <Device.h>
A vulkan dynamic dispatcher that checks if the function is not null.
-
class
Ellipse
: public Vortex2D::Renderer::Shape¶ - #include <Shapes.h>
A solid colour ellipse. Implements the Drawable interface and Transformable interface.
Public Functions
-
void
Initialize
(const RenderState &renderState)¶ Initialize the drawable for a particular state. This might include creating the correct pipeline. If it was already initialized, it will do nothing.
- Parameters
renderState
: the state to initialize with.
-
void
Update
(const glm::mat4 &projection, const glm::mat4 &view)¶ Update the MVP matrix of the drawable.
- Parameters
projection
: the projection matrixview
: the view matrix
-
void
Draw
(vk::CommandBuffer commandBuffer, const RenderState &renderState)¶ Draw for the given render state. This has to be initialized before.
- Parameters
commandBuffer
: the command buffer to record into.renderState
: the render state to use.
-
void
-
class
GenericBuffer
¶ - #include <Buffer.h>
A vulkan buffer which can be on the host or the device.
Subclassed by Vortex2D::Renderer::Buffer< float >, Vortex2D::Renderer::Buffer< glm::ivec2 >, Vortex2D::Renderer::Buffer< glm::vec2 >, Vortex2D::Renderer::Buffer< int >, Vortex2D::Renderer::Buffer< Vortex2D::Fluid::Particle >, Vortex2D::Renderer::Buffer< Vortex2D::Fluid::RigidBody::Velocity >, Vortex2D::Renderer::Buffer< Vortex2D::Renderer::DispatchParams >, Vortex2D::Renderer::IndirectBuffer< Vortex2D::Renderer::DispatchParams >, Vortex2D::Renderer::UniformBuffer< glm::mat4 >, Vortex2D::Renderer::UniformBuffer< glm::vec2 >, Vortex2D::Renderer::UniformBuffer< glm::vec4 >, Vortex2D::Renderer::UniformBuffer< Size >, Vortex2D::Renderer::UniformBuffer< Vortex2D::Fluid::RigidBody::Velocity >, Vortex2D::Renderer::VertexBuffer< glm::vec2 >, Vortex2D::Renderer::VertexBuffer< Vortex2D::Renderer::AbstractSprite::Vertex >, Vortex2D::Renderer::Buffer< T >, Vortex2D::Renderer::IndexBuffer< T >, Vortex2D::Renderer::IndirectBuffer< T >, Vortex2D::Renderer::UniformBuffer< T >, Vortex2D::Renderer::VertexBuffer< T >
Public Functions
-
void
CopyFrom
(vk::CommandBuffer commandBuffer, GenericBuffer &srcBuffer)¶ Copy a buffer to this buffer.
- Parameters
commandBuffer
: command buffer to run the copy on.srcBuffer
: the source buffer.
-
void
CopyFrom
(vk::CommandBuffer commandBuffer, Texture &srcTexture)¶ Copy a texture to this buffer.
- Parameters
commandBuffer
: command buffer to run the copy on.srcTexture
: the source texture
-
vk::Buffer
Handle
() const¶ The vulkan handle.
-
vk::DeviceSize
Size
() const¶ The size in bytes of the buffer.
-
void
Resize
(vk::DeviceSize size)¶ Resize the buffer. Invalidates the buffer handle.
- Parameters
size
: buffer size
-
void
Barrier
(vk::CommandBuffer commandBuffer, vk::AccessFlags oldAccess, vk::AccessFlags newAccess)¶ Inserts a barrier for this buffer.
- Parameters
commandBuffer
: the command buffer to run the barrieroldAccess
: old accessnewAccess
: new access
-
void
Clear
(vk::CommandBuffer commandBuffer)¶ Clear the buffer with 0.
- Parameters
commandBuffer
: the command buffer to clear on
-
void
CopyFrom
(uint32_t offset, const void *data, uint32_t size)¶ copy from data to buffer
- Parameters
offset
: in the bufferdata
: pointersize
: of data
-
void
CopyTo
(uint32_t offset, void *data, uint32_t size)¶ copy buffer to data
- Parameters
offset
: in the bufferdata
: pointersize
: of data
-
void
-
class
GraphicsPipeline
¶ - #include <Pipeline.h>
graphics pipeline which caches the pipeline per render states.
Public Functions
-
GraphicsPipeline &
Shader
(vk::ShaderModule shader, vk::ShaderStageFlagBits shaderStage)¶ Set the shader.
- Return
- *this
- Parameters
shader
: the loaded shadershaderStage
: shader state (vertex, fragment or compute)
-
GraphicsPipeline &
VertexAttribute
(uint32_t location, uint32_t binding, vk::Format format, uint32_t offset)¶ Sets the vertex attributes.
- Return
- *this
- Parameters
location
: location in the shaderbinding
: binding in the shaderformat
: vertex formatoffset
: offset in the vertex
-
GraphicsPipeline &
VertexBinding
(uint32_t binding, uint32_t stride, vk::VertexInputRate inputRate = vk::VertexInputRate::eVertex)¶ Sets the vertex binding.
- Return
- *this
- Parameters
binding
: binding in the shaderstride
: stride in bytesinputRate
: inpute rate
-
GraphicsPipeline &
-
template<typename
T
>
classIndexBuffer
: public Vortex2D::Renderer::GenericBuffer¶ - #include <Buffer.h>
a index buffer type of buffer
-
template<typename
T
>
classIndirectBuffer
: public Vortex2D::Renderer::GenericBuffer¶ - #include <Buffer.h>
an indirect buffer type of buffer, used for compute indirect dispatch
-
class
Instance
¶ - #include <Instance.h>
Vulkan instance, which extensions enabled.
-
class
IntRectangle
: public Vortex2D::Renderer::AbstractShape¶ - #include <Shapes.h>
A solid colour rectangle as Rectangle, however uses integer colors and is meant to be drawn to a framebuffer wiht integer colours.
-
class
LayoutManager
¶ - #include <DescriptorSet.h>
Caches and creates layouts and bindings.
Public Functions
-
void
CreateDescriptorPool
(int size = 512)¶ Create or re-create the descriptor pool, will render invalid existing descriptor sets.
- Parameters
size
: size of the pool
-
DescriptorSet
MakeDescriptorSet
(const PipelineLayout &layout)¶ Create the descriptor set given the layout.
- Return
- built descriptor set
- Parameters
layout
: pipeline/shader layout
-
vk::DescriptorSetLayout
GetDescriptorSetLayout
(const PipelineLayout &layout)¶ Create, cache and return a descriptor layout given the pipeline layout.
- Return
- cached descriptor set layout
- Parameters
layout
: pipeline layout
-
vk::PipelineLayout
GetPipelineLayout
(const PipelineLayout &layout)¶ create, cache and return a vulkan pipeline layout given the layout
- Return
- vulkan pipeline layout
- Parameters
layout
: pipeline layout
-
void
-
class
PipelineCache
¶ - #include <Pipeline.h>
Create pipelines using vulkan’s pipeline cache.
Public Functions
-
void
CreateCache
()¶ Create the pipeline cache.
-
vk::Pipeline
CreateGraphicsPipeline
(const GraphicsPipeline &builder, const RenderState &renderState)¶ Create a graphics pipeline.
- Return
- Parameters
builder
:renderState
:
-
vk::Pipeline
CreateComputePipeline
(vk::ShaderModule shader, vk::PipelineLayout layout, SpecConstInfo specConstInfo = {})¶ Create a compute pipeline.
- Parameters
shader
:layout
:specConstInfo
:
-
void
-
struct
PipelineLayout
¶ - #include <DescriptorSet.h>
Represents the layout of a pipeline: vertex + fragment or compute.
-
class
Rectangle
: public Vortex2D::Renderer::AbstractShape¶ - #include <Shapes.h>
A solid colour rectangle defined by two triangles. Implements the Drawable interface and Transformable interface.
-
class
RenderCommand
¶ - #include <CommandBuffer.h>
A special command buffer that has been recorded by a RenderTarget. It can be used to submit the rendering. The object has to stay alive untill rendering is complete.
Public Functions
-
RenderCommand &
Submit
(const glm::mat4 &view = glm::mat4(1.0f))¶ Submit the render command with a transform matrix.
- Return
- *this
- Parameters
view
: a transform matrix
-
void
Wait
()¶ Wait for the render command to complete.
-
operator bool
() const¶ explicit conversion operator to bool, indicates if the command was properly recorded and can be sumitted.
-
RenderCommand &
-
class
RenderpassBuilder
¶ - #include <RenderTarget.h>
Factory for a vulkan render pass.
Public Functions
-
RenderpassBuilder &
Attachement
(vk::Format format)¶ Format of the render pass.
- Return
- Parameters
format
:
-
RenderpassBuilder &
AttachementLoadOp
(vk::AttachmentLoadOp value)¶ operation to perform when loading the framebuffer (clear, load, etc)
- Return
- Parameters
value
:
-
RenderpassBuilder &
AttachementStoreOp
(vk::AttachmentStoreOp value)¶ operation to perform when storing the framebuffer (clear, save, etc)
- Return
- Parameters
value
:
-
RenderpassBuilder &
AttachementInitialLayout
(vk::ImageLayout layout)¶ Layout of the image to be before render pass.
- Return
- Parameters
layout
:
-
RenderpassBuilder &
AttachementFinalLayout
(vk::ImageLayout layout)¶ Layout of the image to be after render pass.
- Return
- Parameters
layout
:
-
RenderpassBuilder &
Subpass
(vk::PipelineBindPoint bindPoint)¶ Define subpass of the render pass.
- Return
- Parameters
bindPoint
:
-
RenderpassBuilder &
SubpassColorAttachment
(vk::ImageLayout layout, uint32_t attachment)¶ Set the color attachment with index.
- Return
- Parameters
layout
:attachment
: index of the attachment
-
RenderpassBuilder &
Dependency
(uint32_t srcSubpass, uint32_t dstSubpass)¶ Dependency of the subpasses.
- Return
- Parameters
srcSubpass
:dstSubpass
:
-
vk::UniqueRenderPass
Create
(vk::Device device)¶ Create the render pass.
- Return
- Parameters
device
:
-
RenderpassBuilder &
-
struct
RenderState
¶ - #include <RenderState.h>
the various state to render to a target: size, render pass and blend.
Public Functions
-
RenderState
(const RenderTarget &renderTarget)¶ Initialize for a render target with default blend.
- Parameters
renderTarget
:
-
RenderState
(const RenderTarget &renderTarget, ColorBlendState blendState)¶ Initialize for a render target with a given blend.
- Parameters
renderTarget
:blendState
:
-
-
struct
RenderTarget
¶ - #include <RenderTarget.h>
A target that can be rendered to. This is implemented by the RenderWindow and the RenderTexture.
Subclassed by Vortex2D::Renderer::RenderTexture, Vortex2D::Renderer::RenderWindow
-
class
RenderTexture
: public Vortex2D::Renderer::RenderTarget, public Vortex2D::Renderer::Texture¶ - #include <RenderTexture.h>
A render target that renders into a texture.
Subclassed by Vortex2D::Fluid::Density, Vortex2D::Fluid::LevelSet, Vortex2D::Fluid::ParticleCount, Vortex2D::Fluid::Velocity
-
class
RenderWindow
: public Vortex2D::Renderer::RenderTarget¶ - #include <RenderWindow.h>
Render to a swapchain, i.e. to the window/surface.
Public Functions
-
RenderWindow
(const Device &device, vk::SurfaceKHR surface, uint32_t width, uint32_t height)¶ Initialize with a given surface and size.
- Parameters
device
: vulkan devicesurface
: vulkan surfacewidth
:height
:
-
void
Display
()¶ Submits all the render command and present the surface for display.
-
-
class
SamplerBuilder
¶ - #include <Texture.h>
Factory for a vullkan sampler.
Public Functions
-
SamplerBuilder &
AddressMode
(vk::SamplerAddressMode mode)¶ Mode of the sampler: repeat, clamp, etc.
- Return
- *this
- Parameters
mode
: vulkan mode
-
SamplerBuilder &
Filter
(vk::Filter filter)¶ Filter of the sampler: linear, nearest, etc.
- Return
- *this
- Parameters
filter
: vulkan filter
-
vk::UniqueSampler
Create
(vk::Device device)¶ Create the vulkan sampler.
- Return
- unique sampler
- Parameters
device
: vulkan device
-
SamplerBuilder &
-
struct
ShaderLayout
¶ - #include <DescriptorSet.h>
Represents the layout of a shader (vertex, fragment or compute)
-
class
Shape
: public Vortex2D::Renderer::Drawable, public Vortex2D::Renderer::Transformable¶ - #include <Shapes.h>
Shape interface which is drawable, transformable and has a color.
Subclassed by Vortex2D::Renderer::AbstractShape, Vortex2D::Renderer::Ellipse
-
struct
SpecConstInfo
¶ - #include <Pipeline.h>
Defines and holds value of the specification constants for shaders.
-
class
SpirvBinary
¶ - #include <Device.h>
A binary SPIRV shader, to be feed to vulkan.
-
class
Sprite
: public Vortex2D::Renderer::AbstractSprite¶ - #include <Sprite.h>
A sprite that renders a texture with a simple pass-through fragment shader.
Subclassed by Vortex2D::Fluid::Density
-
class
Texture
¶ - #include <Texture.h>
A texture, or in vulkan terms, an image.
Subclassed by Vortex2D::Renderer::RenderTexture
-
class
Timer
¶ - #include <Timer.h>
Calculates the ellapsed time on the GPU.
Public Functions
-
void
Start
(vk::CommandBuffer commandBuffer)¶ Start the timer after the current last command buffer.
- Parameters
commandBuffer
: command buffer to write timestamp
-
void
Stop
(vk::CommandBuffer commandBuffer)¶ Start the timer after the current last command buffer.
- Parameters
commandBuffer
: command buffer to write timestamp
-
void
Start
()¶ Start the timer after the current last command buffer.
-
void
Stop
()¶ Stop the timer after the current last command buffer.
-
uint64_t
GetElapsedNs
()¶ Get the elapsed time between the Start and Stop calls. Blocking function which will download the timestamps from the GPU.
- Return
- timestamp in nanoseconds.
-
void
-
struct
Transformable
¶ - #include <Transformable.h>
Class to represent the transformation of an object: position, scale, rotation and anchor.
Subclassed by Vortex2D::Fluid::Circle, Vortex2D::Fluid::Polygon, Vortex2D::Fluid::RigidBody, Vortex2D::Renderer::AbstractSprite, Vortex2D::Renderer::Shape
-
template<typename
T
>
classUniformBuffer
: public Vortex2D::Renderer::GenericBuffer¶ - #include <Buffer.h>
a uniform buffer type of buffer
-
template<typename
T
>
classVertexBuffer
: public Vortex2D::Renderer::GenericBuffer¶ - #include <Buffer.h>
a vertex buffer type of buffer
-
class
Work
¶ - #include <Work.h>
Represents a compute shader. It simplifies the process of binding, setting push constants and recording.
Public Functions
-
Work
(const Device &device, const ComputeSize &computeSize, const SpirvBinary &spirv, const SpecConstInfo &additionalSpecConstInfo = {})¶ Constructs an object using a SPIRV binary. It is not bound to any buffers or textures.
- Parameters
device
: vulkan devicecomputeSize
: the compute size. Can be a default one with size (1,1) or one with an actual size.spirv
: binary spirvadditionalSpecConstInfo
: additional specialization constants
-
Bound
Bind
(const std::vector<BindingInput> &inputs)¶ Bind the buffers and/or textures.
- Return
- a bound object, ready to be recorded in a command buffer.
- Parameters
inputs
: a list of buffers and/or textures
-
Bound
Bind
(ComputeSize computeSize, const std::vector<BindingInput> &inputs)¶ Bind the buffers and/or textures. This overrides the provided compute size in Work.
- Return
- a bound object, ready to be recorded in a command buffer.
- Parameters
computeSize
: the compute shader compute size.inputs
: a list of buffers and/or textures
-
class
Bound
¶ - #include <Work.h>
Is a bound version of Work. This means a buffer or texture was bound and this can be recorded in a command buffer.
Public Functions
-
template<typename ...
Args
>
voidPushConstant
(vk::CommandBuffer commandBuffer, Args&&... args)¶ Adds a constant value, i.e. a push constant.
- Parameters
commandBuffer
: the command buffer where the compute work will also be recorded.args
: the data to push. A total of 128 bytes can be used.
-
void
Record
(vk::CommandBuffer commandBuffer)¶ Record the compute work in this command buffer. This will also set two additional push constants: the 2D domain size.
- Parameters
commandBuffer
: the command buffer to record into.
-
void
RecordIndirect
(vk::CommandBuffer commandBuffer, IndirectBuffer<DispatchParams> &dispatchParams)¶ Record the compute work in this command buffer. Use the provided parameters to run the compute shader.
- Parameters
commandBuffer
: the command buffer to record into.dispatchParams
: the indirect buffer containing the parameters.
-
template<typename ...
-
-
namespace
Detail
¶ Functions
-
void
InsertSpecConst
(SpecConstInfo &specConstInfo)¶
-
template<typename
Arg
, typename ...Args
>
voidInsertSpecConst
(SpecConstInfo &specConstInfo, Arg &&arg, Args&&... args)¶
-
void
-
using
Fluid API reference¶
Classes¶
Vortex2D::Fluid::Advection
Vortex2D::Fluid::Circle
Vortex2D::Fluid::ConjugateGradient
Vortex2D::Fluid::Density
Vortex2D::Fluid::Depth
Vortex2D::Fluid::Diagonal
Vortex2D::Fluid::DistanceField
Vortex2D::Fluid::Extrapolation
Vortex2D::Fluid::GaussSeidel
Vortex2D::Fluid::IncompletePoisson
Vortex2D::Fluid::Jacobi
Vortex2D::Fluid::LevelSet
Vortex2D::Fluid::LinearSolver
Vortex2D::Fluid::LocalGaussSeidel
Vortex2D::Fluid::Multigrid
Vortex2D::Fluid::ParticleCount
Vortex2D::Fluid::Polygon
Vortex2D::Fluid::Preconditioner
Vortex2D::Fluid::Pressure
Vortex2D::Fluid::Rectangle
Vortex2D::Fluid::Reduce
Vortex2D::Fluid::ReduceJ
Vortex2D::Fluid::ReduceMax
Vortex2D::Fluid::ReduceSum
Vortex2D::Fluid::RigidBody
Vortex2D::Fluid::SmokeWorld
Vortex2D::Fluid::Transfer
Vortex2D::Fluid::Velocity
Vortex2D::Fluid::WaterWorld
Vortex2D::Fluid::World
API Reference¶
-
namespace
Fluid
¶ Enums
Functions
-
LinearSolver::Parameters
FixedParams
(unsigned iterations)¶ Create a linear solver parameters object with fixed solver type.
- Return
- parameters
- Parameters
iterations
: number of iterations to do
-
LinearSolver::Parameters
IterativeParams
(float errorTolerance)¶ Create a linear solver parameters object, solver will continue until error tolerance is reached.
- Return
- parameters
- Parameters
errorTolerance
: tolerance to reach before exiting
-
float
DefaultParticleSize
()¶
-
class
Advection
¶ - #include <Advection.h>
Advects particles, velocity field or any field using a velocity field.
Public Functions
-
Advection
(const Renderer::Device &device, const glm::ivec2 &size, float dt, Velocity &velocity, Velocity::InterpolationMode interpolationMode)¶ Initialize advection kernels and related object.
- Parameters
device
: vulkan devicesize
: size of velocity fielddt
: delta time for integrationvelocity
: velocity field
-
void
AdvectVelocity
()¶ Self advect velocity.
-
void
AdvectBind
(Density &density)¶ Binds a density field to be advected.
- Parameters
density
: density field
-
void
Advect
()¶ Performs an advection of the density field. Asynchronous operation.
-
void
AdvectParticleBind
(Renderer::GenericBuffer &particles, Renderer::Texture &levelSet, Renderer::IndirectBuffer<Renderer::DispatchParams> &dispatchParams)¶ Binds praticles to be advected. Also use a level set to project out the particles if they enter it.
- Parameters
particles
: particles to be advectedlevelSet
: level set to project out particlesdispatchParams
: contains number of particles
-
void
AdvectParticles
()¶ Advect particles. Asynchrounous operation.
-
-
class
Cfl
¶ - #include <Cfl.h>
Calculates the CFL number of the velocity field. It’s an indication on how to choose your time step size. Ideally, the time step should be smaller than the CFL number.
-
class
Circle
: public Vortex2D::Renderer::Transformable, public Vortex2D::Renderer::Drawable¶ - #include <Boundaries.h>
Signed distance field of circle.
Public Functions
-
Circle
(const Renderer::Device &device, float radius, float extent = 10.0f)¶ Initialize the circle with radius and extend of signed distance.
- Parameters
device
: vulkan device.radius
: radius of circle.extent
: extend how far from the circle the signed distance field is calculated.
-
void
Initialize
(const Renderer::RenderState &renderState)¶ Initialize the drawable for a particular state. This might include creating the correct pipeline. If it was already initialized, it will do nothing.
- Parameters
renderState
: the state to initialize with.
-
void
Update
(const glm::mat4 &projection, const glm::mat4 &view)¶ Update the MVP matrix of the drawable.
- Parameters
projection
: the projection matrixview
: the view matrix
-
void
Draw
(vk::CommandBuffer commandBuffer, const Renderer::RenderState &renderState)¶ Draw for the given render state. This has to be initialized before.
- Parameters
commandBuffer
: the command buffer to record into.renderState
: the render state to use.
-
-
class
ConjugateGradient
: public Vortex2D::Fluid::LinearSolver¶ - #include <ConjugateGradient.h>
An iterative preconditioned conjugate linear solver. The preconditioner can be specified.
Public Functions
-
ConjugateGradient
(const Renderer::Device &device, const glm::ivec2 &size, Preconditioner &preconditioner)¶ Initialize the solver with a size and preconditioner.
- Parameters
device
: vulkan devicesize
:preconditioner
:
-
void
Bind
(Renderer::GenericBuffer &d, Renderer::GenericBuffer &l, Renderer::GenericBuffer &b, Renderer::GenericBuffer &x)¶ Bind the buffers for the linear solver.
- Parameters
d
: the diagonal of the matrxil
: the lower matrixb
: the right hand sidex
: the unknowns
-
void
BindRigidbody
(float delta, Renderer::GenericBuffer &d, RigidBody &rigidBody)¶ Bind rigidbody with the linear solver’s matrix.
- Parameters
delta
: solver deltad
: diagonal matrixrigidBody
: rigidbody to bind to
-
void
Solve
(Parameters ¶ms, const std::vector<RigidBody *> &rigidbodies = {})¶ Solve iteratively solve the linear equations in data.
-
float
GetError
()¶ - Return
- the max error
-
-
class
Density
: public Vortex2D::Renderer::RenderTexture, public Vortex2D::Renderer::Sprite¶ - #include <Density.h>
Density field, used to represent smoke swirling.
-
class
Depth
¶ - #include <Multigrid.h>
Contains the sizes of the multigrid hierarchy.
Public Functions
-
Depth
(const glm::ivec2 &size)¶ Initialize with the finest size.
- Parameters
size
: the base size.
-
int
GetMaxDepth
() const¶ The calculated depth of the multigrid.
- Return
- the depth.
-
glm::ivec2
GetDepthSize
(std::size_t i) const¶ Gets the depth for a given level.
- Return
- the size
- Parameters
i
: the level
-
-
class
Diagonal
: public Vortex2D::Fluid::Preconditioner¶ - #include <Diagonal.h>
Diagonal preconditioner. Simplest of preconditioner, useful to verify if the preconditioned conjugate gradient works.
Public Functions
-
void
Bind
(Renderer::GenericBuffer &d, Renderer::GenericBuffer &l, Renderer::GenericBuffer &b, Renderer::GenericBuffer &x)¶ Bind the linear equation buffers.
- Parameters
d
: the diagonal of the matrixl
: the lower matrixb
: the right hand sidex
: the unknown buffer
-
void
Record
(vk::CommandBuffer commandBuffer)¶ Record the preconditioner.
- Parameters
commandBuffer
: the command buffer to record into.
-
void
-
class
DistanceField
: public Vortex2D::Renderer::AbstractSprite¶ - #include <Boundaries.h>
Sprite of a distance field.
Public Functions
-
DistanceField
(const Renderer::Device &device, Renderer::RenderTexture &levelSet, float scale = 1.0f)¶ Initialize the price with the level set and scale.
- Parameters
device
: vulkan devicelevelSet
: level set to use as spritescale
: scale of the level set
-
void
Draw
(vk::CommandBuffer commandBuffer, const Renderer::RenderState &renderState)¶ Draw for the given render state. This has to be initialized before.
- Parameters
commandBuffer
: the command buffer to record into.renderState
: the render state to use.
-
-
class
Extrapolation
¶ - #include <Extrapolation.h>
Class to extrapolate values into the neumann and/or dirichlet boundaries.
Public Functions
-
void
Extrapolate
()¶ Will extrapolate values from buffer into the dirichlet and neumann boundaries.
-
void
ConstrainBind
(Renderer::Texture &solidPhi)¶ Binds a solid level set to use later and constrain the velocity against.
- Parameters
solidPhi
: solid level set
-
void
ConstrainVelocity
()¶ Constrain the velocity, i.e. ensure that the velocity normal to the solid level set is 0.
-
void
-
class
GaussSeidel
: public Vortex2D::Fluid::LinearSolver, public Vortex2D::Fluid::Preconditioner¶ - #include <GaussSeidel.h>
An iterative black and red successive over relaxation linear solver.
Public Functions
-
void
Bind
(Renderer::GenericBuffer &d, Renderer::GenericBuffer &l, Renderer::GenericBuffer &b, Renderer::GenericBuffer &x)¶ Bind the buffers for the linear solver.
- Parameters
d
: the diagonal of the matrxil
: the lower matrixb
: the right hand sidex
: the unknowns
-
void
BindRigidbody
(float delta, Renderer::GenericBuffer &d, RigidBody &rigidBody)¶ Bind rigidbody with the linear solver’s matrix.
- Parameters
delta
: solver deltad
: diagonal matrixrigidBody
: rigidbody to bind to
-
void
Solve
(Parameters ¶ms, const std::vector<RigidBody *> &rigidbodies = {})¶ Iterative solving of the linear equations in data.
-
float
GetError
()¶ - Return
- the max error
-
void
Record
(vk::CommandBuffer commandBuffer)¶ Record the preconditioner.
- Parameters
commandBuffer
: the command buffer to record into.
-
void
Record
(vk::CommandBuffer commandBuffer, int iterations)¶ Record a determined number of iterations.
- Parameters
commandBuffer
:iterations
:
-
void
SetW
(float w)¶ Set the w factor of the GS iterations : x_new = w * x_new + (1-w) * x_old.
- Parameters
w
:
-
void
SetPreconditionerIterations
(int iterations)¶ set number of iterations to be used when GS is a preconditioner
- Parameters
iterations
:
-
void
-
class
IncompletePoisson
: public Vortex2D::Fluid::Preconditioner¶ - #include <IncompletePoisson.h>
Incomplete poisson preconditioner. Slightly better than a simple diagonal preconditioner.
Public Functions
-
void
Bind
(Renderer::GenericBuffer &d, Renderer::GenericBuffer &l, Renderer::GenericBuffer &b, Renderer::GenericBuffer &x)¶ Bind the linear equation buffers.
- Parameters
d
: the diagonal of the matrixl
: the lower matrixb
: the right hand sidex
: the unknown buffer
-
void
Record
(vk::CommandBuffer commandBuffer)¶ Record the preconditioner.
- Parameters
commandBuffer
: the command buffer to record into.
-
void
-
class
Jacobi
: public Vortex2D::Fluid::Preconditioner¶ - #include <Jacobi.h>
An iterative jacobi linear solver.
Public Functions
-
void
Bind
(Renderer::GenericBuffer &d, Renderer::GenericBuffer &l, Renderer::GenericBuffer &b, Renderer::GenericBuffer &x)¶ Bind the linear equation buffers.
- Parameters
d
: the diagonal of the matrixl
: the lower matrixb
: the right hand sidex
: the unknown buffer
-
void
Record
(vk::CommandBuffer commandBuffer)¶ Record the preconditioner.
- Parameters
commandBuffer
: the command buffer to record into.
-
void
SetW
(float w)¶ Set the w factor of the GS iterations : x_new = w * x_new + (1-w) * x_old.
- Parameters
w
:
-
void
SetPreconditionerIterations
(int iterations)¶ set number of iterations to be used when GS is a preconditioner
- Parameters
iterations
:
-
void
-
class
LevelSet
: public Vortex2D::Renderer::RenderTexture¶ - #include <LevelSet.h>
A signed distance field, which can be re-initialized. In other words, a level set.
Public Functions
-
void
Reinitialise
()¶ Reinitialise the level set, i.e. ensure it is a correct signed distance field.
-
void
ShrinkWrap
()¶ Shrink wrap wholes.
-
void
ExtrapolateBind
(Renderer::Texture &solidPhi)¶ Bind a solid level set, which will be used to extrapolate into this level set.
- Parameters
solidPhi
:
-
void
Extrapolate
()¶ Extrapolate this level set into the solid level set it was attached to. This only performs a single cell extrapolation.
-
void
-
struct
LinearSolver
¶ - #include <LinearSolver.h>
An interface to represent a linear solver.
Subclassed by Vortex2D::Fluid::ConjugateGradient, Vortex2D::Fluid::GaussSeidel, Vortex2D::Fluid::Multigrid
Public Functions
-
virtual void
Bind
(Renderer::GenericBuffer &d, Renderer::GenericBuffer &l, Renderer::GenericBuffer &b, Renderer::GenericBuffer &x) = 0¶ Bind the buffers for the linear solver.
- Parameters
d
: the diagonal of the matrxil
: the lower matrixb
: the right hand sidex
: the unknowns
-
virtual void
BindRigidbody
(float delta, Renderer::GenericBuffer &d, RigidBody &rigidBody) = 0¶ Bind rigidbody with the linear solver’s matrix.
- Parameters
delta
: solver deltad
: diagonal matrixrigidBody
: rigidbody to bind to
-
virtual void
Solve
(Parameters ¶ms, const std::vector<RigidBody *> &rigidBodies = {}) = 0¶ Solves the linear equations.
- Parameters
params
: solver iteration/error parametersrigidBodies
: rigidbody to include in solver’s matrix
-
virtual float
GetError
() = 0¶ - Return
- the max error
-
struct
Data
¶ - #include <LinearSolver.h>
The various parts of linear equations.
-
struct
DebugCopy
¶ - #include <LinearSolver.h>
Copies the linear solver data in the debug linear solver data.
Public Functions
-
void
Copy
()¶ Copies the linear solver data in the debug linear solver data.
-
void
-
struct
DebugData
¶ - #include <LinearSolver.h>
Contains the linear equations as texture, so it can easily be visualised in RenderDoc.
-
class
Error
¶ - #include <LinearSolver.h>
Calculates the max residual error of the linear system.
Public Functions
-
void
Bind
(Renderer::GenericBuffer &d, Renderer::GenericBuffer &l, Renderer::GenericBuffer &div, Renderer::GenericBuffer &pressure)¶ Bind the linear system.
- Parameters
d
: the diagonal of the matrxil
: the lower matrixb
: the right hand sidex
: the unknowns
-
float
GetError
()¶ Get the maximum error.
- Return
- The error.
-
void
-
struct
Parameters
¶ - #include <LinearSolver.h>
Parameters for an iterative linear solvers.
Public Types
Public Functions
-
Parameters
(SolverType type, unsigned iterations, float errorTolerance = 0.0f)¶ Construct parameters with max iterations and max error.
- Parameters
type
: fixed or iterative type of solveriterations
: max number of iterations to performerrorTolerance
: solver stops when the error is smaller than this.
-
bool
IsFinished
(float initialError) const¶ Checks if we’ve reacched the parameters.
- Return
- if we can stop the linear solver.
- Parameters
initialError
: the initial error
-
void
Reset
()¶ Sets the out error and out iterations to 0.
-
-
virtual void
-
class
LocalGaussSeidel
: public Vortex2D::Fluid::Preconditioner¶ - #include <GaussSeidel.h>
A version of the gauss seidel that can only be applied on sizes (16,16) or smaller.
Public Functions
-
void
Bind
(Renderer::GenericBuffer &d, Renderer::GenericBuffer &l, Renderer::GenericBuffer &b, Renderer::GenericBuffer &x)¶ Bind the linear equation buffers.
- Parameters
d
: the diagonal of the matrixl
: the lower matrixb
: the right hand sidex
: the unknown buffer
-
void
Record
(vk::CommandBuffer commandBuffer)¶ Record the preconditioner.
- Parameters
commandBuffer
: the command buffer to record into.
-
void
-
class
Multigrid
: public Vortex2D::Fluid::LinearSolver, public Vortex2D::Fluid::Preconditioner¶ - #include <Multigrid.h>
Multigrid preconditioner. It creates a hierarchy of twice as small set of linear equations. It applies a few iterations of jacobi on each level and transfers the error on the level above. It then copies the error down, adds to the current solution and apply a few more iterations of jacobi.
Public Functions
-
Multigrid
(const Renderer::Device &device, const glm::ivec2 &size, float delta, int numSmoothingIterations = 3, SmootherSolver smoother = SmootherSolver::Jacobi)¶ Initialize multigrid for given size and delta.
- Parameters
device
: vulkan devicesize
: of the linear equationsdelta
: timestep delta
-
void
Bind
(Renderer::GenericBuffer &d, Renderer::GenericBuffer &l, Renderer::GenericBuffer &b, Renderer::GenericBuffer &x)¶ Bind the buffers for the linear solver.
- Parameters
d
: the diagonal of the matrxil
: the lower matrixb
: the right hand sidex
: the unknowns
-
void
BuildHierarchiesBind
(Pressure &pressure, Renderer::Texture &solidPhi, Renderer::Texture &liquidPhi)¶ Bind the level sets from which the hierarchy is built.
- Parameters
pressure
: The current linear equationssolidPhi
: the solid level setliquidPhi
: the liquid level set
-
void
BuildHierarchies
()¶ Computes the hierarchy to be used by the multigrid. Asynchronous operation.
-
void
Record
(vk::CommandBuffer commandBuffer)¶ Record the preconditioner.
- Parameters
commandBuffer
: the command buffer to record into.
-
void
BindRigidbody
(float delta, Renderer::GenericBuffer &d, RigidBody &rigidBody)¶ Bind rigidbody with the linear solver’s matrix.
- Parameters
delta
: solver deltad
: diagonal matrixrigidBody
: rigidbody to bind to
-
void
Solve
(Parameters ¶ms, const std::vector<RigidBody *> &rigidBodies = {})¶ Solves the linear equations.
- Parameters
params
: solver iteration/error parametersrigidBodies
: rigidbody to include in solver’s matrix
-
float
GetError
()¶ - Return
- the max error
-
-
class
ParticleCount
: public Vortex2D::Renderer::RenderTexture¶ - #include <Particles.h>
Container for particles used in the advection of the fluid simulation. Also a level set that is built from the particles.
Public Functions
-
void
Scan
()¶ Count the number of particles and update the internal data structures.
-
int
GetTotalCount
()¶ Calculate the total number of particles and return it.
- Return
-
Renderer::IndirectBuffer<Renderer::DispatchParams> &
GetDispatchParams
()¶ Calculate the dispatch parameters to use on the particle buffer.
- Return
-
void
LevelSetBind
(LevelSet &levelSet)¶ Bind a solid level set, which will be used to interpolate the particles out of.
- Parameters
levelSet
:
-
void
Phi
()¶ Calculate the level set from the particles.
-
void
VelocitiesBind
(Velocity &velocity, Renderer::GenericBuffer &valid)¶ Bind the velocities, used for advection of the particles.
- Parameters
velocity
:valid
:
-
void
TransferToGrid
()¶ Interpolate the velocities of the particles to the velocities field.
-
void
TransferFromGrid
()¶ Interpolate the velocities field in to the particles’ velocity.
-
void
-
class
Polygon
: public Vortex2D::Renderer::Transformable, public Vortex2D::Renderer::Drawable¶ - #include <Boundaries.h>
Signed distance field of a poylgon.
Subclassed by Vortex2D::Fluid::Rectangle
Public Functions
-
Polygon
(const Renderer::Device &device, std::vector<glm::vec2> points, bool inverse = false, float extent = 10.0f)¶ Initialize polygon with set of points and extent of signed distance.
- Parameters
device
: vulkan devicepoints
: clockwise oriented set of points (mininum 3).inverse
: flag if the distance field should be inversed.extent
: extend how far from the poylon the signed distance field is calculated.
-
void
Initialize
(const Renderer::RenderState &renderState)¶ Initialize the drawable for a particular state. This might include creating the correct pipeline. If it was already initialized, it will do nothing.
- Parameters
renderState
: the state to initialize with.
-
void
Update
(const glm::mat4 &projection, const glm::mat4 &view)¶ Update the MVP matrix of the drawable.
- Parameters
projection
: the projection matrixview
: the view matrix
-
void
Draw
(vk::CommandBuffer commandBuffer, const Renderer::RenderState &renderState)¶ Draw for the given render state. This has to be initialized before.
- Parameters
commandBuffer
: the command buffer to record into.renderState
: the render state to use.
-
-
struct
Preconditioner
¶ - #include <Preconditioner.h>
An interface to represent a linear solver preconditioner.
Subclassed by Vortex2D::Fluid::Diagonal, Vortex2D::Fluid::GaussSeidel, Vortex2D::Fluid::IncompletePoisson, Vortex2D::Fluid::Jacobi, Vortex2D::Fluid::LocalGaussSeidel, Vortex2D::Fluid::Multigrid
Public Functions
-
virtual void
Bind
(Renderer::GenericBuffer &d, Renderer::GenericBuffer &l, Renderer::GenericBuffer &b, Renderer::GenericBuffer &x) = 0¶ Bind the linear equation buffers.
- Parameters
d
: the diagonal of the matrixl
: the lower matrixb
: the right hand sidex
: the unknown buffer
-
virtual void
Record
(vk::CommandBuffer commandBuffer) = 0¶ Record the preconditioner.
- Parameters
commandBuffer
: the command buffer to record into.
-
virtual void
-
class
PrefixScan
¶ - #include <PrefixScan.h>
The prefix sum operator.
void PrefixSym(int input[], int n, int output[]) { output[0] = input[0]; for (int i = 1; i < n; i++) output[i] = output[i-1] + input[i]; }
-
class
Bound
¶ - #include <PrefixScan.h>
A prefix scan object bound with input/output buffers, ready to be dispatched.
-
class
-
class
Pressure
¶ - #include <Pressure.h>
build the linear equation and compute the divergence from the resulting solution.
Public Functions
-
Renderer::Work::Bound
BindMatrixBuild
(const glm::ivec2 &size, Renderer::GenericBuffer &diagonal, Renderer::GenericBuffer &lower, Renderer::Texture &liquidPhi, Renderer::Texture &solidPhi)¶ Bind the various buffes for the linear system Ax = b.
- Return
- Parameters
size
: size of the linear systemdiagonal
: diagonal of Alower
: lower matrix of AliquidPhi
: liquid level setsolidPhi
: solid level set
-
void
BuildLinearEquation
()¶ Build the matrix A and right hand side b.
-
void
ApplyPressure
()¶ Apply the solution of the equation Ax = b, i.e. the pressure to the velocity to make it non-divergent.
-
Renderer::Work::Bound
-
class
Rectangle
: public Vortex2D::Fluid::Polygon¶ - #include <Boundaries.h>
Signed distance field of a rectangle.
Public Functions
-
Rectangle
(const Renderer::Device &device, const glm::vec2 &size, bool inverse = false, float extent = 10.0f)¶ Initialize rectangle with size and extend of signed distance.
- Parameters
device
: vulkan device.size
: rectangle sizeinverse
: flag if the distance field should be inverted.extent
: extent how far from the rectangle the signed distance field is calculated.
-
void
Initialize
(const Renderer::RenderState &renderState)¶ Initialize the drawable for a particular state. This might include creating the correct pipeline. If it was already initialized, it will do nothing.
- Parameters
renderState
: the state to initialize with.
-
void
Update
(const glm::mat4 &projection, const glm::mat4 &view)¶ Update the MVP matrix of the drawable.
- Parameters
projection
: the projection matrixview
: the view matrix
-
void
Draw
(vk::CommandBuffer commandBuffer, const Renderer::RenderState &renderState)¶ Draw for the given render state. This has to be initialized before.
- Parameters
commandBuffer
: the command buffer to record into.renderState
: the render state to use.
-
-
class
Reduce
¶ - #include <Reduce.h>
Parallel reduction of a buffer into one value. The operator and type of data is specified by inheriting the class.
Subclassed by Vortex2D::Fluid::ReduceJ, Vortex2D::Fluid::ReduceMax, Vortex2D::Fluid::ReduceSum
Public Functions
-
Reduce::Bound
Bind
(Renderer::GenericBuffer &input, Renderer::GenericBuffer &output)¶ Bind the reduce operation.
- Return
- a bound object that can be recorded in a command buffer.
- Parameters
input
: input bufferoutput
: output buffer
-
Reduce::Bound
-
class
ReduceJ
: public Vortex2D::Fluid::Reduce¶ - #include <Reduce.h>
Reduce operation on a struct with a 2d vector and 1 float (i.e. 3 floats) with addition.
-
class
ReduceMax
: public Vortex2D::Fluid::Reduce¶ - #include <Reduce.h>
Reduce operation on float with max of absolute.
-
class
ReduceSum
: public Vortex2D::Fluid::Reduce¶ - #include <Reduce.h>
Reduce operation on float with addition.
-
class
RigidBody
: public Vortex2D::Renderer::Transformable¶ - #include <Rigidbody.h>
Rigidbody that can interact with the fluid: either be push by it, or influence it, or both.
Public Functions
-
virtual void
ApplyForces
()¶ function to override and apply forces from this rigidbody to the external rigidbody
-
virtual void
ApplyVelocities
()¶ Override and apply velocities from the external rigidbody to the this rigidbody.
-
void
SetMassData
(float mass, float inertia)¶ Sets the mass and inertia of the rigidbody.
- Parameters
mass
: of the bodyinertia
: of the body
-
void
SetVelocities
(const glm::vec2 &velocity, float angularVelocity)¶ sets the velocities and angular velocities of the body
- Parameters
velocity
:angularVelocity
:
-
void
UpdatePosition
()¶ Upload the transform matrix to the GPU.
-
void
RenderPhi
()¶ Render the current object orientation in an internal texture and the external one.
-
void
BindPhi
(Renderer::RenderTexture &phi)¶ Bind the rendertexture where this rigidbodies shape will be rendered.
- Parameters
phi
: render texture of the world
-
void
BindDiv
(Renderer::GenericBuffer &div, Renderer::GenericBuffer &diagonal)¶ Bind a the right hand side and diagonal of the linear system Ax = b. This is to apply the rigid body influence to the system.
- Parameters
div
: right hand side of the linear system Ax=bdiagonal
: diagonal of matrix A
-
void
BindVelocityConstrain
(Fluid::Velocity &velocity)¶ Bind velocities to constrain based on the body’s velocity.
- Parameters
velocity
:
-
void
BindForce
(Renderer::GenericBuffer &d, Renderer::GenericBuffer &pressure)¶ Bind pressure, to have the pressure update the body’s forces.
- Parameters
d
: diagonal of matrix Apressure
: solved pressure buffer
-
void
BindPressure
(float delta, Renderer::GenericBuffer &d, Renderer::GenericBuffer &s, Renderer::GenericBuffer &z)¶ Bind pressure, to have the pressure update the body’s forces.
- Parameters
delta
:d
:s
:z
:
-
void
Div
()¶ Apply the body’s velocities to the linear equations matrix A and right hand side b.
-
void
Force
()¶ Apply the pressure to body, updating its forces.
-
void
VelocityConstrain
()¶ Constrain the velocities field based on the body’s velocity.
-
vk::Flags<Type>
GetType
()¶ Type of this body.
- Return
-
void
SetType
(Type type)¶ Set the type of the body.
- Parameters
type
:
-
Renderer::RenderTexture &
Phi
()¶ the local level set of the body
- Return
-
virtual void
-
class
RigidBodySolver
¶ - #include <Rigidbody.h>
Interface to call the external rigidbody solver.
Public Functions
-
virtual void
Step
(float delta) = 0¶ perfoms a single step of the solver.
- Parameters
delta
: of simulation
-
virtual void
-
class
SmokeWorld
: public Vortex2D::Fluid::World¶ - #include <World.h>
A concrete implementation of World to simulate ‘smoke’, or more accurately dye in a liquid. The liquid cannot change location or size.
-
class
Transfer
¶ - #include <Transfer.h>
Prolongates or restrict a level set on a finer or coarser level set.
Public Functions
-
Transfer
(const Renderer::Device &device)¶ Initialize prolongate and restrict compute pipelines.
- Parameters
device
:
-
void
ProlongateBind
(std::size_t level, const glm::ivec2 &fineSize, Renderer::GenericBuffer &fine, Renderer::GenericBuffer &fineDiagonal, Renderer::GenericBuffer &coarse, Renderer::GenericBuffer &coarseDiagonal)¶ Prolongate a level set on a finer level set. Setting the 4 cells to the value of the coarser grid. Multiple level sets can be bound and indexed.
- Parameters
level
: the index of the bound level set to prolongatefineSize
: size of the finer level setfine
: the finer level setfineDiagonal
: the diagonal of the linear equation matrix at sizefineSize
coarse
: the coarse level setcoarseDiagonal
: the diagonal of the linear equation matrix at size half offineSize
-
void
RestrictBind
(std::size_t level, const glm::ivec2 &fineSize, Renderer::GenericBuffer &fine, Renderer::GenericBuffer &fineDiagonal, Renderer::GenericBuffer &coarse, Renderer::GenericBuffer &coarseDiagonal)¶ Restricing the level set on a coarser level set. Averages 4 cells into one. Multiple level sets can be bound and indexed.
- Parameters
level
: the index of the bound level set to prolongatefineSize
: size of the finer level setfine
: the finer level setfineDiagonal
: the diagonal of the linear equation matrix at sizefineSize
coarse
: the coarse level setcoarseDiagonal
: the diagonal of the linear equation matrix at size half offineSize
-
void
Prolongate
(vk::CommandBuffer commandBuffer, std::size_t level)¶ Prolongate the level set, using the bound level sets at the specified index.
- Parameters
commandBuffer
: command buffer to record into.level
: index of bound level sets.
-
void
Restrict
(vk::CommandBuffer commandBuffer, std::size_t level)¶ Restrict the level set, using the bound level sets at the specified index.
- Parameters
commandBuffer
: command buffer to record into.level
: index of bound level sets.
-
-
class
Velocity
: public Vortex2D::Renderer::RenderTexture¶ - #include <Velocity.h>
The Velocity field. Can be used to calculate a difference between different states. Contains three fields: intput and output, used for ping-pong algorithms, and d, the difference between two velocity fields.
Public Types
Public Functions
-
Renderer::Texture &
Output
()¶ An output texture used for algorithms that used the velocity as input and need to create a new velocity field.
- Return
-
Renderer::Texture &
D
()¶ A difference velocity field, calculated with the difference between this velocity field, and the output velocity field.
- Return
-
void
CopyBack
(vk::CommandBuffer commandBuffer)¶ Copy the output field to the main field.
- Parameters
commandBuffer
:
-
void
Clear
(vk::CommandBuffer commandBuffer)¶ Clear the velocity field.
- Parameters
commandBuffer
:
-
void
SaveCopy
()¶ Copy to the difference field.
-
void
VelocityDiff
()¶ Calculate the difference between the difference field and this velocity field, store it in the diference field.
-
Renderer::Texture &
-
class
WaterWorld
: public Vortex2D::Fluid::World¶ - #include <World.h>
A concrete implementation of World to simulate water.
Public Functions
-
Renderer::RenderCommand
RecordParticleCount
(Renderer::RenderTarget::DrawableList drawables)¶ The water simulation uses particles to define the water area. In fact, the level set is built from the particles. This means to be able to set an area, we can’t use RecordLiquidPhi. To define the particle area, simply draw a regular shape. The colour r is used to determine if we add or remove particles, use r = 4 to add and r = -4 to remove.
- Return
- render command
- Parameters
drawables
: list of drawables object with colour 8 or -8
-
void
ParticlePhi
()¶ Using the particles, create a level set (phi) encompassing all the particles. This can be viewed with LiquidDistanceField.
-
Renderer::RenderCommand
-
class
World
¶ - #include <World.h>
The main class of the framework. Each instance manages a grid and this class is used to set forces, define boundaries, solve the incompressbility equations and do the advection.
Subclassed by Vortex2D::Fluid::SmokeWorld, Vortex2D::Fluid::WaterWorld
Public Functions
-
World
(const Renderer::Device &device, const glm::ivec2 &size, float dt, int numSubSteps = 1, Velocity::InterpolationMode interpolationMode = Velocity::InterpolationMode::Linear)¶ Construct an Engine with a size and time step.
- Parameters
device
: vulkan devicesize
: dimensions of the simulationdt
: timestamp of the simulation, e.g. 0.016 for 60FPS simulations.numSubSteps
: the number of sub-steps to perform per step call. Reduces loss of fluid.
-
void
Step
(LinearSolver::Parameters ¶ms)¶ Perform one step of the simulation.
-
Renderer::RenderCommand
RecordVelocity
(Renderer::RenderTarget::DrawableList drawables, VelocityOp op)¶ Record drawables to the velocity field. The colour (r,g) will be used as the velocity (x, y)
- Return
- render command
- Parameters
drawables
: a list of drawable fieldop
: operation of the render: add velocity or set velocity
-
void
SubmitVelocity
(Renderer::RenderCommand &renderCommand)¶ submit the render command created with RecordVelocity
- Parameters
renderCommand
: the render command
-
Renderer::RenderCommand
RecordLiquidPhi
(Renderer::RenderTarget::DrawableList drawables)¶ Record drawables to the liquid level set, i.e. to define the fluid area. The drawables need to make a signed distance field, if not the result is undefined.
- Return
- render command
- Parameters
drawables
: a list of signed distance field drawables
-
Renderer::RenderCommand
RecordStaticSolidPhi
(Renderer::RenderTarget::DrawableList drawables)¶ Record drawables to the solid level set, i.e. to define the boundary area. The drawables need to make a signed distance field, if not the result is undefined.
- Return
- render command
- Parameters
drawables
: a list of signed distance field drawables
-
DistanceField
LiquidDistanceField
()¶ Create sprite that can be rendered to visualize the liquid level set.
- Return
- a sprite
-
DistanceField
SolidDistanceField
()¶ Create sprite that can be rendered to visualize the solid level set.
- Return
- a sprite
-
void
RemoveRigidBody
(RigidBody &rigidbody)¶ Remove a rigidbody from the solver.
- Parameters
rigidbody
:
-
void
AttachRigidBodySolver
(RigidBodySolver &rigidbodySolver)¶ Attach a rigidbody solver, e.g. box2d.
- Parameters
rigidbodySolver
:
-
float
GetCFL
()¶ Calculate the CFL number, i.e. the width divided by the max velocity.
- Return
- CFL number
-
-
LinearSolver::Parameters