Custom HLSL Nodes (block and operator)
These Custom HLSL nodes let you execute custom HLSL code during particle simulation. You can use an operator for horizontal flow or block for vertical flow (in contexts). To be valid and correctly interpreted by the VFX Graph some conventions must be adopted.
Node settings
Setting name | UI | Location | Action |
---|---|---|---|
Name | Text field | Inspector | Choose the name of the block |
HLSL Code | Button | Graph node | Opens a code editor window |
Available Function | Drop down | Graph node | Pick which function to execute |
HLSL Code
The HLSL code can be either embedded in the node or an HLSL file can be used.
In both cases you are allowed to include any other valid HLSL file (using #include "
Function declaration
To be properly recognized by VFX Graph the function must fulfill the following requirements:
- Return a supported type Supported types
- Each function parameter must have the
in
,out
orinout
access modifier - Each function parameter must be of a Supported types
- If you declare multiple functions, they must have unique names.
- Function can take a maximum of 4 parameters
Inline documentation
You can specify a tooltip for each function parameter using the three slash comment notation as shown below:
/// <parameter-name>: the tooltip's text
These comments must be right above the function declaration.
/// a: the tooltip for parameter a
/// b: the tooltip for parameter b
float Distance(in float3 a, in float3 b)
{
return distance(a, b);
}
You may want to write some helper function that you don't want to be exposed in the node's choice list. In that case, simply put this special comment:
/// Hidden
Supported types
HLSL Type | Port Type | Description |
---|---|---|
bool | bool | Simple scalar value represented as a boolean. |
uint | uint | Simple scalar value represented as an uint. |
int | int | Simple scalar value represented as a int. |
float | float | Simple scalar value represented as a float. |
float2 | Vector2 | A structure containing two float. |
float3 | Vector3 | A structure containing three float. |
float4 | Vector4 | A structure containing four float. |
float4x4 | Matrix4x4 | A structure representing a matrix. |
VFXSampler2D | Texture2D | A two dimensions texture. |
VFXSampler3D | Texture3D | A three dimensions texture. |
VFXGradient | Gradient | A three dimensions texture. |
VFXCurve | AnimationCurve | A structure that describe a curve that can be sampled |
StructuredBuffer | GraphicsBuffer | A structure that describe a curve that can be sampled |
ByteAddressBuffer | GraphicsBuffer | A structure that describe a curve that can be sampled |
Sampling
Textures
To sample a texture you must use the VFX Graph structure called VFXSampler2D (or VFXSample3D) which is defined as shown below:
struct VFXSampler2D
{
Texture2D t;
SamplerState s;
};
The easiest way to sample a texture is to use a function provided by the VFX Graph common HLSL code: SampleTexture(VFXSampler2D texure, float2 coordinates)
.
But you can also use HLSL built-in functions to sample a texture using the VFXSampler2D fields.
In that case, since this is used in a compute shader you must specify which mipmap level to sample (use SampleLevel
for instance).
Buffers
You can use two types of buffers: ByteAddressBuffer
and StructuredBuffer<>
.
In both cases the usage is the same as in any HLSL code:
ByteAddressBuffer
: use theLoad
function
uint char = buffer.Load(attributes.particleId % count);
StructuredBuffer<>
: use classic index accessor
float angle = phase + freq * buffer[attributes.particleId % bufferSize];
Gradient
Gradients are handled specifically in VFX Graph (they are packed in a single texture) so you must use a dedicated function to sample them.
Here is the function definition: SampleGradient(VFXGradient gradient, float t)
float3 color = SampleGradient(grad, t);
Curve
Sampling a curve is really similar to sampling a gradient.
Here is the function definition: SampleCurve(VFXCurve curve, float t)
float r = SampleCurve(curve, t);