如何将变量从顶点着色器移动到几何着色器？

Question

我具有以下顶点着色器：

#version 330

layout (location = 0) in ivec4 inOHLC;
layout (location = 1) in int inVolume;
layout (location = 2) in int inTimestamp;

out ivec4 outOHLC;
out int outVolume;
out int outTimestamp;

void main()
{
    outOHLC = inOHLC;
    outVolume= inVolume;
    outTimestamp = inTimestamp;
}

并且我想在几何着色器中接收outOHLC，outVolume和outTimestamp。我对此进行编码：

#version 330

in ivec4 inOHLC;
in int inVolume;
in int inTimestamp;

layout (line_strip, max_vertices = 2) out;

void main() {  

    float x = (float)inTimestamp / 100.0;
    float y1 = (float)inOHLC[1] / 100.0;
    float y2 = (float)inOHLC[1] / 100.0;
    gl_Position = vec4(x, y1, 0.0, 0.0);
    EmitVertex();    
    gl_Position = vec4(x, y2, 0.0, 0.0);
    EmitVertex();    
    EndPrimitive();

}  

但出现以下错误：

run:
. . . vertex compilation success.
. . . geometry compilation failed.
Shader Info Log: 
ERROR: 7:1: 'inOHLC' : geometry shader input varying variable must be declared as an array
ERROR: 8:1: 'inVolume' : geometry shader input varying variable must be declared as an array
ERROR: 9:1: 'inTimestamp' : geometry shader input varying variable must be declared as an array
ERROR: 15:1: ')' : syntax error syntax error

---

在@ Rabbid76注释之后，我将代码修改如下：

#version 330

in ivec4 ohlc[];
in int volume[];
in int timestamp[];

layout (line_strip, max_vertices = 2) out;

void main()
{

    float x = (float)timestamp / 100.0;
    float y1 = (float)ohlc[1] / 100.0;
    float y2 = (float)ohlc[2] / 100.0;
    gl_Position = vec4(x, y1, 0.0, 0.0);
    EmitVertex();    
    gl_Position = vec4(x, y2, 0.0, 0.0);
    EmitVertex();    
    EndPrimitive();

} 

现在我只收到一个错误：

run:
. . . vertex compilation success.
. . . geometry compilation failed.
Shader Info Log: 
ERROR: 40:1: ')' : syntax error syntax error

我想这与(float)强制转换（甚至在GLSL中存在吗？）或浮点变量定义有关。

Answer 1

仔细查看消息：

几何着色器输入变化变量必须声明为数组

Geometry shader的输入是基元。这意味着构成原始图元的顶点着色器的所有输出都组成了。因此，几何着色器的输入将重新排列：

in ivec4 inOHLC[];
in int inVolume[];
in int inTimestamp[];

此外，您必须为几何着色器指定input primitive type。例如一行：

layout(lines​) in;

原始类型lines表示几何着色器接收2个顶点（输入的数组大小为2）。

---

GLSL没有像C这样的转换运算符。如果要将int转换为float，则必须构造一个新的float。重载的float构造函数接受int参数：

float y1 = (float)ohlc[1] / 100.0;

float y1 = float(ohlc[1]) / 100.0;

---

索引从0开始而不是1（例如在C，C ++，C＃，Java，JavaScript，Python等中：）]

float y1 = float(ohlc[0]) / 100.0;
float y2 = float(ohlc[1]) / 100.0;