[Question] Why position can assign for texcoord?

Unity Engine
,
1

I am learning Unlit Shader. I have the code:

CGPROGRAM
            #pragma vertex vert
            #pragma fragment frag
            #include "UnityCG.cginc"

            struct appdata
            {
                float4 vertex : POSITION;
                float2 uv : TEXCOORD0;
            };

            struct v2f
            {
                float2 uv : TEXCOORD0;
                float4 position: TEXCOORD1;
                float4 vertex : SV_POSITION;
            };

            v2f vert (appdata v)
            {
                v2f o;
                o.vertex = UnityObjectToClipPos(v.vertex);
                o.uv = v.uv;
                o.position = v.vertex;
                return o;
            }

            fixed4 frag (v2f i) : SV_Target
            {
                float inCircle = step(0.25, length(i.position.xy));
                half4 col = (1, 1, 1, 1) * inCircle;
                return col;
            }
            ENDCG

why o.position is TEXCOORD1 = v.vertex is POSITION?
I think the vertex includes UV, right?

2

TEXCOORD# semantics are used to pass arbitrary data between the vertex and fragment shaders. It’s named “texcoord” because when the semantics where first thought up, they didn’t imagine anyone passing anything but texture coordinates (aka UVs), and colors from the vertex shader to the fragment shader. But there’s not anything actually special about the TEXCOORD# semantics, unlike the SV_POSITION one, so it can be used to pass any kind of data you want.

And the v.vertex has the local vertex position data in it.

3

Thank bgolus, but i applied this shader to Quad. therefore, it has 4 vertices (0, 0, 0), (0, 1, 0), (1, 1, 0), (1, 0, 0), so why o.position has a lot of value than 4?

4

Due to how GPUs use matrix transforms, the vertex position is a float4 with a w of 1.0. When you’re using a 3D scale, rotation, and translation transform matrix, you need a 4x4 matrix, and you need to use a float4 with a w of 1.0 to have it apply the translation otherwise a w of 0.0 only applies the scale and rotation. For legacy reasons of GPUs not using “shaders” like we think of them today, that means the vertex data needs to have the position be a float4 with the w of 1.0 already set. Today there’s no real reason for it as the shader can override the w value to be whatever is needed.

There’s no need for it to be a float4 when being passed to the fragment shader since only the x and y values are used.

Also, the default quad’s vertex positions are (-0.5, -0.5, 0.0, 1.0), (-0.5, 0.5, 0.0, 1.0), (0.5, 0.5, 0.0, 1.0), (0.5, -0.5, 0.0, 1.0), which is why the fragment shader draws a circle and not a quarter slice of a circle.

5

I’m sorry but I don’t understand. if o.position = v.vertex so o.position just has 4 values (-0.5, -0.5, 0.0, 1.0), (-0.5, 0.5, 0.0, 1.0), (0.5, 0.5, 0.0, 1.0), (0.5, -0.5, 0.0, 1.0)

in this line: float inCircle = step(0.25, length(i.position.xy)); if it has 4 positions, the shader will draw 4 vertices, but this draws a circle like i.position.xy has values are the pixels like UV.

6

Unity’s default quad mesh has 4 vertices. The v.vertex values for those 4 vertices are as I stated above, and the v.uv values are (0.0, 0.0), (0.0, 1.0), (1.0, 1.0), (1.0, 0.0). Those are the values the vertex shader sees as inputs, and passes onto the fragment shader.

The fragment shader never sees those exact values. It instead sees the interpolated values depending on what position within each triangle it appears. Using the barycentric coordinate (proportionally how close that position on the triangle it is to each of the 3 vertices that make that triangle) the fragment shader gets a blend of 3 of the values output by the vertex shader. It doesn’t matter if the values come from UVs, or the vertex position, or the vertex color, or are calculated by the vertex shader some other way. They’re all treated the same way … they get interpolated and passed onto the fragment shader.

The only thing this isn’t true for is the SV_Position, which is the clip space (aka projective screen space) position of those 4 vertices. Because GPUs don’t draw quads, they draw triangles, the quad mesh is two triangles. So normally you’d be drawing a square on screen, but instead you’re getting the local position and clipping the area outside of a radius from the center of the quad.

7

Thank you very much, I understood it.