[SOLVED] Unity 5.6.0b4 Use of UNITY_MATRIX_XXX vs UnityObjectToClipPos

Graphics Shaders & Materials
1

Hi everyone,

in previous version of unity I used the following like to get the object position with a offset.

 float3 pivot =
-mul(UNITY_MATRIX_V, mul(unity_ObjectToWorld, float4(v.vertex.xyz - float3(0, _Scale.y * 0.5, 0), 1)));

Now in in 5.6.bf4 using :

float3 pivot = UnityObjectToClipPos(v.vertex - float3(0, _Scale.y * 0.5, 0));
// Worked perfectly in 5.4.0f1

Cause completely wrong results.

Does any one know to use UnityObjectToClipPos() in the correct context to get pivot correctly in meters?
Thanks a lot.

2

https://forum.unity3d.com/threads/unityobjecttoclippos.400520/
?

3

This is a about an year old stuff :slight_smile:
The result is wrong, not the shader call.

4

Through the new inline definitions in the UnityCG.cgnic you must take care of the swizzle commands.

float3 pivot = -UnityObjectToClipPos(v.vertex.xyz - float3(0, _Scale.y * 0.5, 0));

1 Like
5

In the newest UnityCG it takes care of the swizzle by having a float4 input version that calls the float3 using input.xyz.
I think your problem was actually that you were trying to subtract a float3 from a float4(vertex), and thus had to define a float3 result to subtract from with .xyz.

UnityCG:

// Tranforms position from object to homogenous space
inline float4 UnityObjectToClipPos( in float3 pos )
{
    // More efficient than computing M*VP matrix product
    return mul(UNITY_MATRIX_VP, mul(unity_ObjectToWorld, float4(pos, 1.0)));
}
inline float4 UnityObjectToClipPos(float4 pos) // overload for float4; avoids "implicit truncation" warning for existing shaders
{
    return UnityObjectToClipPos(pos.xyz);
}
6

The strange thing is

float3 pivot = UnityObjectToClipPos(v.vertex - float3(0, _Scale.y * 0.5, 0));

Worked perfect in previous build of 5.4.xx

Now I have to invert the result?

float3 pivot = -UnityObjectToClipPos(v.vertex.xyz - float3(0, _Scale.y * 0.5, 0));

Perhaps it has anything to do with new HSL compiler, new unified the transforming, or inverted zbuffer.

I have some examples where UnityObjectToClipPos does not work generally under dx11 and does not write into the depth.

7

Stange, it’s been the opposite for me. It fixed the MVP method not correctly writing to the depth buffer when used as in a forward rendered shader under deferred player mode. Haven’t seen any other issues now. If you can reproduce that it would make a good bug report.