Pingpong function

Hello,

I am creating a vertex shader to expand the model, and contract it, expand, and contract etc etc. I currently have the expanding, and the contracting I can do. Its the expanding AND THEN contracting, AND THEN expanding. I can only seem to do one at a time.

Does CG have an equivalent to the Mathf.PingPong?

Thanks

 

As far as I know, there isn't really a function that does this for you. You can probably make this yourself by swapping values after a specific condition is met, but I think you will be better of by just passing the pingpong value from script to the shader.

 

how about using a sin or cos value?

 

If you want a triangle wave, you can use abs(frac(_Time.y*speed)-0.5)*magnitude
There is also _SinTime and _CosTime you can read about it here: https://docs.unity3d.com/Documentation/Components/SL-BuiltinValues.html

 

_SinTime! I knew I had forgotten something obvious. Thanks heaps, and thanks for the link to the values page. I was googling 'Shader math / Shader math functions' and couldnt find anything. Thanks again.

 

It might not really matter in your case, but would passing a value that you can use for pingpong purposes be cheaper than using a sin/cos function as those are generally expensive in use?

 

_SinTime isn't a function

 

But it still uses the sin function I guess? Except that it's likely to be set from within the engine.

 

If you look in UnityShaderVariables.cginc, you see it says:

Code (csharp):

1. CBUFFER_START(UnityPerCamera)
2.     // Time values from Unity
3.     uniform float4 _Time;
4.     uniform float4 _SinTime;
5.     uniform float4 _CosTime;
6.     uniform float4 unity_DeltaTime; // dt, 1/dt, smoothdt, 1/smoothdt

So I'm pretty sure they're set from the CPU, instead of being calculated on the GPU, and I assume they take about as much time as other shader properties. I should measure that of course...

 

I checked the cg standard library and could not find any function named Pingpong or something like that, but I think we can write one, just like this :

// return value falls in [0, 1] section //
float Pingpong()
{
int remainder = fmod(floor(_Time.y), 2);
return remainder==1?1-frac(_Time.y):frac(_Time.y);
}

if you want to expand the return value's range just multiply by another value, I tested and it worked well. Hope this could do some help.