Fast Fourier Transform ocean waves simulation for Unity.

Hi. Really nice work.

However I notice rendering artefacts on top/right edge of each Ring. It is present both in the Editor and in the gameview. Looks like it is off by one?

Really nice work. I have been working on FFT ocean waves in UE4 for a couple of weeks. I really appreciate you since I referred to your repository heavily. But now I am struggling with normals when putting cascaded waves together. I don't understand why it works. How do you compute normal from derivatives?

float4 derivatives = tex2D(_Derivatives_c0, IN.worldUV / LengthScale0);
#if defined(MID) || defined(CLOSE)
derivatives += tex2D(_Derivatives_c1, IN.worldUV / LengthScale1) * IN.lodScales.y;
#endif

#if defined(CLOSE)
derivatives += tex2D(_Derivatives_c2, IN.worldUV / LengthScale2) * IN.lodScales.z;
#endif

float2 slope = float2(derivatives.x / (1 + derivatives.z),
    derivatives.y / (1 + derivatives.w));
float3 worldNormal = normalize(float3(-slope.x, 1, -slope.y));

o.Normal = WorldToTangentNormalVector(IN, worldNormal);

Hello

Could you please refer me to sources on your ocean spectrum equations? I am not familiar with half of whats going on for your InitSpectrum shaders.

I am very familiar with Tessendorf's initial spectrum math but you don't seem to be using that one. You seem to be using multiple functions i am unfamiliar with and there is a lot of magic numbers with no explanations for them also.

Hi! I was wondering if you planned to convert the "Ocean" shader to URP 🙏 Or you can help me to do so?

Best,

Hi

Me again, I've slowly been learning the math for this as I want to understand it before I use it.

I read tessendorf's paper but wanted to ask something about the complex number generation for the Fourier amplitudes.

For h0 it is defined as : 1/sqrt(2) * (a + ib) * sqrt(Phillips(kVector));

And h0* (the conjugate) is: 1/sqrt(2) * (a - ib) * sqrt(Phillips(kVector));

Does the conjugate h0* use the same a and b values as the h0 term (but b is negative) or do you generate a second complex number entirely?

Hey, me again.

I've so far got most of the FFT understood now but I am confused by some parts of your code.

You have permute shader which I don't know what that is or where it comes from? I see no mention of it in Tessendorf.

Also your scale shader pass is also confusing you have:

void Scale(uint3 id : SV_DispatchThreadID)
{
	Buffer0[id.xy] = Buffer0[id.xy] / Size / Size;
}

This equates to (xy*Scale) / Scale which basically simplifies to xy so you're not scaling anything since scale/scale =1..?

Shouldn't it be:

[numthreads(8, 8, 1)]
void Scale(uint3 id : SV_DispatchThreadID)
{
	Buffer0[id.xy] = Buffer0[id.xy] / (Size * Size);
}

I am basing it on this iDFT equation:

Also i am very confused by the cascade stuff, does each cascade require its own FFT for DY DX DZ ? How did you chose the numbers:

    [SerializeField]
    float lengthScale0 = 250;
    [SerializeField]
    float lengthScale1 = 17;
    [SerializeField]
    float lengthScale2 = 5;

Are these numbers arbitrary or is there a calculation you made to determine the ideal cascade scales?

Thanks!