• Advertisement
Sign in to follow this  

Using partial derivatives to light circular waves

This topic is 2084 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I'm doing an exercise in a book that asks to add ambient, diffuse and specular lighting to an existing demo that animates circular waves over a basic flat grid. My attempt as it stands doesn't seem to work as expected though. Instead of looking circular in tune with the waves themselves, my lighting has a strange moire like pattern to it. I've forgotten most of my calculus, but fortunately the book's exercise spoon feeds me the formula for computing the partial derivatives neccessary for calcuating the vertex normals. The way I understand it, once I compute the partial derivatives, I simply plug them into the tangent vectors u and v and take the cross product to get my vertex normal.

Here is my shader...

// CHAPTER 10 10.12 EXERCISE 6
// LIGHTING SHADER FOR WAVES

uniform extern float4x4 gWorld;
uniform extern float4x4 gWorldInverseTranspose;
uniform extern float4x4 gWVP;
uniform extern float gTime;
uniform extern float4 gAmbientMtrl;
uniform extern float4 gAmbientLight;
uniform extern float4 gDiffuseMtrl;
uniform extern float4 gDiffuseLight;
uniform extern float4 gSpecularMtrl;
uniform extern float4 gSpecularLight;
uniform extern float gSpecularPower;
uniform extern float3 gLightVecW;
uniform extern float3 gEyePosW;
struct OutputVS
{
float4 posH : POSITION0;
float3 normalW : TEXCOORD0;
float3 posW : TEXCOORD1;
};
// Amplitudes
static float a[3] = {0.8f, 0.2f, 0.1};

// Angular wave numbers.
static float k[3] = {1.0, 8.0f, 10.0f};

// Angular frequency.
static float w[3] = {1.0f, 8.0f, 0.6f};

// Phase shifts.
static float p[3] = {0.0f, 1.0f, 2.0f};
float SumOfRadialSineWaves(float x, float z)
{
// Distance of vertex from source of waves (which we set
// as the origin of the local space).
float d = sqrt(x*x + z*z);

// Sum the waves.
float sum = 0.0f;
for(int i = 0; i < 2; ++i)
sum += a*sin(k*d - gTime*w + p);

return sum;
}
void Partials(float x, float z, out float dhOverdx, out float dhOverdz)
{
float d = sqrt(x*x + z*z);
dhOverdx = 0.0f;
dhOverdz = 0.0f;
for(int i = 0; i < 3; ++i)
{
dhOverdx += (a*k*x*cos(k*d - gTime*w + p))/d;
dhOverdz += (a*k*z*cos(k*d - gTime*w + p))/d;
}
}
OutputVS ColorVS(float3 posL : POSITION0, float3 normalL : NORMAL0)
{
// Zero out our output.
OutputVS outVS = (OutputVS)0;

// Get the height of the vertex--the height is given by
// summing sine waves.
posL.y = SumOfRadialSineWaves(posL.x, posL.z);

// calc normal
float derivx;
float derivz;
Partials(posL.x, posL.z, derivx, derivz);

float3 v = float3 (1, derivx, 0);
float3 u = float3 (0, derivz, 1);
normalL = cross(u, v);
// Transform normal to world space.
outVS.normalW = mul(float4(normalL, 0.0f), gWorldInverseTranspose).xyz;
outVS.normalW = normalize(outVS.normalW);
// Transform vertex position to world space.
outVS.posW = mul(float4(posL, 1.0f), gWorld).xyz;
// Transform to homogeneous clip space.
outVS.posH = mul(float4(posL, 1.0f), gWVP);

// Done--return the output.
return outVS;
}
float4 ColorPS(float3 normalW : TEXCOORD0, float3 posW : TEXCOORD1) : COLOR
{
normalW = normalize(normalW);
// Compute the vector from the vertex to the eye position.
float3 toEye = normalize(gEyePosW - posW);

// Compute the reflection vector.
float3 r = reflect(-gLightVecW, normalW);
// Determine how much (if any) specular light makes it into the eye.
float t = pow(max(dot(r, toEye), 0.0f), gSpecularPower);
// Determine the diffuse light intensity that strikes the vertex.
float s = max(dot(gLightVecW, normalW), 0.0f);

// Compute the ambient, diffuse and specular terms separatly.
float3 spec = t*(gSpecularMtrl*gSpecularLight).rgb;
float3 diffuse = s*(gDiffuseMtrl*gDiffuseLight).rgb;
float3 ambient = gAmbientMtrl*gAmbientLight;
return float4(ambient + diffuse + spec, gDiffuseMtrl.a);
}
technique LightedWaveTech
{
pass P0
{
// Specify the vertex and pixel shader associated with this pass.
vertexShader = compile vs_2_0 ColorVS();
pixelShader = compile ps_2_0 ColorPS();
}
}


What am I doing wrong?

Share this post


Link to post
Share on other sites
Advertisement
Please note, this is not homework. I'm simply studying this on my own for fun. So, can anyone help me?

Share this post


Link to post
Share on other sites
If you can post the entire code in a ZIP file and let me know which SDK I need to compile it, I'd be happy to try.

Share this post


Link to post
Share on other sites
Hi, your normal calculation seem correct to me, maybe you are doing something wrong in the pixel shader when calculating the lighting? Can you post some code or executable we can try? What you exactly mean by "a strange moire like pattern". Can you show an image of the result?

Share this post


Link to post
Share on other sites
Here's my project and a screenshot attached. I'm using June 2010 DirectX SDK 9.29.1962

[attachment=9294:mywaves.PNG]
[attachment=9295:ColoredWavesDemo.zip]

Share this post


Link to post
Share on other sites
I see that it's almost right, but it's like one or two axes are being mixed up in the calculation somewhere.

I'm downloading the SDK right now and will play around tonight, if no one else fixes it first.

Share this post


Link to post
Share on other sites
I can't seem to download your ZIP file. The download dies at about 11MB or so.

Is there any chance that you exclude any PCB files in the ZIP file? Apologies. Not sure why the download is failing.

Share this post


Link to post
Share on other sites
Confirmed, download loops at 9.8MB for me, seems to be broken. Did you upload it fully before posting? Don't know why this happens...

Share this post


Link to post
Share on other sites
I couldn't find any files with extension pcb, though I removed a file that was taking way too much space. I'm re-uploading it here.

[attachment=9300:ColoredWavesDemo.zip] Edited by Hseptic

Share this post


Link to post
Share on other sites
Whoops! I only have 5.17MB of space left to upload. Therefore, I cannot post the sdf file that I removed. Why am I only allowed a certain amount of upload space? I'm sorry.

Share this post


Link to post
Share on other sites
You could always create a Google Code account, create a new project, then put the ZIP file in your Downloads. Lots of space.

Share this post


Link to post
Share on other sites
I succeeded in downloading your program, but I'm unable to run (I suggest to link with the static libraries or distribute the required dlls..) and compile it since I do not have Visual Studio on this PC. I will try on the other computer later. Can you modify your shader to show the normals as color and post an image? It may help understanding if the error is in the normal generation or somewhere else.

Share this post


Link to post
Share on other sites
How do I link the libraries or distribute the dlls, and get them to you? Also, how to I make the normals visible and colorize them?

Share this post


Link to post
Share on other sites
You can change the project settings to use static libraries instead of dynamic ones (dlls). To show the normals as color you can simply return the normal in the pixel shader, i.e.

return float4(normalW, 1.0);

Share this post


Link to post
Share on other sites
I have successfully tried your program in another computer. The strange effect is clearly given by the normals. The strange pattern is indeed also visible when showing the normals instead of the result of the light calculation. If we limit the calculation to a single sine wave*, the normals and lighting seem correct. The math formulation is also correct. Thus, I think the strange effect is actually correct in general and the pattern is probably motivated by the interaction of the different waves at different amplitudes, angular velocities and phases.

When you calculate the y-component of the position you are using only two waves instead of three as in the derivative.

* I suggest to see the second and third waves in isolation. These waves clearly follows the pattern you are seeing. The number of vertices is probably too low to show the correct result. You can try to use more polygons or implement some form of displacement mapping. Edited by apatriarca

Share this post


Link to post
Share on other sites
An artifact from using higher wave numbers than can be represented by the mesh perhaps?

Share this post


Link to post
Share on other sites
Is the flaw basically isotropic? No.

Share this post


Link to post
Share on other sites
You are right, it isn't isotropic but the vertices aren't arranged in a circular fashion and the light "agree" with the geometry (look at the border of the mesh for example). I think that may motivate the non-isotropic behaviour.

Share this post


Link to post
Share on other sites
I realize this. I'm merely using the basic sample that comes with the D3D SDK to compare. It still looks like crap, but at least it's isotropic crap.

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement