Need help in Casting ray

[plastic]

I have been trying to implement a ray tracer in my fragment shader. And I need to clear off some confusions first:
i) I have scan through some example ray tracer, and I find that most of them tend to use the fragCoord to cast the ray. But I have been wondering why don’t they use eye and vertex coordinate to cast the ray? Please correct me if I am wrong here, the camera/eye is at (0,0,0) in eye coordinate. And suppose we multiply vertex with modelview matrix: myvertex = (ModeviewMAtrix * vertex) , we get a result of vertex in eye coordinate stored in myvertex. Is it appropriate to cast a ray with ray.Origin at (0,0,0) and direction of the ray is the position of myvertex in normalized form?

I am trying to do a very basic stuff only here, drawing 2 sphere with different ambient different material, no lighting calculation involve yet, I just want my shader to be able to correctly map the correct ambient color onto my spheres

size 640 480 // window size
camera 0 -4 4 0 0 0 0 1 1 45 //eye, center, up, fovy

pushTransform // first sphere should look grey color
ambient .7 .7 .7
sphere 0 0 0 1 // xyz radius
popTransform;

pushTransform // 2nd sphere, purple
translate 2 0 0
ambient .1 .7 .7
sphere 0 0 0 1
popTransform;

//vertex shader

void main() {
        gl_Position = pm * mv * vec4(vertices, 1.0) ;   // pm and mv are uniforms for projection and modelview mat
	myvertex = mv * vec4(vertices, 1.0) ;            // vertices is the varying input passed from window with
}

//Fragment shader

//data for raytracer//
const int numObj = 2;
uniform vec4 ambData[numObj];
uniform vec4 diffData[numObj];
uniform vec4 specData[numObj];
uniform vec4 emiData[numObj];
uniform mat4 transfData[numObj];
uniform float shnData[numObj];
uniform int typeData[numObj];
uniform float sizeData[numObj];

uniform mat4 lookAt;
uniform int maxDepth;

//////////////////////


bool circleIntersect(in vec3 cen, in float r, in vec3 ori, in vec3 dir, inout float t ){
	
	vec3  RC = ori - cen;
        float DD = dot(dir, dir);
        float DdRC = dot(dir, RC);
	float sqtN, sqtP;

	t = r*r - dot(RC,RC) + DdRC*DdRC;

        if( t > 0.0 )	// 2 root
        {
        sqtP = sqrt(t) - DdRC;
		sqtN = -sqtP - DdRC;
		if (sqtP <= 0 && sqtN <= 0){ 
			return false;
		}

		if (sqtN < sqtP){
			t = sqtN;
		}
		else{
			t = sqtP;
		}

		if (t <= 0)
			return false;
		return true;
    }
    return false;
}

vec4 intersection(in vec3 rayO, in vec3 rayD){
	vec4 retClr = vec4(0.0);
	float tMin = t_inf;                  //t_inf is a constant = 100 000.0f
	float t = tMin; 
	vec3 norm;
	int closestIdx;
	for (int i = 0; i < numObj; i++){
		if (typeData[i] == 2){
			vec4 c = lookAt * transfData[i] * vec4(0.0, 0.0, 0.0, 1.0);
			if (circleIntersect(c.xyz, 6.0, rayO, rayD, t)){
				if (t < tMin){
					closestIdx = i;
					tMin = t;
				}
			}
			else{
				continue;
			}
		}
		else if (typeData[i] == 4){
			if (cubeIntersect(transfData[i], rayO, rayD, t, norm)	){
				if (t < tMin){
					closestIdx = i;
					tMin = t;
				}
			}
			else{
				continue;
			}
		}
	}

	retClr = vec4(0,0,0,1) + ambData[closestIdx];

	return retClr;
}


void main (void) 
{   
	vec4 eye = vec4(0,0,0,1); //lookAt * vec4(0.0, -4.0, 4.0, 1.0);
	
	vec3 rayOri = eye.xyz / eye.w;
	vec3 rayDir = normalize(myvertex.xyz - eye.xyz);
	
	gl_FragColor += intersection(rayOri, rayDir);
}

The result I get is 2 circle are drawn on the screen. But sadly both of them in purple color…I rotate the camera to see these 2 sphere from different directions but both of them look completely purple, no other color. I highly doubt my ray setup is totally wrong.

[plastic]

I discovered where the problem at, i copy the wrong transfData and ambData in opengl, that is why everything is so out of control. Using eye position at (0,0,0) and ray direction toward vertex position in eye coordinate system is not wrong.

Sorry for my clumsiness.