#define SAMPLECUTDIST2 0.01
#define WEIGHTCUTOFF 0.01

int nRadionColumns;

texture radions;
sampler radionSampler = sampler_state
{
   Texture = <radions>;
   MinFilter = Point;
   MagFilter = Point;
   MipFilter = None;
};


struct vsInputComputeWeights
{
    float4	pos			: POSITION;
    float2	tex			: TEXCOORD0;
};

struct vsOutputComputeWeights
{
    float4	pos			: POSITION;
    float2	tex			: TEXCOORD0;
};

vsOutputComputeWeights
	vsComputeWeights(vsInputComputeWeights input)
{
    vsOutputComputeWeights output = (vsOutputComputeWeights)0;
    output.pos = input.pos;
    
    output.tex = (input.pos + 1.0) / 2.0;
    output.tex.y = 1.0 - output.tex.y;
    
    //output.tex = input.tex;
    return output;
}

float4
	psComputeWeights(vsOutputComputeWeights input) : COLOR0
{
	float dataColumnWidth = 1.0 / (float)nRadionColumns;
	int bushIndex = input.tex.x * 4096 + input.tex.y * nRadionColumns * 4096;
	int werx = bushIndex % nRadionColumns;
	int wery = bushIndex / nRadionColumns;
	float3 pos = tex2D(radionSampler, float2((werx + 0.25) * dataColumnWidth, (wery  + 0.5) / 4096.0) ).xyz;
	float3 dir = tex2D(radionSampler, float2((werx + 0.75) * dataColumnWidth, (wery  + 0.5) / 4096.0) ).xyz;
	/*
	float3 diff = lightPos - pos;
	float dist2 = dot(diff, diff);
	if(dist2 < SAMPLECUTDIST2)
		dist2 = SAMPLECUTDIST2;
	diff = normalize(diff);
	float cosa = - dot(lightDir, diff);
	float cosb = dot(dir, diff);	
	
	float4 occProjPos = mul(float4(pos, 1), occWorldToProjMatrix);	
	occProjPos /= occProjPos.w;
	float2 occTexPos = mul( occProjPos.xyw, occProjToTexMatrix);
	float visibility = tex2Dproj(depthMapSampler, float4(occTexPos, occProjPos.z - 0.1, 1) );

	float3 lightToPos = pos - lightPos;
	float actualDepth = length(lightToPos);

	float ret;
	if(  visibility < 0.5 || cosa < 0 || cosb < 0)
		ret = 0.0;
	else
		ret = pow(cosa, 9) * cosb / dist2;

	if(ret < 0.000001)
		ret = 0.0;
//	if(ret > WEIGHTCUTOFF)
//		ret = WEIGHTCUTOFF;
	*/
	
	
	float3 diff = lightPos - pos;
	float dist2 = dot(diff, diff);
	if(dist2 < SAMPLECUTDIST2)
		dist2 = SAMPLECUTDIST2;
	diff = normalize(diff);
	float cosa = max(dot(lightDir, -diff), 0);
	float cosb = max(dot(dir, diff), 0);
	float ret = pow(cosa, 9) * cosb / dist2;
	return ret;
	
	return float4(ret, ret, ret, ret);
//	return float4((wery  + 0.5) / 4096.0, werx, werx, 1);
//	return float4(actualDepth, 1, 1, 1);
}

technique ComputeWeights{
	pass P0
    {
        VertexShader = compile vs_3_0 vsComputeWeights();
        PixelShader  = compile ps_3_0 psComputeWeights();
    }
}
/*
float4
	psAggregateWeights(vsOutputComputeWeights input, in int rid : VPOS) : COLOR0
{
	float sumWeight = 0.0;
	float sumProb = 0.0000001;
	
	int iRadion = clusterStarts[rid];
	float2 wTex = float2((iRadion / 4096 + 0.5) * dataColumnWidth,
						 (iRadion % 4096  + 0.5) / 4096.0) );
	while(iRadion < clusterStarts[rid+1])
	{
		sumWeight += tex2Dlod(weightSampler, float4(wTex, 0, 1));
		float sumProb += 1.0;
		wTex.y += 1.0 / 4096.0;
	}
	return (sumWeight / sumProb).xxxx;
}

technique AggregateWeights{
	pass P0
    {
        VertexShader = compile vs_3_0 vsComputeWeights();
        PixelShader  = compile ps_3_0 psAggregateWeights();
    }
}
*/