#define SAMPLECUTDIST2 0.01
#define WEIGHTCUTOFF 0.01

uniform int nRadionColumns;
uniform sampler2D radionSampler       : register(s0);
uniform sampler2D entryPointCountSampler : register(s1);
uniform sampler2D allWeightsSampler   : register(s2);
uniform float3 lightPos;
uniform float3 lightDir;
uniform float clusterCount;

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.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 = max(dot(lightDir, -diff), 0);
	float cosb = max(dot(dir, diff), 0);	
	
	/*
	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 visibility = 1;
	
	float4 ret = cosa * cosb;

	//return float4(lightDir + ret.x * 0.00000000001, 1);
	return ret;
	//return ret.x*0.0000000001 + 0.5;
}

float4 psSumWeights(vsOutputComputeWeights input) : COLOR0
{
	float halfPixel = 0.5 / clusterCount;
	float iCluster = input.tex.x + halfPixel;
	int entryPointCount = tex2D(entryPointCountSampler, float2(iCluster, 0.25));
	int currentEntryPoint = tex2D(entryPointCountSampler, float2(iCluster, 0.75));
	//sum entrypoint weights
	float weight = 0;
	float clusterRad = 0;
	for(int i = 0; i < entryPointCount; i++)
	{
		float hp = float2(0.5 / (float) nRadionColumns, 0.5 / 4096.0);
		float2 coord1 = float2((float)(currentEntryPoint % nRadionColumns)  / (float) nRadionColumns,
								currentEntryPoint / (float) nRadionColumns / 4096.0) + hp;
		coord1.y = 1.0 - coord1.y;
		float2 coord2 = coord1 + float2(hp.x / 2.0, 0);
		float radrad = tex2Dlod(radionSampler, float4(coord2, 0, 0)).a;
		//weight += tex2Dlod(allWeightsSampler, float4(coord1, 0, 0)).r * radrad;
		weight += tex2Dlod(allWeightsSampler, float4(coord1, 0, 0)).r;
		//clusterRad += radrad;
		currentEntryPoint++;
	}	
	/*
	if(clusterRad >= 0)
		weight /= clusterRad;
	else
		weight = 0;*/
	
    weight /= (float) entryPointCount;
		
	return weight;
}