Hi there,

I have been trying to implement SSAO and I was reading for quite a while now, so I should be quite familiar with the theory.

How ever the comparatively easy task to create the depth normal map happens to be a problem...

Here is what i am doing:

struct VS_INPUT
{
float3 position : POSITION;
float3 normal    : NORMAL;
};

struct VS_OUTPUT
{
float4 position : SV_Position;
float3 normal : TEXCOORD0;
float depth : TEXCOORD1;
};


VS_OUTPUT VSMAIN (in VS_INPUT input)
{
    VS_OUTPUT output;

    output.position = mul (float4 (input.position, 1.0f), WorldMatrix);
    output.position = mul (output.position, ViewMatrix);
    output.position = mul (output.position, ProjMatrix);

    float4 norm = mul (float4 (input.normal, 0.0), WorldMatrix);
    output.normal = mul (norm, ViewMatrix).xyz;
    output.normal = normalize (output.normal);
    output.normal = output.normal * 0.5 + 0.5;

    output.depth = output.position.w;

    return output;
}

float PSMAIN (in VS_OUTPUT input) : SV_Target
{
    return float4 (input.normal, input.depth);
}

Unfortunately the output generated is not quite the thing you would expect, the picture is flickering with "noise":

The Question now is, what the devil could cause such behaviour? I am confused.

Yeah that was kind of obvious, wasn't that...

Btw. Further in the ssao, how are you converting the depth into view space?

I intended to strip down the sample code from the hieroglyph3 engine, but I'm not entirely familiar with cumpute shader yet (but theres plenty of examples to go through)... I will reply again if theres an acceptaple solution for me.

Thaks for your quick response!

First problem: the return type is declared as a float but you are actually returning a float4. I'm surprised this is even compiling.

It is legal to cast any floatn to float, so no surprise here. However, there may be warning for it.

Cheers!

Sorry to bother you again...

I think i got the depth normal pass working and went on to the actual algorithm,

but theres weired stuff happening again...

This is how i generate normals and positions:

struct VS_INPUT
{
float3 position : POSITION;
float3 normal	: NORMAL;
};

struct PS_INPUT
{
float4 position : SV_Position;
float4 viewpos : POSITION;
float3 normal : NORMAL;
};

PS_INPUT VSMAIN (in VS_INPUT input)
{
	PS_INPUT Out;

	Out.viewpos = mul (float4 (input.position, 1.0f), WorldMatrix);
	Out.viewpos = mul (Out.viewpos, ViewMatrix);

	Out.position = mul (Out.viewpos, ProjMatrix);

	float3 viewSpaceNormal = mul (input.normal, (float3x3)WorldMatrix);
	Out.normal = mul (viewSpaceNormal, (float3x3)ViewMatrix);
	

	return Out;
}

struct PS_OUTPUT
{
	float4 normal : SV_Target0;
	float4 viewpos : SV_Target1;
};

PS_OUTPUT PSMAIN (in PS_INPUT In)
{
	PS_OUTPUT Out;

	Out.normal =  float4(normalize(In.normal) * 0.5f + 0.5f, 1.0f);
	Out.viewpos = In.viewpos;

	return Out;
}

This is what i got so far, it is pretty much like the article from here: http://www.gamedev.net/page/resources/_/technical/graphics-programming-and-theory/a-simple-and-practical-approach-to-ssao-r2753

The Ambien Occlusion shader here is applied to a full screen triangle

float3 getPosition (in float2 uv)
{
	return PositionBuffer.Sample (LinearSampler, uv).xyz;
}

float3 getNormal (in float2 uv)
{
	return DepthNormalBuffer.Sample (LinearSampler, uv).xyz;
}

float2 getRandom (in float2 uv)
{
	return normalize (RandomTexture.Sample (LinearSampler, uv).xy * 2.0f - 1.0f);
}

float doAmbientOcclusion (in float2 tcoord, in float2 occluder, in float3 p, in float3 cnorm)
{
	// vector v from the occludee to the occluder
	float3 diff = getPosition (tcoord + occluder) - p;
	const float3 v = normalize (diff);

	// distance between occluder and occludee
	const float d = length (diff);

	return max (0.0, dot (cnorm,v)) * (1.0 / (1.0 + d));
}


float PSMAIN (in PS_INPUT In) : SV_Target
{
	const float2 vec[4] = {
		float2 (1,0), float2 (-1,0),
		float2 (0,1), float2 (0,-1)
	};

	float3 p = getPosition (In.tex);
	float3 n = getNormal (In.tex);
	float2 rand = getRandom(In.tex);

	float ao = 0.0f;
	float rad = 0.125 / p.z;

	int iterations = 4;
	for (int j = 0; j < iterations; ++j)
	{
		float2 coord1 = reflect(vec[j], rand) * rad;
		float2 coord2 = float2 (coord1.x*0.707 - coord1.y*0.707,
		                        coord1.x*0.707 + coord1.y*0.707);

		ao += doAmbientOcclusion (In.tex, coord1*0.25, p, n);
		ao += doAmbientOcclusion (In.tex, coord2*0.5, p, n);
		ao += doAmbientOcclusion (In.tex, coord1*0.75, p, n);
		ao += doAmbientOcclusion (In.tex, coord2, p, n);
	}
	ao /= (float)iterations*4.0;

	return ao;
}

after that I render the scene again and modulate the output by using this occlusion buffer

how ever theese are the images it produces:

I dont't know how that soapy screen effect is produced but i think it has something to do with the randomTexture.

I can't get the image look right to me (aside from the soap screen) by tweaking the parameters.

For example: If I look 45° to the left the front faces grow dark (why?).

Another thing is, that if I normalize the random normals the frame rate drops to a tenth.