Strange striations in my bloom effect?

Graphics and GPU Programming Programming

Started by JackOfAllTrades August 13, 2011 11:28 PM

8 comments, last by JackOfAllTrades 12 years, 8 months ago

109

Author

August 13, 2011 11:28 PM

[attachment=4912:Bloom.jpg]

The separate bloom texture does not appear to have this problem but it only happens when I mix the two textures using this shader:



	struct FS_INPUT

	{

		float4 pos				: POSITION;

		float2 tex				: TEXCOORD0;

	};



	struct FS_OUTPUT

	{

		float4 color				: COLOR;

	};



	uniform sampler2D txDiffuse1;

	uniform sampler2D txDiffuse2;



	FS_OUTPUT FS( FS_INPUT In)

	{

		FS_OUTPUT Out;

		Out.color = tex2D(txDiffuse1, In.tex);

		Out.color += tex2D(txDiffuse2, In.tex);

		Out.color.w = 1.0;

		return Out;

	}

Any ideas why this may be happening?

Want an easy to use open source 3D video game SDK that handles Physics, Graphics, Audio, and AI right out of the box?

EsorU

115

August 14, 2011 02:05 PM

What issue did you mean? The less bloom or the over-brighten textures.

if issue is over-bright and txDiffuse2 is the bloom sampler try this:



Out.color = tex2D(txDiffuse1, In.tex);

Out.color.rgb += tex2D(txDiffuse2, In.tex).rgb * ( 1.0f - Out.color).rgb;

Hodgman

52,717

August 14, 2011 02:24 PM

Can you explain what you mean by the strange striations and/or post a close-up picture of a problem area without JPEG compression artefacts?

. 22 Racing Series .

JackOfAllTrades

109

Author

August 14, 2011 11:28 PM

Can you explain what you mean by the strange striations and/or post a close-up picture of a problem area without JPEG compression artefacts?

[attachment=4931:Untitled.png]

Ignore the blockiness that is just due to too few samples. Pay attention to the horizontal lines. What you saw before was not JPEG compression artifacts but the actual problem I speak of. That is how it looks in my game...

Want an easy to use open source 3D video game SDK that handles Physics, Graphics, Audio, and AI right out of the box?

JackOfAllTrades

109

Author

August 15, 2011 01:53 AM

Can you explain what you mean by the strange striations and/or post a close-up picture of a problem area without JPEG compression artefacts?

Actually after boosting the brightness and doubling the samples it appears to be my blur shader that is causing the problem.



	struct FS_INPUT

	{

	  float4 pos				: POSITION;

	  float2 tex				: TEXCOORD0;

	};



	struct FS_OUTPUT

	{

	  float4 color				: COLOR;

	};



	float PixelKernel[27] =

	{

		-13,

		-12,

		-11,

		-10,

		-9,

		-8,

		-7,

		-6,

		-5,

		-4,

		-3,

		-2,

		-1,

		0,

		1,

		2,

		3,

		4,

		5,

		6,

		7,

		8,

		9,

		10,

		11,

		12,

		13,

	};



	static const float BlurWeights[27] =

	{

		0.0005540,

		0.0011080,

		0.0027450,

		0.0043820,

		0.0089395,

		0.0134975,

		0.0229385,

		0.0323795,

		0.0464360,

		0.0604925,

		0.0742545,

		0.0880165,

		0.0938760,

		0.0997355,

		0.0938760,

		0.0880165,

		0.0742545,

		0.0604925,

		0.0464360,

		0.0323795,

		0.0229385,

		0.0134975,

		0.0089395,

		0.0043820,

		0.0027450,

		0.0011080,

		0.0005540,

	};



	float2 direction;



	uniform sampler2D txDiffuse;



	FS_OUTPUT FS( FS_INPUT In)

	{

		FS_OUTPUT Out;

		float2 samp = In.tex;

		for (int i = 0; i < 27; i++)

		{

			samp = In.tex + (direction * PixelKernel);

			Out.color += tex2D(txDiffuse, samp) * BlurWeights;

		}

	return Out;

	}

Any idea what I can do to fix it?

Want an easy to use open source 3D video game SDK that handles Physics, Graphics, Audio, and AI right out of the box?

Hodgman

52,717

August 15, 2011 02:23 AM

It could be that the direction variable is not set correctly. If this is larger than (1.5/width,1.5/height) then you'll skip pixels during the blurring step.

. 22 Racing Series .

JackOfAllTrades

109

Author

August 15, 2011 02:30 AM

It could be that the direction variable is not set correctly. If this is larger than (1.5/width,1.5/height) then you'll skip pixels during the blurring step.

Actually I've found that doing more passes and less samples fixes the problem. Here is what I did:



		for( float i = 0; i < (OvglPi*2); i = i + (OvglPi/4))

		{

			float x = sin(i) * 0.005f;

			float y = cos(i) * 0.005f;

			cgGLSetParameter2f( CgDirection2, x, y );

			glBegin(GL_QUADS);

				glTexCoord2f( 0.0f, 0.0f );

				glVertex3f(-1.0f,-1.0f, -1.0f);

				glTexCoord2f( 1.0f, 0.0f );

				glVertex3f(1.0f,-1.0f, -1.0f);

				glTexCoord2f( 1.0f, 1.0f );

				glVertex3f(1.0f, 1.0f, -1.0f);

				glTexCoord2f( 0.0f, 1.0f );

				glVertex3f(-1.0f, 1.0f, -1.0f);

			glEnd();

		}

Works great though it's a bit slower than I was hoping for.

Want an easy to use open source 3D video game SDK that handles Physics, Graphics, Audio, and AI right out of the box?

Hodgman

52,717

August 15, 2011 02:49 AM

Is the value of the direction variable in any way related to the resolution of the input texture?

. 22 Racing Series .

JackOfAllTrades

109

Author

August 15, 2011 02:51 AM

Is the value of the direction variable in any way related to the resolution of the input texture?

Nope... why?

Want an easy to use open source 3D video game SDK that handles Physics, Graphics, Audio, and AI right out of the box?

JackOfAllTrades

109

Author

August 15, 2011 07:51 AM

Fixed! The problem was kinda obvious. I was writing into the same texture I was reading from. I'm now flip-flopping between two different textures for each pass and it works perfect!

Want an easy to use open source 3D video game SDK that handles Physics, Graphics, Audio, and AI right out of the box?

Strange striations in my bloom effect?

This topic is closed to new replies.

Popular Topics

Recommended Tutorials

Strange striations in my bloom effect?

This topic is closed to new replies.

Popular Topics

Recommended Tutorials

Reticulating splines