I'm trying to get some particles to work properly - they should blend among themselves but not with the background color. For this I'm using additive blending, textures with premultipled alpha and set the color as (r*a, g*a, b*a, 0) on textures I want to add up. I'm using the blend mode: glBlendFunc(GL11.GL_ONE, GL_ONE_MINUS_SRC_ALPHA); for everything.

I found out that what I want to do can be achieved by rendering and additively blending the particles in a framebuffer and then render that framebuffer on top of the background with standard alpha blending. In theory I think this should work..however i'm still getting the same result as I did before which is very bright particles at bright areas of the background.

So what I'm doing is simply: 1)Render the background as usual with color (r*a,g*a,b*a,a)

2)Bind my fbo

3)Use GL11.glClear(GL11.GL_COLOR_BUFFER_BIT); GL11.glClearColor(0f, 0f, 0f, 0.0f);

4)Render all the particles with color as (r*a,g*a,b*a,0)

5)Unbind the fbo and render it on top of the bg with (1,1,1,1) color.

Why doesn't this work?


1 Answer 1


I think you may not entirely understand how destination blending works.

You claim that you set your output to (r*a, g*a, b*a, 0.0) and then use GL_ONE_MINUS_SRC_ALPHA for the destination blend factor. Because source alpha is constant (0.0) in this example, that works out to be equivalent to GL_ONE.

Likewise, if you apply GL_ONE, GL_ONE_MINUS_SRC_ALPHA in your second pass that effectively discards anything already in the color buffer:

SrcColor.rgba * 1.0   +   (1.0 - 1.0) * DstColor.rgba    =  SrcColor.rgba  =  (1,1,1,1)

In this case, you would be better off simply disabling blending in the second pass.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .