Lately i have been experimenting with 2D shaders for various post process effects but i'm at the stage were i would like to be able to start masking them to be able to distinguish for example the lens flare sources from the glow sources.

So what i need is the 2D shader to access things like a MaterialSpecular() or MaterialGlow() buffer or to be able to sample some specific textures for each fragment it is processing.

The obvious solution would have been to use the RTT technique to store the information i want and then pass it to my 2D shader for further sampling/masking but the things you have access outside the object shaders are the rendered RGB color, the alpha which looks like it is the MaterialGlow() * GlowTextureValue and the newly added Renderer Stages.

But what if you want something more specific like the MaterialSpecular() for the fragment you are sampling in the 2D Shader?

Is there a specific way i should approach this? Or even can the following idea be considered viable?

For Every Frame
1) Render the whole scene normally and store to rttRGBA.
2) Change at runtime the shaders of all the objects by using "D.setGetShaderFunc(GetShader)" and "D.setShader(null)" with a single shader that would expose the wanted values.
3) Re-render the whole scene and store to rttMASK.
4) Change the shader function back to default with "D.setGetShaderFunc(null)" and "D.setShader(null)".
5) Send rttRGBA and rttMASK to 2D Shader.

Would this be viable performance-wise or is there a better way to approach this? Or maybe the access to the source code is necessary for a viable solution?

Hello,

The specular intensity of the material in deferred renderer is stored in w channel of the normal render target.
To access that you would need to grab the source code, which I highly recommend

Awesome, thanks Esenthel.