stage

Type:

Remarks:

• If the following stage has no PROCESS_SCREEN flag set, the view renders into a target bitmap rather then on the screen. If the view has no bmap parameter, a render target bitmap is automatically generated with the current size of the view. After rendering, the global render_target pointer is set to the target bitmap, making it available to the following stage through the TargetMap texture parameter.
• If a view renders into a render target, the content from the screen or from a previous stage is not automatically copied into the target bitmap before rendering. This must be kept in mind when the view post processes an existing scene, for instance by blurring the stencil buffer content on the screen. In such a case the resulting image must be composed from the content of the previous stage ( TargetMap) blended with the new content (e.g. StencilMap).
• The following stage does not need the VISIBLE flag for rendering. For disabling a postprocessing stage, it is sufficient to set stage at NULL.
• The following stage either renders the whole scene again (PROCESS_TARGET not set), or only the TargetMap (PROCESS_TARGET set) for 2D postprocessing effects. You can use pixel shaders for 2D postprocessing, but no vertex shaders.
• An arbitrary number of stages can be linked together for a view rendering chain. For instance, shadow mapping requires a view rendering from a light source into a depth map, followed by a view that renders the shadow map using the depth map as a source map, followed by one or two post processing stages for blurring, and finally by a view that renders the scene combined with the blurred shadow map.
• Depending on the complexity of the stage shader and the size of the view, postprocessing can remarkably reduce the frame rate. However in some cases postprocessing can even be used for increasing the frame rate, especially when complex effects and shaders are used. For this, the first view renders only into the z buffer and the stage view then renders the full scene with shaders, this way reducing the number of shaded pixels through early z-buffer clipping.

Edition:

Example (lite-C):

MATERIAL* mtlEmboss = // a postprocessing material
{
  effect = "
  Texture TargetMap;
  sampler2D smpSource = sampler_state { texture = <TargetMap>; };
  float4 vecViewPort; // contains viewport pixel size in zw components

  float4 embossPS( float2 Tex : TEXCOORD0 ) : COLOR0 
  {
    float4 Color = float4( 1.0, 0.5, 0.5, 0.5 );
	  Color -= tex2D( smpSource, Tex.xy-vecViewPort.zw)*2.0;
	  Color += tex2D( smpSource, Tex.xy+vecViewPort.zw)*2.0;
	  Color.rgb = (Color.r+Color.g+Color.b)*0.333; 
    return Color;
  }

  technique emboss { pass one { PixelShader = compile ps_2_0 embossPS(); } }
  technique fallback { pass one { } }
  ";
}

VIEW* viewEmboss = { material = mtlEmboss; flags = CHILD | PROCESS_TARGET; } // a postprocessing view
...
camera.stage = viewEmboss; // enable postprocessing for the camera view

See also:

► latest version online