The Camera Map Per Pixel map lets you project a map from the direction of a particular camera. It is meant as an aid to 2D matte painting: You can render a scene, adjust the rendering using an image-editing application, then use this adjusted image as a matte that is projected back onto the 3D geometry.
To get good results, the plate should be at least 2K pixels; 3K to 6K, or higher is recommended.
Adjusting (which usually means increasing) the value of ZFudge can improve the quality of edges of the projection.
Usually it is a good idea to have Remove Back Face Pixels turned on. Another way to adjust the projection edge is to adjust this control's Angle Threshold.
If you have multiple mattes to project, you might need to slice geometry to make each map's target a separate object.
Click to turn on, and then choose a camera in the scene by selecting it in a viewport, or press H to use the Pick Object dialog.
Once you have assigned a camera, its name appears on this button.
This camera should be the one used to render the map used in the Texture and ZBuffer Mask components.
Click to assign the texture to project. You can assign any kind of map, but typically this is a Bitmap that contains an image file that you first rendered from the same camera, and then possibly edited with a different application.
Click to assign a map that contains Z-depth data used to mask the projection from unwanted surfaces. Typically this is an RPF file or an RLA file rendered using the same camera, with the Z channel option turned on.
Use the toggle to turn use of the ZBuffer Mask on or off. By default, it is off, and it is not turned on automatically when you assign the ZBuffer Mask.
Behaves like the mask in the Mask map by letting you view one map through another. Black areas of the mask are transparent, white areas are opaque, and gray areas are partially transparent, based on the percentage of the gray.