In it’s most basic form a matte is an image that tells a second image where to be opaque and where to be transparent. Sometimes this opacity information comes in the form of an “alpha channel”. In this special instance, the transparency information is embedded inside the matte image in question. So, a RGB image, with an embedded alpha channel, actually has 4-channels, not just three: RGBA (Red, Green, Blue and Alpha). The alpha channel embedded in the one file can be used to dictate the transparency in another file. So let’s talk about mattes and alpha channels. Specifically, luma mattes and alpha mattes—what’s the difference? And why would you need to invert a matte?
When you assign a luma matte to your subject image, you are telling the subject to use the lightness values of the matte image as it’s source of transparency. Typically, the areas of the matte that are perfectly white will cause the subject to be completely opaque: 100% visible. Conversely, areas of the matte that are completely black force the opposite: those areas of the subject will disappear completely. Then, if areas of the matte are gray, there will be partial transparency in these areas of the subject.