Almost every digital image captured from the real world contains grain or visual noise caused by the recording, encoding, scanning, or reproduction processes and by the equipment used to create the image. Examples include the faint static of analog video, compression artifacts from digital cameras, halftone patterns from scanned prints, CCD noise from digital image sensors, and the characteristic speckle pattern of chemical photography, known as film grain.
Noise isn’t necessarily bad; it’s often added to images to create a mood or tie elements together, such as adding film grain to a computer-generated object to integrate it into a photographed scene. However, noise can be unwanted for aesthetic reasons. Archival footage or high-speed photography may appear unpleasantly grainy; digital compression artifacts or halftone patterns may mar an image; or noise may interfere with technical processes such as bluescreen compositing.
Technical reasons also exist for reducing noise. For example, compression algorithms usually achieve smaller file sizes if the input material is less noisy, so noise reduction is a valuable preprocessing step for work such as DVD creation and video streaming.
The Add Grain, Match Grain, and Remove Grain effects allow you to manipulate grain that appears more or less evenly over an entire image. Grain effects can’t correct image problems that affect only a few pixels, such as dust, salt and pepper noise, or analog video dropouts.
The Add Grain effect generates new grain from nothing; it doesn’t take samples from existing grain. Instead, parameters and presets for different types of film can be used to synthesize different types of grain.
The Remove Grain and Match Grain effects use a two-step process to manipulate grain without affecting the edges, sharpness, or highlights of an image. First, the grain is sampled, either automatically or manually; second, the grain is analyzed and portrayed by a mathematical model, which the effect uses to add, remove, or match the grain.