Noise Filters

Value Noise

Value noise adds noise to the luminance of an image.  It may be useful when you want to add noise to a grayscale image, or when you want to reduce banding without adversely effecting the color of an image.

Value noise can also be the basis for many textures, by applying additional filters.  A nice brushed metal texture can be achieved by applying a motion blur filter to value noise.  Even fractal noise is based on multiple iterations of value noise.


Color Noise

Color noise adds random values to the red, green, and blue channels of the buffer.  Small amounts of color noise can actually improve the look on many images.  Banding can occur on images with smooth color blends and gradients, and roughing them up a little bit with noise can actually make them look smoother an more natural instead of computer generated.


Jitter Noise

Jitter noise roughens an image by displacing pixels by random values.


Jitter blur +

Jitter blur plus is similar to the Jitter filter, but gives some additional options, such as multi-sampling, and working in different color models.

The Samples parameter controls how many samples will be taken from the surrounding pixels.

The Scatter modes sample from a pixels color. In the case of RGB, HSV, and YUV modes, the samples are taken randomly from each channel of the color model independantly.

Film Grain

This filter is used to simulate the look of film grain.  It is often useful to add a little fake film grain to a computer generated image to give it that added realism.

You can adjust the color of the grain, as well as the scale.


Noise removal

This filter can remove low-frequency noise (noise in low detail areas) from an image, using an iterative approach. A simple box filter is applied the specified number of times, however, areas of high-frequency noise (high detail areas) are preserved. This is based on the Threshold. The threshold parameter is a literal count of pixel differences in a small box around a given pixel. If the count is small, that means the neighboring pixels have little change and this is a low frequency area. If the count is high, that means this is a high frequency area, ie, an area with high detail.

Since our eyes are more sensitive to grain in areas of low detail, it is helpful to remove grain there, and leave it in the higher detail areas.

The Mode drop-down control should be kept on the “Threshold box filter” mode for all practical purposes.

Note, digital cameras will often perform their own noise reduction.