Sharpen Node

The Sharpen node enables you to enhance the detail in an image, effectively sharpening it.

You can choose from spatial analysis or frequency analysis. The default is spatial analysis, because it is the simpler method. Spatial analysis uses a convolution operation for sharpening.

Note:
If you need more information about using a convolution filter for sharpening, refer to the Convolve Node description.

Frequency analysis, on the other hand, performs several filtering steps and requires that the parameters on the Advanced tab be set appropriately for the image being sharpened.

Using Frequency Analysis

The results of Frequency Analysis are well worth the effort of learning the controls involved. Start by reading the descriptions of the parameters on the Advanced tab. Then decide if the image needs prefiltering before sharpening; that is, if the image is especially noisy or grainy.

For the actual sharpen operations, start by setting the Noise Level and Scale parameters. (The Scale parameters are only applicable to film.) These parameters set the threshold for the high frequency signal being analyzed: anything above the threshold is considered content (image detail) and anything below is considered noise and removed.

The second part of the sharpen operation is to control how much of the "cleaned up" high frequency signal gets added back into the original image. Open the Sharpen node monitor and adjust the Amount and Exponent parameters until you see the result you want. Higher values mean more sharpening for both of these parameters. The difference is that the Exponent is used to concentrate the sharpening effect in the higher frequencies (i.e., edges) or spread it out over the entire image.

Sharpen Parameter Tab

Operation

The Operation menu enables you to select the method to use for the sharpen operation:

Spatial Analysis

This is a 3 x 3 convolve kernel that sharpens by assigning greater weight to the value of the central pixel in the convolution than to the surrounding pixels in the kernel.

Frequency Analysis

This method treats the image as a continuous signal of changing intensity. To sharpen the image, the high-frequency part of the signal--which tends to include important image detail, like edges or small features--is separated out, smoothed if possible to remove extra noise, and then added back into the original image.

There are two frequency analysis methods, one optimized for video imagery and one for film. When either is selected, the parameters on the Advanced tab are enabled. These parameters must be set knowledgeably to get the best results, so it is recommended that you review the Advanced tab parameter descriptions below.

Amount

This parameter enables you to specify the amount of sharpening. With either method, lower values produce less sharpening, and higher values, more (from no sharpening at 0 to a very harsh result at 1):

  • For spatial analysis, the value you set is used in the convolution kernel.
  • For frequency analysis, this value specifies the amount of high-frequency detail to be added back into the original image.

Channel Mask

The Channel Mask parameter enables you to inhibit the modification of any channel by deselecting the Red, Green, Blue, Alpha, or Other channel icons.

Control Image Parameters

The Control Action and Control Channel menus enable you to specify how an optional control image input will govern the node operation, as explained in "Using Control Images with Filter Nodes ."

Advanced Parameter Tab

The parameters on the Advanced tab become available when a frequency analysis option is selected from the Operation menu on the Sharpen tab.

Note:
Enabling the prefiltering parameters is optional and they may be used singly or together. In either case, any prefiltering is done before the sharpen operation.

Noise Level and Scale Parameters

These parameters enable you to set a threshold for noise and grain filtering of the high frequency signal before it is added back into the original image. The Noise Level parameter applies to all channels equally, whereas the Scale parameters are used to weight the influence of the noise level value differently for each channel:

Noise Level

Large Noise Level values will remove more of the original signal--they will remove more noise, but also more picture detail. Smaller values will preserve detail, but allow more noise through.

Scale (Red, Green, Blue, Other)

For film grain analysis, the Scale parameters enable you to adjust threshold values for each channel individually because each film layer responds differently to light. In particular, the blue layer typically has much more noise than the green layer.

The default values are a good starting point, as they reflect a ratio typical for many film stocks. (For video, however, the Scale controls are not active, because video signals do not exhibit noise differences between the red, green, and blue components.)

Exponent

The Exponent parameter, in conjunction with the Amount parameter on the Sharpen tab, modifies how much of the high frequency signal falling above the threshold levels (set in the Noise Level and Scale parameters) will be added back into the original image.

The Exponent differs from the Amount parameter in that the exponent is a power function that controls the slope of the nonlinear curve that represents the frequency distribution. This means basically that it controls the steepness of the slope:

  • Higher exponent values concentrate the sharpening effect (as specified in the Amount tab) in areas of greatest frequency change in the image, typically edges.
  • Lower exponent values will tend to spread the effect throughout the image, making it look soft.

Kernel Size

The Kernel Size menu is used to specify the size of the convolution kernel used by a blur operation that is performed initially to help identify the high frequency component of the signal. (The high frequency component is the difference between the original image and the blurred image.)

Prefiltering Parameters

The Prefiltering parameters give you the option to specify that a prefiltering operation be done before the actual sharpen operations:

Prefilter Blue

In film, the blue component is subject to more noise than either red or green. Therefore, you can check the Prefilter Blue option to instruct Chalice to prefilter the blue channel only, before sharpening. Prefilter Blue is a basic 3 x 3 convolution similar to a blur.

Prefilter Grain

The Prefilter Grain option performs grain filtering on all three image channels before sharpening, using the same frequency analysis method performed in the Degrain node.

Grain Size

If Prefilter Grain is checked, the Grain Size parameter becomes active. It enables you to control how much of the new, degrained signal is added back into the original image during the Prefilter Grain operation. Higher values enhance detail up to a point; beyond that, however, the image will begin to look harsh.




Table of Contents | Index



Copyright 1996-1999 by Silicon Grail Corporation