Scale Node
The Scale node enables you to enlarge or reduce the size of an input image. You can specify exact numerical values, in pixels, or use multiplier values to scale the image in the X and Y dimensions. The Scale node accepts one or two inputs. The top input is the image to be scaled. The optional bottom input is used as a size reference for the scale operation.
- Tip:
- If you are scaling an input to match the size of another image in the network, the fastest method will probably be to connect this reference image to the second Scale node input. This will automatically scale the top input image to match.
Scale Parameter Tab
The Scale tab enables you to specify either specific XY dimensions in pixels or the factor by which the input image will be scaled in X and Y. First you choose whether to work in Pixels or Fractions, and then you enter values into the parameter that is activated by the choice.
Work In...
Use this menu to specify whether the parameter values you enter represent pixels or fractional multipliers. The default is to work in Fractional values--floating point values between 0 and 1.
- Tip:
- Always work in fractions (or switch to fractions after entering specific pixel values) if you are working with proxies, where a lower resolution proxy input is eventually replaced by a full resolution input.
This saves you from having to account for the change later by going back into the Scale node to adjust pixel values manually--a fractional value of 0.5 in the X and Y fields will scale both the proxy and the full resolution image by half.
Fractions
First, the Fractions option must be selected from the "Work In" menu to activate this parameter. Then you can key a pair of values into the data entry fields: the left column specifies the multiplier value for the X dimension; the right column specifies the Y dimension.
Pixels
First, the Pixels option must be selected from the "Work In" menu to activate this parameter. Then you can key a pair of values into the data entry fields that represent the exact dimensions, in pixels, to which you want to scale the image: the left column specifies the number of pixels you desire for the X dimension; the right column specifies the Y dimension.
Filtering
The Filtering menu offers filtering options for the scale operation: The default is "Best (Dynamically Chosen)," which means that Chalice will choose between Mitchell and Lanczos filtering. However, you can also select "Good (Bilinear Interpolation)" to speed processing, or the fastest option, "Fast (No Filtering)," when image quality is not a top concern, as when reviewing your initial settings.
The "Advanced - User Set" option activates the parameters on the Advanced tab, which provide control over the type of filtering to use for the current imagery.
Advanced Parameter Tab
This parameters on this tab are activated when the "Advanced - User Set" option is chosen from the Filtering menu on the Scale tab.
Filter Type
The Filter Type menu includes Box, Triangle, Quadratic, Cubic, and Gaussian; as well as Catmull-Rom/Overhauser Spline, Mitchell, Sinc (Windowed), Bessel (Windowed), Lanczos 2-lobe Sinc, and Lanczos 3-lobe Sinc.
Box, Triangle, Quadratic, Cubic and Gaussian refer to the shape of the function curve that defines the filter. Box equals constant, triangle is linear; and so on in increasing precision and processing time.
The other filter options are named for the individuals who developed them, and the best choice will depend on the individual characteristics of the image as well as on the type of operation the node is performing.
The Mitchell and Lanczos filters are among the most popular filter options. Mitchell filtering is often preferred for Jpegs or images with a lot of fine lines. Lanczos filtering provides a good compromise between sharpness, ringing and aliasing reduction. It is best used on images which are reduced or scaled down.
Mitchell B and C
The Mitchell B and C parameters are activated when Mitchell filtering is selected in the Filter Type menu to enable you to make adjustments to the Mitchell filtering process.
The default values, 0.333 for both parameters, fall within a recommended range of roughly 0.25 to 0.5. Since the quality of the result must be judged based on your imagery and intentions, the best course may be to experiment with these settings.
As a guideline, note that values above 0.5 for the Mitchell B parameter may produce unnecessary blurring. Values above 0.5 for the Mitchell C parameter may result in excessive ringing (rippling patterns). When both parameters are set at values above 0.6 or below 0.2, anisotropic artifacts may appear.
Window Type
This parameter is activated only when the Sinc or Bessel filter type is selected. Choices are Hanning, Hamming, Blackman, or Kaiser.
Kaiser A
This parameter is activated only when Kaiser is chosen in the Window Type parameter.
Blur
The Blur parameter can be used with any filter option when heavy blurring is needed. The default value of 1 adds no blurring to the image; values greater than 1 represent increased blurring, up to the maximum of 4.
Table of Contents | Index
Copyright 1996-1999 by Silicon Grail Corporation |