ImageReformat

Reformat lets you resize your image sequence width and height using the incoming displayWindow to determine the scale factor. This also allows you to use plates of varying image resolution on a single recipe without running into issues when combining them.

Note: If no resize is needed, filtering is NOT applied (unlike in the ImageTransform2D node, which always applies filtering).

Connection Type	Connection Name	Function
Input	bg	The background image.
Input	out_mask	An optional image to use as a mask. By default, the merge is limited to the non-black areas of the mask.

Control (UI)	Default Value	Function
[2D node controls]	N/A	Set the controls for the stereo view. For more information, refer to the Common 2D Nodes Widget Type in Common Parameter Widgets.
resolution
[resolution]	Dependent on Project Settings	The format to which you want to output the image sequence. The default setting resizes the image to the format indicated in the Project Settings.
width	globals.width	Sets a custom reformat width. Note: You can only edit this field when the initial resolution is modified.
height	globals.height	Sets a custom reformat height. Note: You can only edit this field when the initial resolution is modified.
preserveAspect	enabled	When enabled, Katana preserves the input image's aspect ratio.
allowOverscan	disabled	This allows the node to generate overscan (if possible). Overscan refers to image pixel data outside of the displayWindow and can be inspected using options in the Monitor.
preserveAspect: enabled
center	enabled	When enabled, Katana pads the output image if any gaps remain after reformatting while preserving the original aspect ratio.
filtering
downFilter	Lanczos3	The filter kernel to use for downsampling: • Gaussian • Triangle • Box • Bell • BSpline • Sinc • Lanczos2 • Lanczos3 • Lanczos5 • Mitchell • Bilinear • Bicubic • Nearest
upFilter	Mitchell	The filter kernel to use for upsampling: • Gaussian • Triangle • Box • Bell • BSpline • Sinc • Lanczos2 • Lanczos3 • Lanczos5 • Mitchell • Bilinear • Bicubic • Nearest
highlightCompensation	enabled	When enabled, Katana adaptively compresses pixel values prior to transform filtering and re-expands them afterward. This helps to reduce the ringing in high-contrast areas that can be a problem in linear floating point images.
clampOutput	enabled	Filtering can introduce negative values and send values above 1.0. When clampOutput is enabled, Katana clamps the RGB channels low at 0 and the alpha channel between 1 and 0 after the image is filtered. This is recommended for transforms on color/alpha images. When clampOutput is disabled, no clamping is done and values below 0 and above 1 are allowed. This is recommended for transforms applied to images that contain data that may (correctly) range more widely.

You are here: