The nodes in the Convert menu are used to perform various conversion operations on the input image.
The Bit Depth node is strictly for converting linear data to a larger or smaller linear colorspace, such as 8-bit linear data to 16-bit linear data and vice versa. Bit Depth is easy to use and allows you to change the default remapping values based on the actual high and low values in your imagery.
Lin To Log and Log To Lin , on the other hand, are strictly for converting nonlinear (log) data to linear colorspace and vice versa. Automatic conversion settings can be used, or you can customize the conversion for specific imagery.
Channel Split is used to split an RGBA image into separate streams of RGB and Alpha channel data.
The Premultiply node multiplies the RGB channels of the input by the alpha channel, while Unpremultiply restores the original RGB channel values of a premultiplied RGBA image.
Interlace and Deinterlace are designed to work with video fields. The Deinterlace node takes a single input and outputs frames of either even or odd fields, as you specify. The Interlace node requires even and odd field inputs, which it reintegrates into interlaced frames.
More video-related nodes are described in Chapter 21: Timing Nodes.
Contents | Top of Page | Index
The Channel Split node is used to split an RGBA image into two separate streams of image data: the RGB image and the Alpha channel.
Some users prefer to send the matte channel through the network separately rather than integrating the Alpha into the RGB image. This is not strictly necessary, insofar as you can use the Channel Select parameters available in most nodes to control whether the RGB or the Alpha of an RGBA input is affected by the node operation. But it can make the node network more "legible" in that you can tell at a glance whether the color or matte data is being modified by a node.
Fig. 20.1 The Channel Split node in the Worksheet, with the RGB output connected to a Color Correct node and the Alpha output connected to Roto.
The Channel Split node accepts one input, which must be a four-channel (RGBA) image. Unlike other nodes, Channel Split has two output connectors:
To merge separate RGB and Alpha channels back into a single RGBA image, use the Channel Swap Node in chapter 16. |
There are no node-specific parameters available in the Channel Split Node Panel. The node automatically splits any RGBA image you connect to it.
Contents | Top of Page | Index
The Bit Depth node converts images to and from 8-bit, 16-bit, and floating point linear data.
To use the Bit Depth node, select the output format from the Bit Depth menu in the Node Panel and RAYZ will do the rest. However, you also have the option of modifying the white and black mapping values to optimize the conversion for your imagery.
By default, the Bit Depth node remaps the maximum values of one bit depth to another. For example, assume that you wish to convert 16-bit data (0 to 65535 bits) to 8-bit data (0 to 255 bits). The 16-bit colorspace is scaled linearly by remapping 65535 to 255.
However, imagine that the data in your 16-bit image actually occupies a range of only 50 to 5050. You can scale the actual range of your input data (50 to 5050) down to the full range of the 8-bit colorspace (0 to 255) by entering a value of 5050 into the White In parameter and a value of 50 into the Black In parameter.
This enables you to spread the actual range of your input data over the maximum possible range of the output you specified.
Select the bit depth to which the input image should be converted: 8-bit linear, 16-bit linear, or floating point linear.
Optionally, you can adjust either the White In or White Out parameter value, or both, to optimize the conversion. These parameters adjust all channels equally; to control individual channel values, expand the parameter group to reveal the channel controls.
The "White" parameters represent the maximum values to be used for the conversion, with White In representing the input data and White Out representing the output data.
The White In and White Out parameters provide an eyedropper tool you can use to sample an area of the image by clicking on an individual pixel or by scrubbing across an image area. (Remember to set the Image Viewer's Source menu to Input Image to sample the input image values.)
The maximum value per channel of the sampled pixels will be used to set each individual color channel parameter. This means that you have expand the group to see the resulting values in the Red, Green, and Blue parameters. The master parameter value is not modified by the eyedropper selection.
Optionally, you can adjust either the Black In or Black Out parameter value, or both, to optimize the conversion. These parameters adjust all channels equally; to control individual channel values, expand the parameter group to reveal the channel controls.
The "Black" parameters represent the minimum values to be used for the conversion, with Black In representing the input data and Black Out representing the output data.
The Black In and Black Out parameters provide an eyedropper tool you can use to sample an area of the image by clicking on an individual pixel or by scrubbing across an image area. (Remember to set the Image Viewer's Source menu to Input Image to sample the input image values.)
The minimum value per channel of the sampled pixels will be used to set each individual color channel parameter. This means that you have expand the group to see the resulting values in the Red, Green, and Blue parameters. The master parameter value is not modified by the eyedropper selection.
Contents | Top of Page | Index
The Deinterlace node enables you to de-interlace video frames into frames of either odd or even fields, as you specify.
You can use two Deinterlace nodes to create separate sequences of the odd and even fields, which can be reintegrated in the complementary Interlace Node.
To create a sequence of alternating odd and even fields for each frame, use the Split Node in chapter 21 instead. |
Fig. 20.2 You can specify whether to output the odd or even field, and whether to fill in the missing field lines or to throw them away.
Use the Field menu to select the odd fields (lines 1, 3, 5, etc.) or the even fields (lines 2, 4, 6, etc.). If you choose "Odd," for example, the Deinterlace node will retain the odd field data.
This parameter enables you to specify how to treat the vacant spaces that remain once the Deinterlace node removes either the odd or even scan lines. Select one of the following options from the menu:
Use Black This option fills the vacant lines with black.
Copy from Previous This option fills the vacant lines with data replicated from each previous line.
Average Above and Below This option fills the vacant lines with data obtained by averaging the preceding and subsequent lines together.
Throw Away This option discards the data that formerly occupied the now-vacant lines. It has the effect of shrinking your file by 50% in the Y dimension. (In other words, an image that was originally 720 x 486 will become 720 x 243.)
Contents | Top of Page | Index
The Interlace node recombines video imagery that has been split into separate sequences of odd and even field frames, which means that the Interlace node requires two inputs.
You can use the Interlace node to recombine frame fields that you de-interlaced using the Deinterlace Node.
The Field Order and Input Style parameter settings should be based on the method used previously to de-interlace the frames using the Deinterlace node. |
The Field Order menu is used to specify which input should be used to create the odd fields (lines 1,3,5, etc.), and which input should be used for the even fields (lines 2, 4, 6 etc.):
The Input Style menu is used to specify how the fields will be interlaced. The options are complementary to those available when using the "Other Field Fill" parameter in the Deinterlace node:
Contents | Top of Page | Index
The Lin To Log node enables you to convert linear imagery to log (10-bit Cineon) format. The node will perform the conversion for you automatically; however, you can adjust the individual parameters if necessary to optimize the conversion for your imagery.
The complement to this node is the Log To Lin Node, which converts Cineon 10-bit log imagery to a linear format. If you need to convert linear imagery to another linear bit depth, use the Bit Depth Node.
The log image output by the node is displayed in the Image Viewer using the Raw Log display conversion method. You can choose Cineview emulation instead |
See also the description of the Log To Lin Node for information about the Kodak specifications on which default Cineon conversion values are based.
By default this menu is set to "Automatic," which converts the imagery to log using default parameter values. Select "Manual" to activate the other conversion parameters and set the values yourself.
These parameters become active when you choose a Manual from the Conversion menu. Each is a master control that affects all channels equally. To adjust any channel individually, expand the parameter group to access the RGB controls.
The Reference Black parameter enables you to modify the default log value (95).
The Log 90% White parameter enables you to modify the default log value of 685 for reference white.
The Linear 90% White parameter enables you to modify the default linear value (65535 in 16-bit) for peak white.
You can use the eyedropper tool associated with this parameter to sample an appropriate area of the image. Set the Image Viewer Source menu to Input Image and click on a pixel or scrub across an area. The maximum value sampled per channel will be used in the Red, Green, and Blue channel parameters (expand the group to access). The master Linear White value is not affected by the eyedropper selection.
The Display Gamma parameter enables you to modify the default conversion value (1.7) for display gamma.
The Film Gamma parameter enables you to modify the default conversion value (0.6) for film gamma.
Contents | Top of Page | Index
The Log To Lin node converts Cineon 10-bit log data to linear format at the bit depth per channel you specify.
To convert linear image data into 10-bit log, use the Lin To Log Node. To change one linear bit depth to another, use the Bit Depth Node.
The default values used in the Log To Lin Node Panel parameters are based on specifications published by Kodak Motion Picture & Television Imaging and Cinesite Digital Film Center in "Grayscale Transformations" (1993) and "Conversion of 10-bit Log Film Data to 8-bit Linear or Video Data" (1995).
These documents can be downloaded in Acrobat PDF format, suitable for printing, from Silicon Grail at ftp://ftp.sgrail.com/pub/reference/cineon. They are also posted on the Technical Documents page of the Cinesite Hollywood website at http://www.cinesite.com/la/scanrec/techdocs.html. |
Select the linear bit depth to which the log input image should be converted: 8-bit linear, 16-bit linear, or floating point linear. The node will then perform the conversion for you automatically; however, you can adjust the individual parameters if necessary to optimize the conversion for your imagery.
You can use the default values for these parameters, or you can modify them to suit your needs and imagery. Each is a master control that affects all channels equally. To adjust any channel individually, expand the parameter group to access the channel-level controls.
This parameter specifies the value used for reference black in the conversion operation. The default value is 95, which represents Dmin (the minimum printing density, or blackest black that can be recorded, about equivalent to the 1% black card).
This parameter specifies the value used for reference white in the conversion operation. The default is 685, which represents the code value of the 90% white card for a normally exposed film negative.
You can use the eyedropper tool associated with this parameter to sample an appropriate area of the image. Set the Image Viewer Source menu to Input Image and click on a pixel or scrub across an area. The maximum value sampled per channel will be used in the Red, Green, and Blue channel parameters (expand the group to access). The master Log White value is not affected by the eyedropper selection.
This parameter specifies the value in linear space to which the log 90% white value will be mapped in the conversion operation. The default is 65535 (in 16-bit), which will clip any values above 685 in the original log file. When converting to linear float, however, values are not clipped at 1.
The Display Gamma parameter enables you to modify the default conversion value (1.7) for display gamma.
The Film Gamma parameter enables you to modify the default conversion value (0.6) for film gamma.
Softclip can be used to reduce the effects of harsh clamping at the high end when remapping log data to 8- or 16-bit linear (highs are not clamped when converting to linear float). Check the Softclip box to enable the Softclip parameter and set a positive value in the range of 0-100.
If the Softclip parameter is left at its default value of 0, the distribution curve used to remap the image data remains linear along its entire length from low to high. In other words, this is a "hard" clip of input values above reference white.
When the Softclip is a positive value, the softclip value is subtracted from the Log 90% White value to create a breakpoint below peak white. Above this breakpoint, the slope of the distribution curve becomes nonlinear, gradually leveling off the highs.
Fig. 20.3 The graph on the right shows how the maximum softclip value affects the redistribution curve, as compared to the default conversion (left).
Contents | Top of Page | Index
The Premultiply node takes a single RGBA input. For every pixel in the image, the Premultiply node multiplies the value of each color component (the red, green, and blue channels) by the corresponding alpha channel value.
Premultiplication is an integral step in the digital compositing process, as it transfers the transparency level represented by the alpha channel to the RGB components themselves before the RGB channels of two separate images are blended in a compositing operation.
In general, CG (computer-generated) imagery is premultiplied, while digitized film footage is not.
It is not necessary to use the Premultiply node to composite imagery in RAYZ. Composite nodes will premultiply the input images for you, unless the image has already been premultiplied.
However, the Premultiply node can be used, if desired, to premultiply imagery that will not be subject to a composite operation, as when RAYZ is used strictly for color correction and you want the output rendered as premultiplied imagery for use in another application.
The Premultiply operation is automatic and does not require user modification; therefore, there are no parameters to set in the Premultiply Node Panel.
The Premultiply node performs the following computation, where "A" represents the input image; and r, g, b, and a represent the red, green, blue, and alpha channels:
The multiplication in this formula is done at floating point precision, and for floating point images the alpha channel value is assumed to be in the range of 0-1. For 8-bit and 16-bit images, however, the alpha channel value (Aa) is converted into a decimal fraction first by dividing it by an appropriate value: for 8-bit, the alpha is divided by 255; for 16-bit, it is divided by 65535.
Contents | Top of Page | Index
The Unpremultiply node attempts to restore the RGB channels of a premultiplied RGBA image to their original, non-premultiplied values.
It is important to note that the accuracy of this operation depends on the alpha channel values in the image. If the alpha channel value of a pixel is zero, you lose all information that would allow you to accurately reconstruct the original value of each color component.
Premultiplication is an essential first step in most composite operations (see also About Premultiplication in chapter 18 for more information), and computer generated imagery, such as the output from a 3D rendering program, is almost invariably premultiplied.
An image is premultiplied with the assumption that it is ready to be composited and that the individual image channel values will not be further adjusted independently.
However, you may very well need to adjust such an image before compositing it over a background--you may need to color correct the RGB channels without affecting the opacity or to adjust the matte edges without affecting the color characteristics. In these cases, you need to work with non-premultiplied image channels to get the results you want.
For example, in a non-premultiplied image, a full red pixel with half alpha coverage would be represented as [1, 0, 0, 0.5]. After premultiplication, the same pixel would be represented as [0.5, 0, 0, 0.5].
Now assume that you adjust the matte such that the alpha channel value for this pixel changes from 0.5 to 0.75.
If this matte adjustment occurs after the pixel has been premultiplied, the value becomes [0.5, 0, 0, 0.75], whereas if the adjustment occurs before premultiplying, the pixel value becomes [0.75, 0, 0, 0.75]. Obviously, these two values would generate a different result for the same pixel in a composite.
Therefore, you may choose to run an image through the Unpremultiply node before adjusting it, and then have the composite node premultiply the adjusted image.
The Unpremultiply operation is automatic and does not require user modification; therefore, there are no parameters to set in the Unpremultiply Node Panel.
The Unpremultiply node performs the following computation, where "A" represents the input image; and r, g, b, and a represent the red, green, blue, and alpha channels:
"Q" represents the alpha channel value, modified as follows:
Q = Aa == 0 then 1.0 else 1.0/Aa
For 8-bit and 16-bit images, the alpha channel value (Aa) is converted into a decimal fraction (floating point value) first by dividing it by an appropriate value: for 8-bit, the alpha is divided by 255; for 16-bit, by 65535.
Contents | Top of Page | Index