The Matte nodes enable you to create and modify matte channel data for use in compositing and other image processing operations.
In RAYZ, the term "matte" is generally used synonymously with "alpha" to describe the image channel that holds the opacity values for each RGB triplet and is necessary to most compositing operations. When appropriate, alpha may be used to denote a channel of image data and matte to describe the channel functionally.
The Chromakey and Lumakey nodes pull mattes from RGB imagery by using color and luminance values respectively. The Roto node enables you to create matte shapes by drawing them. Erode Dilate is used to shrink or expand edges, operations most commonly performed on mattes.
The Ultimatte nodes provide a suite of tools for working with bluescreen imagery: they correct flawed backing, kill grain, create a processed foreground ready for compositing, and match colors in the foreground with equivalent objects in the background (or between any two images).
It is recommended that you start with the introduction to the Ultimatte nodes to find out which node to use for what operation and the best order in which to use them.
The Chromakey node creates an alpha channel matte based on the colors of the input image, within a narrow range of hue, saturation, and luminance that you specify in the Chromakey Node Panel.
You can output all channels or the alpha channel only.
Chromakey accepts two inputs, the primary input image to be chromakeyed and an optional second input that is used for a garbage matte. For more information, refer to Using a Garbage Matte in the Ultimatte node description.
You can connect an input that already has an (unsuitable) alpha channel. The Chromakey node will discard the incoming alpha channel values. |
Start by specifying the range of HSL values to key on. Pixels in the input image that fall within the ranges specified for hue, saturation, and luminance in the Node Panel are assigned alpha channel values of 1 (that is, the maximum value for the color depth) and all other pixels in the alpha channel become black (0).
You can select a range of values in the Chroma wheel in the Node Panel or use the Chromakey eyedropper to sample an area of the RGB image in the Viewer.
Then switch the Image Viewer display to Alpha to examine the resulting matte channel, adjusting the Key parameters as necessary to pull an acceptable matte:
The Key parameter group includes the Hue, Saturation, and Luminance parameters that specify the range of image values to use in the Chromakey operation. All three parameters have a pair of fields: one specifies the low end of the range and the other, the high end.
You can set these parameters by typing numerical values into the fields or by using the associated eyedropper to sample an area of the RGB image in the Viewer. The Hue, Saturation, and Luminance parameters will update accordingly. The Hue and Saturation values can also be set using the Chroma Wheel.
Fig. 15.1 Chromakey parameters.
Select Scrub or Drag Box from the eyedropper menu and click the eyedropper button. Move the eyedropper cursor over the image in the Viewer and scrub an area or drag a bounding box around it to sample it.
The minimum and maximum values of the sampled pixels will be used to set the Hue, Saturation, and Luminance parameter values.
You can specify a range of Hue and Saturation values by adjusting the selection box in the chroma wheel. The selection box and its borders can be dragged to resize and reposition the box in the wheel:
The Hue parameter defines the range of hues to use within a range of 0 to 360 degrees, which represents the circumference of the color wheel.
To key on a typical bluescreen blue, for example, you might specify a range of 220 to 240 and then adjust it as necessary for your imagery.
The Saturation parameter defines the range of saturation to use within a range of 0 to 1, which represents the scale from completely desaturated to fully saturated.
Using a bluescreen shot as an example, you might specify a range of 0.5 to 1, which represents the saturation range typical of bluescreen backings.
The Luminance parameter defines the range of luminance to use within a range of 0 to 1. For some imagery, the luminance values of the input image can be used effectively to help pull the matte.
The Luminance slider control has two vertical slider bars. The one on the left is used to set the minimum luminance value and the one on the right, the maximum luminance value. You can also drag the area between the bars to shift the range up or down without changing its magnitude.
The Softness parameter can be used to soften the alpha channel matte generated by Chromakey. By default, the Softness value is 0, which creates a hard, opaque matte in which all pixels are either white or black. As you increase the Softness value, the matte becomes more transparent.
Check the Alpha Only box to output only the alpha channel from the Chromakey node. Otherwise, the node outputs the RGBA channels.
Check the Invert Alpha box to invert the values of the alpha channel created by the Chromakey node: white pixels become black, and vice versa.
This menu becomes active when an optional garbage matte input is connected to the Chromakey node. It is used to specify which channel of this input to use as the garbage matte, when the image has more than one channel. For more information, see also Using a Garbage Matte in the Ultimatte node description.
The Erode Dilate node will erode or dilate edges in an image, as you specify. Any or all channels can be adjusted, but this node is used primarily to modify matte edges in an alpha channel for better composites.
The dilate operation works by dilating, or spreading, brighter areas in the image while it erodes, or shrinks, the darker areas. The erode operation, on the other hand, erodes brighter areas and dilates the darker areas.
The Ultimatte AE (AdvantEdge) Node in chapter 18 is designed specifically to provide matte edge controls when compositing an Ultimatte foreground image over a background. You may want to use Ultimatte AE rather than Erode Dilate in such cases. |
The Erode Dilate node accepts two inputs, the image to be processed and an optional mask input used to control which pixels of the primary input are processed. For more information, see Using Mask Inputs in chapter 7.
The Operation menu is used to select Erode or Dilate for the operation. Erode is selected by default.
The Magnitude parameter controls the size of the effect. The range of the slider control is 0 to 0.05, to allow fine control over the adjustment, although the upper end of the range is unconstrained. The default value of the Magnitude parameter is 0.01.
The Softness parameter uses subpixel accuracy to modify the softness, or falloff, of the effect in a range of 0-1. When the Softness parameter is set at 0, the original edge pixel values are used to erode or dilate the edge by the specified Magnitude value.
Nonzero Softness values, however, will gradate the pixel values from the original edge inward (for erode) or outward (for dilate) to the new edge. The greater the Softness value, the steeper the slope, or falloff. A Softness value of 1 produces a circular gradient from the original edge to the new edge.
Use this parameter to specify which channels of the input image will be processed by the node. By default, only the alpha channel is selected.
The Lumakey node creates an alpha channel matte based on the luminance of the input image, within a narrow range of black and white that you specify in the Node Panel. You can choose to output all channels or the alpha channel only. You also have the option of inverting the outgoing alpha matte.
Lumakey accepts two inputs, the primary input image to be lumakeyed, and an optional mask input that controls which pixels in the primary input are used in the lumakey operation. For more information, see Using Mask Inputs in chapter 7.
You can input an image that already has an (unsuitable) alpha channel. The Lumakey node will discard the incoming alpha channel values. |
Use this menu to specify whether the Lumakey node should output all channels (the default) or the alpha channel only.
Check this box to invert the matte channel created by the node.
Use this menu to select the type of luminance to key on:
You can use one of the existing color channels, or have the node generate a luminance channel from the RGB image. For the formulas used, refer to Appendix A: How RAYZ Computes Luminance Values.
Use the Black Min and Black Max parameters to set the range of blacks to use in the Lumakey operation.
Use the White Min and White Max parameters to set the range of whites to use in the Lumakey operation.
The Roto node is used to create and manipulate closed spline shapes, typically in the matte channel, although you can apply them to any image channels you choose. You can draw, edit, scale, rotate, invert, and animate these rotospline shapes. Most commonly this functionality is used to create garbage mattes or mask images.
The section of this node description on Using the Roto Node explains how to draw and edit shapes using the Roto tools in the Image Viewer, while the section on Shape Controls in the Roto Node Panel explains how to use the Node Panel parameters provided for each shape you create. |
Garbage mattes are used to remove unwanted elements from an image before compositing it with other imagery. For example, a garbage matte could be used to remove lights or other equipment from a bluescreen shot before the subject was composited over a background scene.
Fig. 15.3 Nodes with dedicated garbage matte inputs, such as Ultimatte, assume that pixels in the garbage matte with non-zero values represent areas that should be matted out of the image being processed.
The Ultimatte and Chromakey nodes accept an optional garbage matte input, as described in Garbage and Holdout Matte Inputs.
The output of a Roto node can just as easily be used as a holdout matte input to an Ultimatte node to prevent print-through in a foreground element. To create a holdout matte you would draw matte shapes defining the areas that should be opaque when composited.
A Roto matte can also be used as a mask; that is, to control which areas of an image are processed in another node, such as a color correction node. For example, a matte could be drawn to isolate an automobile that was filmed in black-and-white. The original shot could then become the primary input to an Indexed Color node, and the output of the Roto node would become the mask input. This would enable you to tint only the car while leaving the background monochrome.
For more information about using mask inputs in RAYZ, refer to How Masks Control Node Processing in chapter 7.
The Roto node accepts one input, which may or may not have an alpha channel:
If you will be working on the RGB channels only, you can prevent the node from creating an alpha channel for an RGB input by unchecking the Add Alpha box. |
The node will output all of the image channels, or the alpha channel only, as you specify in the Output menu in the Node Panel. The default is to output All Channels.
Fig. 15.4 The Roto Node Panel provides menus to control the incoming alpha channel and the channels to output.
The Roto node provides tools directly in the Image Viewer for drawing closed splines around portions of an image, so you always start by displaying the Roto node image in a Viewer.
Fig. 15.5 Roto node tools available in the Image Viewer.
Display the RGB image in the Viewer to use it as a reference when drawing a spline shape that will be used to isolate a specific object or area in the image.
Display the Alpha channel to view the roto matte and adjust its transparency level or edge width. To evaluate an animated matte, use the Flipbook controls in the Viewer to play the animated roto sequence.
You can create as many roto shapes as you need for an image. For each shape you create, a corresponding entry is created in the Roto Node Panel. The Node Panel list enables you to control overall characteristics of each shape individually, as described in Shape Controls in the Roto Node Panel.
If you intend to animate the outline or position of a shape over time, you can turn on Autokey before you start. This will animate the shape automatically by creating a keyframe whenever you go to a new frame in the sequence and modify the shape in any way. You can also use the Animation menu for the shape entry in the Node Panel. See also Animating Parameter Values in chapter 7.
You can add, delete, and modify shapes and points at any frame. For example, you could draw a shape at frame 1, reposition it at frame 15, and add a point at frame 25 to reshape the outline. And you could create another shape starting at frame 7 that held its position until you deleted it at frame 100, and so on.
The Roto tool strip offers buttons used to specify drawing and editing modes. Freehand drawing mode is selected by default when a new Roto node is created, however, you can select Ellipse or Rectangle instead. Once you have closed a shape, the node automatically switches to an edit mode. To draw another shape, select a draw mode again.
Fig. 15.6 The default Draw mode is Freehand, in which you draw one segment at a time in order to follow the outline of an object in the image. The other two drawing modes, Ellipse and Rectangle, enable you draw entire shapes in one motion.
Click the Freehand button to select it, if necessary.
To draw a linear spline (straight line), click anywhere in the image area to create an anchor point, move the mouse, and click again to create a second anchor point. RAYZ will draw a linear spline segment connecting the two points. Continue moving and clicking the mouse to add additional segments.
To draw a curved spline (with point handles), drag in the image to create the anchor point instead of clicking (the farther you drag, the longer the control handle on the point). Release the mouse button and a spline will appear "attached" to the cursor. Move the cursor to wherever you want to place the next point and either click (to start a new linear segment) or drag again (to start a new curved segment).
To close the shape, click on the first point you created. Once the shape is closed, you can manipulate the shape as a whole or any of its points as described in Editing a Shape.
For maximum efficiency, use as few points as necessary to create a shape. |
In addition to drawing a shape freehand, you can select one of the other two drawing modes to create an ellipse or rectangle. With the appropriate drawing mode selected, drag in the image to draw the shape in one motion.
To constrain the proportions of the shape to a circle or square, hold down the Shift key while dragging. |
Once you have drawn a shape, you will probably want to adjust it. One of the editing modes should be selected by default; if not, click an Edit Mode button.
There are two point editing modes to choose from, however, the choice is only relevant when you are manipulating Bezier handles (the handles on points that control curved spline segments).
In the default point editing mode (the Edit button on the left), dragging a point handle adjusts the curve on both sides of the point equally. In split mode (the Edit button on the right), each end of the point handle is adjusted independently.
You can modify an entire shape or shapes, or any individual point or points, depending on what is currently selected:
Fig. 15.8 A selected point is solid, while an unselected point is hollow. Point handles appear when a point that controls a curved segment is selected.
As you move the cursor over the image, the line, point, or point handle currently under the cursor will highlight in a contrasting color so that you can tell which object you will be selecting when you click.
To change the outline of a roto shape, adjust the position, type, and number of points in the shape:
To reposition a selected shape, drag it by any of the selected points. (You can't drag a shape by its splines.)
Use the arrow keys on the keyboard to nudge a selected point or shape in one-pixel increments in the direction indicated by the arrow. You can also hold down the Shift key as you press an arrow key to move in larger increments (4 pixels).
To change the default increment values, go to Edit > Preferences > Settings and select Nudge Size (Small) or Nudge Size (Large).
You can also adjust the slope and acceleration of a curved spline by manipulating the point handles, which appear whenever the point is selected. To adjust the slope and acceleration of a curve, drag the handles that extend from the anchor point:
To adjust the segments on each side of a spline point separately, switch to Split Handle mode, as described in Choosing an Edit Mode.
As described above in Freehand Drawing, you create linear or curved segments as you draw a shape. However, you can change any control point after the shape has been drawn. Click the point to select it, and then hold down the Control key and click it again: Linear points will become curve points and vice versa.
Click the Rotate or Scale buttons to choose one of these modes. When an entire shape is selected, the rotation or scaling applies to the entire shape. When an individual point is selected, the operation applies to the selected point (or points) only.
Fig. 15.9 The Transform modes are Rotate, which rotates selected points or shapes, and Scale, which scales selected shapes up or down. (You can translate selected points or shapes in Edit mode.)
In Rotate mode, a pivot point marker appears in the image. This indicates the pivot around which the selected items will rotate. By default, the pivot appears in the center of the image. To change the pivot, drag it to a new location.
When animating a shape across time, you may want to use Rotate mode to reposition the shape (or individual points) in a smooth arc to mimic certain types of motion. |
The Roto tool strip also provides buttons to control the color of a shape's outline; that is, the color used to display the spline segments and points of the shape overlay.
Fig. 15.10 You can change the color of any shape outline to make it easier to distinguish over the image, or to color-code related shapes.
The Outlines button toggles auto-increment mode on and off. In auto-increment mode, each new shape you create is assigned the next color in the palette. If you turn this feature off, however, each new shape is assigned the color currently selected in the palette.
After a shape has been created, you can change the color of its outline display by selecting a different color in the Overlay Color menu for the shape, which is located in the Roto Node Panel. |
An entry is created in the Splines list of the Node Panel for each shape you create.
Fig. 15.11 Shape-specific parameters available in the Roto Node Panel for every shape you create.
Using the controls in the top line of the shape entry, you can
When a shape entry is expanded, you have access to additional parameters that control shape characteristics such as opacity, edge width, and blend operation, as well as which image channels will be affected by the shape.
This parameter enables you to change the color of the shape overlay; that is, of the splines with their points and handles. You can use this feature to color code related shapes or make the overlay more visible over the image.
To change the overlay color, click and hold the Color button to access a menu of color swatches and select a different color.
The Operation menu specifies how the roto shape will be blended with the underlying image and any other overlapping roto shapes. The default blend mode is Mix, however you can also choose Add, Subtract, Minimum, or Maximum.
The Edge Width parameter is used to feather the edges of a shape by creating a gradual falloff either inward or outward from the edges. The default value is 0, which creates a hard edge that follows the spline outlines. Negative values create an inward falloff and positive values create an outward falloff.
The Opacity parameter specifies how much of the underlying image shows through the shape; that is, it controls the opacity of the shape and not the image channel.
This means that the effect of the opacity setting depends on the color of the shape, the image channel values under the shape, and whether there is any overlap with other shapes.
By default a shape is fully opaque. If you draw a roto shape over an area of the alpha channel that has a mixture of values, for example, you can use the Opacity parameter to control how much the alpha channel values show through the roto shape.
Fig. 15.12 The Opacity value controls how much of the underlying image shows through the shape. As this example illustrates, zero opacity does not make opaque pixels in the alpha channel transparent.
This parameter group specifies the color values to apply to the channel data of the pixels delineated by the roto shape. Expand the parameter to access the standard RAYZ color parameters. (See also Using the Color Parameters in chapter 14.)
Most commonly this parameter group is used to specify the alpha channel value for the shape. The default value, white, makes the area of the outgoing alpha channel under the shape fully opaque. Changing the value to 0 would make the area black, that is, completely transparent.
You can matte out an unwanted element in an otherwise adequate matte by drawing a roto around problem area and setting the Alpha channel value in the Render Color parameters to 0. |
Fig. 15.13 In the image on the left, the Alpha channel value in the Render Color parameters is set to maximum (255 for an 8-bit image). The example on the right shows the effect of an Alpha value of 0.
Use the Channel Select parameters to specify which channels of the image will be affected by the roto shape. By default, only the alpha channel is selected, but you can actually apply the roto shape to any or all channels in the image.
Ultimatte is bluescreen compositing software for film and video. It was designed by the Ultimatte Corporation, and its functionality is included in RAYZ. (You do not need a separate Ultimatte license to use it.)
Ultimatte is designed to optimize the bluescreen compositing process by seamlessly compositing imagery while retaining shadows, fine detail, and transparent objects in the foreground image.
As an example of the bluescreen process, a filmmaker might shoot an actor in front of a background that has been painted entirely blue, using paints formulated specifically for this purpose such as Rosco Labs Ultimatte Blue #5720, or Ultimatte Super Blue #5722, for wire removal. This bluescreen background area is referred to as the backing area or screen area .
Later, using the Ultimatte nodes, a digital artist can prepare a processed foreground image (an image of the actor, with the backing matted out) and seamlessly composite it over a background clip of a burning building, lunar landscape, or other dramatic scene.
In actuality, the screen may be blue, green, or red; in fact greenscreen is used quite commonly. However, this manual uses the term "bluescreen" generically. |
RAYZ includes several Ultimatte matte nodes, each of which is optimized to solve a particular type of bluescreen image processing problem:
This node generates a processed foreground and matte; that is, an image of the foreground subject with any spill removed and the backing area suppressed to black, and the Ultimatte matte attached as the alpha channel. You may later composite this image using a node such as Multi-comp or Ultimatte AE.
If you have a reference clip (clean plate), this node can be used to correct uneven areas in the bluescreen backing behind the foreground subject. Use Ultimatte CSC before generating the matte in the Ultimatte node.
This node can be used to remove film grain or video noise from the bluescreen backing area of a foreground before generating the matte in the Ultimatte node.
This node enables you to adjust the color balance of one film clip to make it appear natural or appropriate for the color values of a background clip when composited. Use Ultimatte CC after the other Ultimatte matte nodes.
RAYZ also includes an Ultimatte composite node. While any suitable composite node in RAYZ can be used to composite an Ultimatte processed foreground image, the Ultimatte AE node employs a method developed by Ultimatte to help fix any edge problems in the composite. For more information, see the description of the Ultimatte AE (AdvantEdge) Node in chapter 18. |
The following steps provide an overview of how the Ultimatte nodes are used in a typical compositing network. For detailed information on each node and how it is used, refer to the individual node descriptions that follow this overview.
Fig. 15.14 Example of a compositing network using the Ultimatte matte nodes.
The foreground image is connected to an Ultimatte CSC node to correct any flaws in the backing, or screen area. In an ideal situation, the screen area against which an actor is filmed would be of perfectly even tone. In reality, the backing almost always shows imperfections due to variations in ambient lighting and flaws in the surface. The Ultimatte CSC (Classic Screen Correction) node enables you to perfect the screen area while retaining shadow information.
A clean plate of the stage, shot at the same time as the foreground subject, is connected to the second input of the Ultimatte CSC node to be used as the reference screen for the correction process.
For filmed imagery (or for a noisy video clip) the Ultimatte GK node is used to remove grain from the backing area of the foreground image. This will prevent the appearance of a double layer of grain when the foreground is composited over a filmed background.
The Ultimatte GK node should be used after the screen area has been corrected using the Ultimatte CSC node and before the final matte is created in the Ultimatte node.
The output of the Ultimatte CSC node, or GK node, if used, is connected to the Ultimatte node to create a processed foreground that is ready to be composited. The Ultimatte node enables you to remove flare or spill on foreground objects and create a matte that is output as the alpha channel of the image.
The processed foreground image is composited with the background image using a composite node. If the foreground image is one of many foreground elements to be composited, you would use a Multi-comp node to composite all the layers.
When you use the Ultimatte nodes to prepare bluescreen images, the best way to ensure the optimal result for any particular shot is to review the imagery as it looks when composited over the background in one of the RAYZ composite nodes. This is the true test of whether the various node parameter settings need further adjustment.
In some cases you may find that the color balance of the foreground image does not match the background image perfectly in the composite. To correct this situation, you would insert an Ultimatte CC node into the network between the Ultimatte node and the composite node and use the Ultimatte CC controls to correct the color balance of the foreground image.
The Ultimatte matte nodes enable you to create even bluescreen backings, remove film grain, process the foreground image, and correct the color balance of an image to be composited.
The output of the process will usually be a single processed foreground image suitable for seamless compositing in another RAYZ node.
The Ultimatte process is designed to create highly accurate matte data based on the color values of the foreground image. Most colors occurring in nature have at least some amount of red, green, and blue when represented in RGB colorspace. Given this fact:
Using a blue backing screen as an example, the colors used for the bluescreen backing will have extreme differences between the blue channel value and the red and green channel values.
The Ultimatte process starts by examining the relationships among the values of the three channels of an RGB image. For each pixel, Ultimatte determines which has the greater value, the red channel or the green channel. This value is then subtracted from the blue channel value to obtain a difference value for the pixel.
Areas where this difference value is pronounced are interpreted as backing. The greatest difference value is called the peak point, and is considered to be totally transparent for purposes of the final composite; that is, 100 percent of the background will show through the foreground wherever a pixel has this value.
In reality, unobstructed areas of the backing will display some variation in color. This means that the pixel with the highest peak point value might not be the most representative blue in the backing, so Ultimatte enables you to specify the ideal blue to use as the peak point.
Otherwise, unobstructed areas of the backing that have slightly lower difference values than the peak point could be interpreted as if they were not entirely transparent (as if they were in a faint shadow, perhaps).
Areas where the difference value is 0 or less are interpreted as foreground and considered to be totally opaque for purposes of the final composite. For example, a pure white pixel has equal values for R, G, and B, so its difference value is 0; and negative numbers will result for colors where B is less than the max of R or G.
Pixels with values that fall between the peak and minimum difference values are assigned varying levels of opacity in a linear distribution. This is what allows Ultimatte images to retain fine details, shadows in the backing area, and soft edges around foreground objects when composited over a background.
However, this also means that blue areas in the foreground subject may be treated as backing and result in print-through in the final composited image. "Print-through" refers to areas where the background is visible, or partially visible, through foreground objects that are supposed to be opaque.
Ultimatte can correct print-through problems in a matte by altering the formula used to determine the difference value for each pixel: a matte density value is assigned for the image, and the max of R or G is first multiplied by this matte density value before it is subtracted from blue.
Boosting the value of the channel that is subtracted from blue results in a lower difference value. When the difference values for an image are decreased in this way, pixels which previously had some transparency can become entirely opaque, eliminating print-through.
Raising matte density too much, however, can harden matte edges and eliminate fine detail, with adverse affects on the realism of the composite.
The Ultimatte nodes include a number of controls that enable you to adjust these values as needed for any particular image.
In the Ultimatte node, for example, adjusting the Matte Controls affects the alpha channel that is output for subsequent compositing. In the CSC node, on the other hand, the matte density parameters affect the matte data that is used in the screen correction operation applied to the RGB channels.
The Ultimatte node is used to create and output a clean, processed foreground image: that is, an image of the foreground subject with any flare removed, the backing area suppressed to black, and the Ultimatte matte attached as the alpha channel. You will then be able to composite this image using one of the RAYZ composite nodes.
You can specify whether the node will output only the matte channel or all four (RGBA) channels.
Matte density values can be adjusted to correct any holes in foreground areas of the matte caused by spill or flare that would result in print-through in the final composite. The Ultimatte node also provides flare controls to remove any discoloration caused by spill from foreground objects.
Refer to Using Ultimatte for general guidelines and to Ultimatte Parameters for descriptions of specific Node Panel parameters.
The Ultimatte node accepts three inputs:
The optional inputs are designed to take a garbage or holdout matte. The Garbage matte input is labeled with a "G" and the holdout matte input with an "H."
Usually a simple shape, a garbage matte is used to eliminate, or matte out, unwanted elements from an image that will be used as a foreground in a composite. Typically this "garbage" consists of rigging, microphones, or other equipment that showed up in frame, or a flaw in a bluescreen backdrop. A garbage matte input can also be used to extend the backing area in cases where the bluescreen was not large enough to fill the frame.
White areas in a garbage matte define the areas in the foreground image that should be matted out--that should become transparent in the matte channel that will be output from the node. Garbage matte pixels with a value of zero (that is, black) are simply ignored.
A holdout matte, on the other hand, is used to specify that certain areas of the foreground do show up in the composited image; that is, to specify that these areas will be opaque in the foreground matte that is output from the Ultimatte node.
It may be necessary to use a holdout matte input to prevent print-through of blue or very shiny foreground elements--to fill in holes in the foreground matte--when the Matte Controls cannot be used for this purpose without adversely affecting the matte edges.
Unlike a garbage matte, the white areas in a holdout matte define areas in the foreground image that should be opaque when composited.
The Roto Node is often used to create garbage and holdout mattes.
Display the RGB image in an Image Viewer. Use the eyedropper tool in the Reference Color parameter in the Ultimatte Node Panel to pick a pixel in the backing area of the image. Ultimatte Intelligence will use the color values as a reference for setting color logic and generating matte data. See How Ultimatte Generates Matte Data for more information.
If the backing is a homogenous field of color (blue, green, or red), you can select any pixel in the backing that is totally unobstructed by shadows. If it is not, use an Ultimatte CSC node to correct the backing areas before using Ultimatte. Refer to the introduction, Overview of the Ultimatte Nodes, for more information about when and in what order to use each Ultimatte node.
Next switch from RGB to Alpha channel display in the Image Viewer to view the matte channel. This display represents the alpha channel that the node will output. Any holes in the matte will appear as gray or black areas within white foreground objects.
Fix any matte density problems caused by spill to prevent print-through using the Matte Controls.
After you are satisfied with the matte density values, you can use the Flare Controls to correct discoloration, especially of magentas, yellows, and browns, that can be caused by spill and flare. Compare the composited image with the original foreground image to identify these problem areas.
For the best result, be sure to refer to the Matte Controls and Flare Controls descriptions. Adjusting any of these parameters too dramatically can adversely affect the realism of your composite. |
Use the Reference Color eyedropper to click on the backing area of the image. This sets the reference value for the backing color that is used in the Ultimatte node operation.
Use this menu to specify which channels the Ultimatte node will output:
Check the Invert Matte box to invert the values of the outgoing alpha channel, so that transparent (black) areas become opaque (white) and vice versa.
The Matte Controls are used to adjust matte density levels.
The Matte Density and Black Gloss parameters enable you to adjust matte density values to eliminate print-through, which occurs when pixels in a foreground object are interpreted as being part of the backing.
Both parameters function by minimizing patches of gray or black within the white foreground objects of the matte. If you set either parameter too high, however, the result may include hard, dark edges around foreground objects and darkening of shadows and backing areas.
Adjust Matte Density and Black Gloss interdependently: increasing Black Gloss may allow Matte Density to be set lower. Readjust both parameters to be as low as possible while still ameliorating the problem.
The Black Gloss 2 and RGB Density parameters, which operate on the RGB channels rather than the matte, may help to minimize edge problems caused by Matte Density and Black Gloss. |
Matte Density To control print-through in bright foreground objects, use Matte Density. This situation may be caused by flare or spill, or when a foreground color is too close to the backing color (a blue-eyed actor against a bluescreen, e.g.). The Matte Density default value is 50.
Black Gloss To control print-through in black glossy or dark foreground objects, use Black Gloss. This situation can be caused when a black glossy object is reflecting color from the backing and is interpreted as being a dark area of the backing. The Black Gloss default value is 0.
The Black Gloss 2 parameter enables you to stop print-through on foreground objects that suffer from excessive spillage from the backing. Unlike Black Gloss, this parameter adjusts RGB values instead of the matte, and setting this parameter too high may alter the color of objects in the foreground. The default value is 0.
The RGB Density parameters enable you to adjust RGB values to reduce hard, dark edges from objects with strong red (skin tones), green (flora), or blue components. Unlike Matte Density, this parameter adjusts the RGB channels instead of the matte.
If you set one of the RGB Density values too low, the result may be print-through in foreground objects of the respective color. (For example, a low blue value may result in print-through in blue foreground objects.) In that case, the Matte Density control may need to be readjusted.
You can use the master control to modify all three channels equally, but in most cases you will probably want to expand it to modify the individual color channels separately. The default value for all channels is 100 (full density).
The Clean Up and Clean Up Balance parameters are interdependent, with Clean Up Balance controlling how Clean Up affects the foreground relative to the screen area. In general, the Clean Up value should be as low as possible.
The Clean Up parameter enables you to eliminate fine imperfections in the backing. However, it can also eliminate subtle details of the foreground, such as strands of hair, smoke, shadows, or reflections, and ruin the realism of the composite. The Clean Up default value is 0.
Use the Clean Up Balance parameter to assign the degree of influence that the Clean Up parameter has on the foreground relative to the backing area of the image. The default value is 50, which specifies that the Clean Up parameter will affect the foreground and backing equally. Try adjusting this parameter when the Clean Up parameter darkens edges or makes them glow.
The Level Balance parameter will help you even out imperfections or textures in the backing that are noticeable in the composite.
Increasing this parameter will not brighten the background, but it will increase the level in the shadows and transparent areas of the composited image. On occasion, lowering this parameter will result in an enhancement to the fine edge detail in foreground objects. The default value is 50.
The Shadow Noise parameter enables you to reduce noise in shadows and glare areas. The default value is 50.
If there are slight variations in color in the backing area, a residue of the backing color may remain in some areas when Ultimatte suppresses the backing color to black. This veil of color can discolor the background image in the composite.
The RGB Veil parameters are used to adjust the RGB color components in the backing area to suppress this residue. Decreasing these values will remove veiling from the backing area, but may cause dark edges around foreground objects. Increasing the values will add a colorized tint or haze over the background.
You can use the master control, which has a default value of 50, to modify all three channels equally, or expand it to access the individual channel controls.
If possible, you should use the Ultimatte CSC node to even out the backing area before the image is input to the Ultimatte node. |
The Flare Controls are used to correct the discoloration of foreground objects caused by spills and flares. During shooting, light reflecting from the backing may "spill" onto the foreground subject, causing discoloration and print-through. This color contamination of the foreground can also occur from reflections within the camera lens, or lens flare.
To correct the matte density problems (print-through) caused by spills and flares, use the Matte Controls first. Then use the Flare Controls to correct the discoloration. |
Ultimatte spill/flare suppression works by analyzing the color components of each foreground pixel. Using a blue backing as an example (this suppression works similarly for green and red screens), for most colors in nature, if the blue component exceeds green, it does so by less than green exceeds red. This is true even for many natural pastel shades of blue. The exceptions are deep blues, which have already been suppressed as part of the backing, and magentas.
Ultimatte assumes, therefore, that any time the blue component in a color exceeds green by more than green exceeds red, it is because of blue spill contaminating the color. Ultimatte adjusts the values so that blue only exceeds green by the amount green exceeds red (or by holding blue to the level of green if green is equal to or less than red), any unnatural bluish tints will be removed while natural blues will be unaffected.
As mentioned before, Ultimatte may interpret foreground objects with a magenta tint as being contaminated by blue spill and change shades of magenta to shades of red. In this case, you will need to adjust the Gate controls to compensate.
Another problem may arise with some shades of yellow and brown that are contaminated by blue spill. The blue component should be less than green, but Ultimatte suppression will allow blue to equal green, slightly desaturating these colors. In this case, you will need to adjust the Black or Gray Balance controls.
The Balance parameters can be used to make foreground colors warmer or cooler by overriding the spill/flare suppression logic that ensures that whites in the foreground have equal amounts of red, green, and blue:
The default value is 50 for all three parameters. Set the parameter by eye while viewing the composited image.
White Balance: Use the White Balance parameter to adjust light foreground colors with minimal effect on darker colors. It can make foreground whites match the cool/warm tint of background whites.
For example, flare suppression logic can cause blond hair, shot against blue backing, to look white around the edges. Adjusting White Balance can restore a warmer, more natural color to the hair without significantly affecting other colors.
Black Balance: Use the Black Balance parameter to adjust dark foreground colors with minimal effect on lighter colors. It can make foreground blacks match the cool/warm tint of background blacks.
For example, adjusting Black Balance can eliminate flare from black and brown hair (and some blond hair) with minimal effect on light colors. You can also compensate for the slight desaturation of some colors that the suppression logic can cause.
Gray Balance: Use the Gray Balance parameter to adjust midrange foreground colors with minimal effect on lighter and darker colors. It can make foreground grays match the cool/warm tint of background grays.
For example, used in conjunction with Black Balance, Gray Balance can help compensate for the desaturation of some colors.
The Gate parameters, Gate 1 & 3 and Gate 2, enable you to adjust specific color ranges, depending on the backing color used to shoot the sequence: blue, green, or red. (The "gate" terminology, by the way, is a holdover from the original Ultimatte hardware, which included physical gate switches to control this function.)
Gate 1 & 3: The following table illustrates the spill suppression effect of the Gate 1&3 parameter on various color families:
Using blue backing as an example, compare the original foreground image and the composite. Look to see if any cyan objects have turned blue, or green objects cyan. If so, you can probably use this parameter to remove the excess blue. The default value is 100.
Gate 2: The following table illustrates the spill suppression effect of the Gate 2 parameter on various color families:
Again using blue backing as an example, compare the original foreground image and the composite. Look to see if any magenta or pink objects have turned red. If so, you can probably use the Gate 2 parameter to restore the true blue component.
The default value for Gate 2 depends on the backing color:
This menu becomes active when an optional garbage matte input is connected to the Ultimatte node. It is used to specify which channel of the input to use as the garbage matte, when the image has more than one channel. For more information, see also Using a Garbage Matte.
This menu becomes active when an optional holdout matte input is connected to the Ultimatte node. It is used to specify which channel of the input to use as the holdout matte, when the image has more than one channel. For more information, see also Using a Holdout Matte.
The Ultimatte CSC (Classic Screen Correction) node enables you to correct flaws in the screen area, or backing, of a foreground image. Flaws can be caused by uneven lighting or seams and patches on the surface of the backing. You can use Ultimatte CSC to transform imperfect bluescreen backings into homogenous fields of blue (or green or red) while retaining shadows and transparency.
You should process imagery with the Ultimatte CSC node before further processing with the Ultimatte GK, Ultimatte, or Ultimatte CC nodes.
The Ultimatte CSC node requires two inputs:
Fig. 15.15 The foreground image is connected to the top input of Ultimatte CSC and the clean plate is connected to the bottom input.
The screen correction image, commonly referred to as a clean plate, is a clip of the lit backing that has been shot in the absence of foreground objects, such as actors or props. The clean plate is used as the reference for the correction. This input may be a single frame, for locked-off shots, or a sequence of frames that matches the number of frames in the foreground image input.
For a locked-off shot (where the camera doesn't move, pan, or zoom during the scene), a single reference frame is adequate. The Ultimatte CSC node will use the single reference frame for comparison with each frame of the foreground image.
In a scene where the camera does move, motion control equipment should be used to shoot a correction clip that matches the actual shot exactly. The Ultimatte CSC node will compare each frame of the foreground image to the corresponding reference frame.
Essentially, to create a corrected foreground image for output, the Ultimatte CSC node must identify, for each pixel in the original foreground image, whether the pixel represents a totally opaque foreground object, a partially transparent foreground object (such as a shadow), or unobstructed backing. In the corrected image
To accomplish this screen correction, the Ultimatte CSC node generates temporary mattes for both the foreground image and the reference frame using the process explained in How Ultimatte Generates Matte Data. Then the node generates a correction frame by subtracting the pixel values of the reference frame from the peak point value.
The value of each pixel in the foreground matte is divided by the corresponding pixel value of the reference matte to generate a correction matte.
Each pixel of the correction frame is multiplied by the corresponding pixel of the correction matte. The result is added to the corresponding pixel of the foreground image to create a new, corrected foreground image.
The first step to using Ultimatte CSC is to select a sample color Ultimatte will use as a reference in the screen correction process.
The sample color should be in an unobstructed area of backing (no shadows or strands of hair, e.g.) and should be a representative backing color. The best blue (or green, or red) to select from an uneven backing will depend on the image:
It is usually best to select a bright, representative value that is close to important foreground information, such as near a person's face. Be sure to avoid clicking on fine foreground details, such as smoke, shadows, strands of hair, or mist. Zoom in if necessary before picking.
Fill a swatch in the Image Viewer's Color Picker with the sample color and use it to fill the Sample Color parameters in any other Ultimatte nodes you may use to process the image further. |
The Ultimatte CSC Node Panel parameters are used to select the backing color and adjust the screen correction.
The Sample Color parameter displays a swatch of the sample pixel you select by clicking on an area of backing in the RGB display of the Image Viewer. You can also expand the parameter to access the full color selection controls, which enable you to set RGB or HSV values numerically. (For a detailed description of these controls, refer to Using the Color Parameters in chapter 14.)
See Selecting a Sample Color in the previous section of this node description for information about how to choose the best color.
The Matte Density and Black Gloss parameters can be used, if necessary, to adjust how Ultimatte interprets what is foreground and what is backing during the screen correction process:
If you set the Matte Density or Black Gloss parameters too high, the result may include enhanced visual noise or other unwanted effects. Be sure to check any adjustments you make to these parameters by viewing the result in a composite node.
The Ultimatte GK (Grain Killer) node removes film grain from the screen area, or backing, of a foreground clip. It is also effective at reducing visual noise such as that introduced by video cameras under low-light conditions.
One of the greatest strengths of Ultimatte is its ability to retain fine details in a composite. However, this means that film grain (or video noise) in the backing area of the foreground image may be composited onto the background and result in a double-grain effect in those areas.
The Ultimatte GK node is designed to eliminate this problem by recognizing and filtering only the backing area of the foreground image. This enables you to avoid softening the foreground subject and eliminating detail. See How Ultimatte Generates Matte Data for an explanation of the process Ultimatte uses to recognize backing areas of a foreground image.
Ultimatte GK filters grain pixels in the screen area by averaging the values of surrounding pixels to create replacement values for the pixel currently being processed.
The Ultimatte GK node accepts one input.
The first step is to specify a sample color of the backing area that represents the ideal color for Ultimatte Intelligence to use as a reference. Then you can adjust the default grain filtering parameter values if necessary.
If the input to the GK node is an Ultimatte CSC node--and this is highly advised, unless you have a foreground image with perfectly even backing--use the same sample color value in the GK node that you used for the Ultimatte CSC node.
If your shot does not require the use of the CSC node, use the eyedropper tool associated with the Sample Color parameter to click an appropriate pixel in the Viewer. See Selecting a Sample Color in the Ultimatte CSC node description for guidelines on selecting the best color.
Select Filter Area from the Source menu of the Image Viewer to examine the Filter Area matte. This display does not represent a normal image channel to be output from the node. Instead, it shows which parts of the image will be grain filtered.
Fig. 15.16 Select Filter Area display from the Viewer's Source menu to find out how the Ultimatte GK node will filter each pixel.
Areas defined as solid foreground subject are never filtered by the GK node. These areas appear green. To increase or decrease the subject area, adjust the Matte Density and Black Gloss parameters in the Advanced parameter group.
The backing area, as well as the transition areas between the subject and backing, can be filtered. These areas are black and white in the Filter Area display: white areas will be filtered and black areas will not. To increase the proportion of white pixels in the filter area, use the Screen Filter parameter in the Advanced parameter group.
The Sample Color parameter displays a swatch of the sample pixel, or pixel area, you select in an area of backing in the RGB display of the Image Viewer. To select the color, click the eyedropper tool associated with the Sample Color parameter to click on an appropriate pixel in the image.
You can also expand the parameter to access the full color selection controls, which enable you to set RGB or HSV values numerically. For a detailed description of these controls, refer to Using the Color Parameters in chapter 14.
See also Using Grain Killer in this node description for general guidelines.
Check this box to eliminate stray single pixels, or "orphans," from the screen area. These orphans are pixels that failed to be filtered by the node.
The Advanced parameter group includes the following parameters, which are best used in conjunction with the Filter Area display (see Fig. 15.16).
The Filter Passes parameter enables you to specify the number of times the image data will pass through the filter. The default value is 1; however, multiple passes may be necessary. If the backing still appears mottled after the initial pass, try increasing the setting.
The filtering process used by the node averages the color values of surrounding pixels to create replacement values for the grain pixels being filtered. Pixels are averaged 25 at a time, in a 5 x 5 matrix, as long as the matrix does not encroach on transition or foreground areas. For pixels too close to a transition edge for an unobstructed 5 x 5 square to be defined, 9 pixels (in a 3 x 3 matrix) are averaged.
The Matte Density and Black Gloss parameters enable you to control what portion of the image is defined as solid subject area. (The subject area is green in the Filter Area display in the Image Viewer.) This is important because the areas defined as totally opaque foreground objects will not be grain-filtered.
For example, if you boost the density values in GK Matte to eliminate holes in the matte, the portion of the image that is defined as solid subject, and will therefore not be filtered, is increased accordingly:
Matte Density defaults to 50, while Black Gloss defaults to 0. Both parameter ranges are 0-100.
As the Ultimatte GK node does not generate an output matte (the Ultimatte node is used for that purpose), you can define transition areas as part of the solid subject area if you do not want them to be grain filtered. On the other hand, you can erode the subject area to ensure filtering of transition areas. It all depends on the nature of the image.
The Screen Filter parameter controls how much of the area defined as backing--the black and white pixels in the Filter Area display--will be grain filtered. White areas in the Filter Area display will be filtered; black areas will not.
If too much of the backing area is filtered, you will begin to filter the edges of the foreground subject, resulting in a loss of detail. To reduce the area of backing that will be filtered (to turn some of the white pixels black), start by expanding the Screen Filter parameter to access the individual color controls.
Decrease the Value parameter, and watch the change to the Filter Area display, which will update accordingly. The white pixels in the transition areas (next to the green subject areas) will turn black first. If you continue to decrease the Value parameter, more pixels in the rest of the backing will turn black.
To ensure that these values are set properly for your shot, be sure to review the result (especially in the final composite image) and judge for yourself.
The Ultimatte CC (Color Control) node enables you to match the color levels of an image to a reference image. Typically, the Ultimatte CC node is used to adjust the output from an Ultimatte node to match the background image over which it will be composited.
One example of how Ultimatte CC is used is the stereotypical "twins" scene, in which a single actor plays dual roles. In such a case, you may be faced with two clips: one of the actor against a background, and one of the actor turned in the opposite direction, speaking twin to self, filmed in front of a bluescreen.
It is likely that each clip was shot under different lighting conditions, and you may therefore need to adjust the flesh tones or clothing colors of one image to match the other. The Ultimatte CC node will enable you to make the colors of two such images compatible.
The Ultimatte CC accepts one input--it is not necessary to connect the reference image to the CC node.
Ultimatte CC is normally used to match the color of the input image to a reference image. However, it can be used to adjust color attributes of an image without using a reference image. To do so, skip step 1 and start with step 2.
The Color Control Node Panel includes parameters for a Sample Color value, which should be picked from the input image to be adjusted, and for a Reference Color value, which should be picked from the image you want to match.
Select representative areas from each image:
Then use the parameters in the Controls group to correct the output image:
For the best result, fine-tune your color adjustments based on viewing the final composited image, rather than the foreground image alone. |
Both parameters work the same way: use the eyedropper tool to sample the color values in a representative area of an image, which will fill the parameter swatch with that color:
Expand the Controls group to access the Level, Gamma, Black, and Saturation parameters. These parameters are master controls that adjust all three color channels equally. However, each of them can be expanded to reveal separate Red, Green, and Blue channel parameters if you need to adjust a channel individually. The range of all parameters in the Controls group is 0-100.
The Level parameter is used to adjust brightness in the highlights; pure blacks are unaffected. This parameter is effectively a gain control on the color components because it sets a multiplier value for each pixel value. The default value is 80.
The Gamma parameter is used to adjust contrast in the midrange tones; pure blacks and whites are unaffected:
When you increase the dynamic range of color values at one end of the range, colors at the opposite end are correspondingly compressed. The default value for the Gamma parameter is 50.
The Black parameter is used to adjust black levels; pure whites are unaffected. If the Black parameter is set below 50 (which is the default value), the darkest color components in the image will clip at 0.
The Saturation parameter is used to control saturation by proportionally replacing each color component with its luminance value. If you set the value to zero, the result will be a grayscale image. The default value is 80.