The Luma Keyer node uses the overall luminance of an image to create an Alpha channel. The label of this node may seem misleading since it allows pulling mattes from almost any channel. In some respects, it is more accurate to call this node an all-purpose channel keyer, but its primary purpose is for extracting alpha channels based on luminance
Luma Keyer Node Inputs
The Luma Keyer node includes four inputs in the Node Editor.
- Input: The orange input accepts a 2D image that contains the luminance values you want to be keyed for transparency.
- Garbage Matte: The gray garbage matte input accepts a mask shape created by polylines, basic primitive shapes, paint strokes, or bitmaps masks. Connecting a mask to this input causes areas of the image that fall within the matte to be made transparent. The garbage matte is applied directly to the alpha channel of the image.
- Solid Matte: The white solid matte input accepts a mask shape created by polylines, basic primitive shapes, paint strokes, or bitmaps masks. Connecting a mask to this input causes areas of the image that fall within the matte to be fully opaque.
- Effect Mask: The optional blue input expects a mask shape created by polylines, basic primitive shapes, paint strokes, or bitmaps masks. Connecting a mask to this input limits the pixels where the luminance key occurs. An effects mask is applied to the tool after the tool is processed.
Luma Keyer Node Setup
The Luma Keyer takes an input from an image with significant luminance difference to extract a key. You can then use the output of the Luma Keyer into any mask input.
Luma Keyer Node Controls Tab
The Controls tab in the Luma Keyer contains all the parameters for adjusting the quality of the matte.
This menu selects the color channel used for creating the matte. Select from the Red, Green, Blue, Alpha, Hue, Luminance, Saturation, and Depth (Z-buffer) channels.
This range slider sets the lower threshold using the handle on the left and sets the upper threshold using the handle on the right. Adjusting them defines a range of luminance values to create a matte.
A value below the lower threshold setting becomes black or transparent in the matte.
Any value above the upper threshold setting becomes white or opaque in the matte.
The values within the range create a grayscale matte.
This control selects the filtering algorithm used when applying a blur to the matte.
- Box: This is the fastest method but at reduced quality. Box is best suited for minimal amounts of blur.
- Bartlett: Otherwise known as a Pyramid filter, Bartlett makes a good compromise between speed and quality.
- Multi-box: When selecting this filter, the Num Passes slider appears and lets you control the quality. At 1 and 2 passes, results are identical to Box and Bartlett, respectively. At 4 passes and above, results are usually as good as Gaussian, in less time and with no edge “ringing.”
- Gaussian: The Gaussian filter uses a true Gaussian approximation and gives excellent results, but it is a little slower than the other filters. In some cases, it can produce an extremely slight edge “ringing” on floating-point pixels.
This blurs the edge of the matte using the method selected in the Filter menu. A value of zero results in a sharp, cutout-like hard edge. The higher the value, the more blur.
This option determines how edges are handled when performing domain of definition rendering. This is profoundly important when blurring the matte, which may require samples from portions of the image outside the current domain.
- Frame: The default option is Frame, which automatically sets the node’s domain of definition to use the full frame of the image, effectively ignoring the current domain of definition. If the upstream DoD is smaller than the frame, the remaining area in the frame is treated as black/transparent.
- Domain: Setting this option to Domain respects the upstream domain of definition when applying the node’s effect. This can have adverse clipping effects in situations where the node employs a large filter.
- None: Setting this option to None does not perform any source image clipping at all. This means that any data required to process the node’s effect that is usually outside the upstream DoD is treated as black/transparent.
This slider shrinks or grows the semitransparent areas of the matte. Values above 0.0 expand the matte, while values below 0.0 contract it.
This control is usually used in conjunction with the blur to take the hard edge of a matte and reduce fringing. Since this control affects only semitransparent areas, it has no effect on a matte’s hard edge.
Matte Gamma raises or lowers the values of the matte in the semitransparent areas. Higher values cause the gray areas to be more opaque, and lower values cause the gray areas to be more transparent. Wholly black or white regions of the matte remain unaffected.
Selecting this checkbox inverts the matte, causing all transparent areas to be opaque and all opaque areas to be transparent.
Solid mattes are mask nodes or images connected to the solid matte input on the node. The solid matte is applied directly to the alpha channel of the image. Generally, solid mattes are used to hold out keying in areas you want to remain opaque, such as someone with blue eyes against a blue screen.
Enabling Invert inverts the solid matte before it is combined with the source alpha.
Garbage mattes are mask nodes or images connected to the garbage matte input on the node. The garbage matte is applied directly to the alpha channel of the image. Generally, garbage mattes are used to remove unwanted elements that cannot be keyed, such as microphones and booms. They are also used to fill in areas that contain the color being keyed but that you wish to maintain.
Garbage mattes of different modes cannot be mixed within a single tool. A Matte Control node is often used after a Keyer node to add a garbage matte with the opposite effect of the matte applied to the keyer.
Enabling Invert inverts the garbage matte before it is combined with the source alpha.
Select this option to cause the keyer to multiply the color channels of the image against the alpha channel it creates for the image. This option is usually enabled and is on by default.
Deselect this checkbox and the image can no longer be considered premultiplied for purposes of merging it with other images. Use the Subtractive option of the Merge node instead of the Additive option.
For more information on these Merge node settings.
Luma Keyer Node Settings Tab
The Settings tab in the Inspector is also duplicated in other Matte nodes. These common controls are described in detail HERE.
About the Author
Justin Robinson is a Certified DaVinci Resolve, Fusion & Fairlight instructor who is known for simplifying concepts and techniques for anyone looking to learn any aspect of the video post-production workflow. Justin is the founder of JayAreTV, a training and premade asset website offering affordable and accessible video post-production education. You can follow Justin on Twitter at @JayAreTV YouTube at JayAreTV or Facebook at MrJayAreTV