Glossary

Writing a glossary for a rapidly evolving field with “unstable terminology” like digital compositing is a tricky but necessary task. A task that is further complicated by an infusion of terms from many other disciplines such as conventional opticals, special effects, photography, computer science, and signal processing. However, it needs be done, so it is done here with the intent of being more helpful to the digital artist than the engineer. Many terms are very technical in nature, but the unabashedly technical definition is often less than helpful to someone trying to make a good composite than a more common sense, task-related definition, so the common sense style prevails in this glossary.

Some of the words in the definitions are highlighted in bold to indicate that those words are also defined in this glossary.

Numerical

2D transformation – moving an image only in two dimensions, such as a translation, scale, or a rotation.

2k – literally 2,000, but in computerland it is 2,048 (211).

2:3 pull-down (or pulldown) – another term for 3:2 pull-down.

2:3 pull-up (or pullup) – another term for 3:2 pull-up.

3-perf – slang for a 35mm film format where each exposed frame on the film spans three perfs (perforations). Saves film and lab costs.

3D animation (3D) – animation created in a computer by creating three-dimensional models of objects, then lighting and animating them.

3D compositing – to incorporate 3D geometry, lights, cameras, and other 3D functions into a 2D compositing program.

3D coordinates – the X, Y, and Z coordinate system used to define the location of a point in three-dimensional space.

3D geometry – a three-dimensional object defined by polygons used in 3D animation. Can include points and particle systems.

3:2 pull-down (or pulldown) – a timing scheme to stretch the 24 frame per second rate of film to the 30 frame per second rate of video. Groups of 4 film frames are spread over 5 video frames by distributing the film frames over 10 video fields in the pattern 2, 3, 2, 3. Also referred to as a 2:3 pulldown.

3:2 pull-up (or pullup) – the process of removing the 3:2 pull-down from film transferred to video in order to restore the frames to their original 24 frame per second rate. Also referred to as a 2:3 pullup.

4:1:1 – video digitized with 4 samples of luminance and 1 sample of chrominance every 4 pixels. When converted to an RGB image the luminance will be full resolution but the colors will be 1/4 resolution.

4:2:0 – video digitized with 4 samples of luminance and 1 sample of chrominance every 4 pixels. When converted to an RGB image the luminance will be full resolution but the colors will be 1/4 resolution like 4:1:1 but with a different sampling pattern.

4:2:2 – video digitized with 4 samples of luminance and 2 samples of chrominance every 4 pixels. When converted to an RGB image the luminance will be full resolution but the colors will be half resolution.

4:4:4 – video digitized with 4 samples of luminance and 4 samples of chrominance every 4 pixels. When converted to an RGB image both the luminance and colors will be full resolution.

4k – literally 4,000, but in computerland it is 4,096 (212).

4-perf – slang for any number of 35mm film formats where each exposed frame on the film spans 4 perfs (perforations).

5-perf – slang for the 65mm/70mm film format where one frame spans 5 perfs (perforations) of film.

65mm/70mm – a large format film stock with a 65mm-wide camera negative and a 70mm-wide projection print with an aspect ratio of 2.2.

8-bit Images – images that use 8 bits of data per channel, which results in a data range between 0 and 255 for each channel.

8-perf – slang for the VistaVision film format where each exposed frame sits sideways on the film and spans 8 perfs (perforations).

10-bit Images – images that use 10 bits of data per channel, which results in a data range between 0 and 1023.

10-bit log – representing log data with 10 bits of precision.

A

Absolute difference – the difference between two values where any negative values are converted to a positive number.

Academy aperture – the rectangular area of a frame of film for cinema projection that leaves room for a sound track between the left edge of the frame and the perfs.

Academy center – the center of the academy aperture. Since the academy aperture is off-center on the film frame, so is the academy center.

Accommodation – in the human visual system the act of focusing the eyes on a target.

ACES (Academy Color Encoding System) Color Management – a color management system that converts all image data to a large gamut linear light space where the capturing camera influences are removed so that it represents the original scene illumination.

Add operation – the processing of summing two images together.

Add–mix composite – to use color curves to make two versions of a matte prior to composite, one used to scale the foreground and the other to scale the background. Allows fine control over the look of semi–transparent regions.

Algorithm – a step-by-step procedure for solving a problem or achieving an objective.

Aliasing – the pixilated, jagged lines introduced into a computer image by failing to compute an image with subpixel accuracy. Results in edge pixels that are either on or off instead of smoothly blended.

Alphasee alpha channel.

Alpha channel – the “matte” channel of a four-channel CGI image. See also key, mask, and matte.

Alpha compositing operations – image-blending operations such as Over, Under, and XOR that require the presences of an alpha channel.

Ambient lighting – an omni-directional low-level soft illumination effect that uniformly lights a 3D object.

Ambient occlusionsee Occlusion pass.

Ambient shader – a shader that renders a low-level soft illumination effect that uniformly lights a 3D object.

Anaglyph – a simple but dated stereo display technology using red- and cyan-colored gasses.

Analog – a signal that is continuously variable without discrete “steps” or values.

Anamorphic – to scale an image along one axis only. Cinemascope frames are an anamorphic format squeezed horizontally on the film.

Animate – to change over time.

Anti-aliasing – the mathematical operation that eliminates aliasing in CGI images to give objects smooth edges. Multiple subpixel samples are made for each pixel to calculate a more accurate color and/or transparency value, but also dramatically increases computing time.

AOV – Arbitrary Output Variable, a CGI render pass that contains data about the image, not the image itself. Examples are depth Z, motion UV, and normals.

Artifacts – unintended side effects of an operation, usually bad.

Aspect ratio – a number that describes the rectangular “shape” of an image (how rectangular or square it is) as the ratio of the width divided by the height. May be expressed as a proportion (4:3), a ratio (4/3), or a floating-point number (1.33).

Atmospheric haze – the attenuation and discoloring of distant objects in a scene due to the color and density of particles (haze) in the atmosphere. Applies to interior scenes as well.

Axis – a straight line about which an object may be rotated or scaled. Also the horizontal and vertical reference lines in a chart or graph that are marked with the units of the chart. See also null object.

B

B-spline – short for “Basis Spline” where the resulting curved line does not pass through the control points. Commonly used for defining mechanical shapes.

Background – the rear-most layer in a composite.

Backing color – the uniform solid color used as the backdrop in a bluescreen or green-screen shot.

Banding – an image artifact that appears as a flat region of the same color, like a contour map that is cause by too few data value steps for the display device. Instead of a smooth gradient, a series of slightly different colored bands would be seen. Also known as contouring.

Base layer – the bottom layer in a two-layer Adobe Photoshop blending operation.

Bayer array – a imaging array that has a colored Bayer filter over it consisting of an alternating pattern of two green, one red and one blue filter.

Beauty pass – a full color render of a 3D object with color, texture maps, and lighting, but excluding other passes such as specular highlights, reflections, and shadows.

Bezier spline – named for French engineer Pierre Bézier where the resulting curved line passes through the control points. Commonly used for defining organic shapes surfaces.

BGsee background.

Bicubic filter – performs edge sharpening as it resamples an image. Can introduce edge artifacts.

Bilinear filter – simply averages pixel values without edge sharpening as it resamples an image. Results look softer than a bicubic filter.

Binarize – to separate an image into just two values, usually zero and one (black and white), about some threshold value.

Bit – the smallest unit of digital data, its value is either zero or one.

Bit depth – the number of bits used to represent one channel of an image (or other data). A bit depth of 8 can represent 256 brightness levels, and increasing the bit depth increases the number of brightness levels that can be represented.

Blend layer – the top layer in a two-layer Adobe Photoshop blending operation.

Bluescreen – 1) General: using a uniform solid primary color as the background so that a foreground object can be isolated by extracting a matte. Usually blue or green, but can be red. 2) Specific: a shot that uses blue as the backing color for extracting a matte.

Bounce light – light that bounces between multiple surfaces that imparts both its brightness and color to nearby surfaces.

Brightness – the subjective human perception of luminance.

Bump matte – creating a matte from just the “bumps” of a textured surface, such as the rough bark of a tree.

C

Cadence – refers to the five possible 3:2 pull-down timings that can be encountered when film is transferred to video.

Camera aperture – the area of the film frame that is exposed by the camera. See also projection aperture.

Camera projection – to create a 3D scene by projecting an image with a 3D-camera onto 3D geometry and rendering the results with a second 3D-camera.

Camera tracking – using sophisticated software to analyze a clip that calculates the camera position and lens on a frame-by-frame basis. Used to render 3D elements to track correctly with the original clip.

CCD – abbreviation for Charged Coupled Device, a solid-state electronic device that converts light into electricity for digitizing images. With the CCD cells arranged in a row it is called a linear array, and arranged into a rectangular grid it is called a 2-dimensional array.

CGIsee Computer Generated Image.

Channel – for digitized images, one of image planes that make up the color image or a matte, such as the red channel or the matte channel. See also color channel.

Channel arithmetic – mathematical operations between the color channels of an image, such as adding 10% of the green channel to the red channel.

Chroma key – a matte that is created based on the color (chrominance) of an element, or the use of such a matte to isolate an object from its background for compositing or other treatment.

Chromatic aberration – a defect in photography lenses that causes each color to focus at a slightly different distance. Results in colored rings/fringes around objects.

Chrominance (chroma) – the color portion of a video signal that is separated into chrominance (color) and luminance (brightness).

Cinemascope – a 35mm film format characterized by an image that is anamorphically squeezed horizontally by 50% on film. When projected with an anamorphic lens the width is doubled to create a 2.35 aspect ratio image. See anamorphic.

Cineon – the 10-bit log image file format developed by Kodak for digitizing feature film.

Cineon File Format – a 10-bit log file format for film scans developed by Kodak.

Circle of confusion – an optical spot where an image starts to perceptively lose focus due to the lens’ depth of field.

Clean plate – a second version of an image with the item of interest gone, such as a bluescreen plate without the foreground object.

Clipping – to set any data value above an established threshold to be equal to that threshold. With a threshold value of 200, a pixel value of 225 would be clipped to 200.

Clipping plane – a flat plane that bisects a three-dimensional space and deletes (clips out) all objects on one side.

Color channel – one of the three color components of an RGB image, such as the red channel.

Color correction – altering an image to match another image by adjusting it separately in three brightness zones – the shadows, midtones, and highlights.

Color curve – remapping (altering) the color values of an image based on drawing a graph of the relationship between the input and output values.

Color difference key – creating a key by subtracting color channels from each other.

Color difference keyer – a bluescreen matte extraction technique that uses the difference in value between the primary backing color (the blue) and the other two color records to create a matte.

Color grading – applying an initial color grade to a raw image to correct exposure or white balance.

Color resolution – the total number of colors that can be produced with the available number of bits. An 8-bit image has a color resolution of 256 colors while a 24-bit image has 16.7 million.

Color space – the selection of appropriate color attributes plus their units of measure used to describe a color. For example, the attributes could be hue, saturation, and value for HSV colorspace, or red, green, and blue for RGB colorspace.

Color timing – the process of color-correcting each shot of a feature film. Originally done chemically in a film lab, it is now being done with computers in the Digital Intermediate process.

Composite – the process of combining two or more image layers together with the use of a matte to define the transparencies of the layers.

Computational photography – to create an image from an array of data rather than to directly image a scene. An example is light field cinematography.

Computer Generated Image (CGI) – an image created totally in the computer such as 3D animation or a digital matte painting, as opposed to live action images imported into the computer then manipulated, such as composites and morphs.

Contact shadow – the very dense shadow created by the contact of an object to a surface, such as a foot standing on the ground.

Contouring – another term for banding.

Contrast – the degree of dissimilarity between the brightest regions of an image and the darkest.

Control points – the points on a spline that control its position and shape.

Convergence – the angle between each eye as they focus on an object. Close objects create a large convergence, while objects at infinity have parallel lines of sight.

Conversion – to convert a movie originally photographed “flat” using one camera into a stereo film by synthesizing the left and right views in post-production.

Convolution kernel – the two-dimensional array of numbers used to weight the pixel values in a convolve operation. Usually a square array of odd dimensions (3 × 3, 5 × 5, etc.), the kernel is centered over each pixel in an image and the surrounding pixels are scaled by the values in the kernel, then summed together to produce the output pixel value. Different convolution kernels give very different effects such as edge detection, blurring, or sharpening.

Cool – a general term to describe an image with a bluish cast. See also warm.

Core matte – a matte designed to clear stray pixels from the solid core (center) of a matte.

Corner pinning – a geometric transformation of an image where each corner of the image can be moved to an arbitrary location.

Correlation number – in motion tracking, the number the computer calculates to quantify how close the match is between the current frame’s match box and the initial reference image.

Correlation points – for a spline warp, the points located on the source and destination splines that controls how the pixels will be moved between them.

Crop – to make an image smaller by trimming the edges.

Cross-dissolvesee dissolve.

Cscope – short for Cinemascope.

D

Dailies – projecting film in a screening room to review new shots, or the film itself.

Darken operation – the Adobe Photoshop term for a “minimum” image blending operation.

Dashboard effect – the difficulty the eye has in adjusting to two adjacent scenes that have an abrupt change in depth.

Data compression – using any one of a number of schemes to reduce the size of the data required to represent an image.

DCI (Digital Cinema Initiative) – the film industry specifications for images to be projected in a theater digitally. The 2k format is 2048 × 1080 with a 1.89 aspect ratio.

DCP (Digital Cinema Package) – the packaged delivered to theaters for digitally projecting a movie that contains the movie, sound, and control data.

De-interlace – to use one of several schemes to convert video frames composed of two slightly differing interlaced fields into a single unified frame. See interlaced video.

Deep compositing – to composite with deep images.

Deep images – images that contain multiple samples of color, transparency and depth for each pixel.

Deformation lattice – a bounding box placed around a 3D object that is used to deform it.

Denoise – to remove the digital noise or film grain from an image.

Depth compositing – compositing images using their depth Z values to automatically sort their front-to-rear layering order.

Depth grading – to adjust the relative depth of the objects in a stereoscopic scene.

Depth of field – the inner and outer distance from the camera lens in which objects appear in focus.

Depth Z pass – a 3D rendering technique where a file is rendered containing pixel-by-pixel information of the distance from the lens to a 3D object. Used during compositing to add various effects such as depth-of-field blurring.

Despill – the removal of backing color spill contamination on a foreground object in a bluescreen shot.

DIsee Digital Intermediate.

Difference matte – the matte generated by taking the absolute value of the difference between two images, one with the item of interest present and an identical one without the item of interest.

Difference tracking – to track the difference in position between two moving objects in order to track one of them onto the other.

Diffraction glow – an imaging artifact where very bright objects have a glow around them due to the light waves diffracting around sharp edges.

Diffuse passsee beauty pass.

Diffuse shader – a shader that renders light and dark regions of a 3D object based on the surface’s angle to the light sources.

Digital cinema – to project a feature film using digital images and a digital projector rather than film and a 35mm projector.

Digital Intermediate (DI) – the process of color timing a feature film digitally.

Digitize – to convert a continuously varying analogue signal into discrete numbers. To quantize.

Dilate – to expand the outer perimeter of a white matte equally all around.

Discreet nodes – simple, one-function nodes such as “add” or “invert” that are combined to create more complex functions such as a screen operation.

Disparity – the pixel offset of an object between the left and right views of a stereo pair due to parallax.

Disparity map – a computed table of the motion vectors that indicate how far each object in a stereo pair has shifted between the left and right views.

Displacement map – an image that controls how the pixels of a second image are displaced (shifted in x and y).

Dissolve – a transition between two scenes using an optical effect where one image is faded down at the same time as the other is faded up.

Divergence – an unnatural and uncomfortable condition created in poorly designed stereo scenes where the eyes actually “toe out”.

Double Exposure – exposing the same frame of film twice with different images.

DPX file format – Digital Picture Exchange, a file format that can hold 10-bit log film scans as well as 10-bit linear Hi–Def video frames. A very flexible image file format, it can hold many types of images and supports extensible features.

DXsee double exposure.

Dynamic range – the ratio of the largest to the smallest brightness values in an image, display device or scene. A dynamic range of 100 would mean the brightest element was 100 times brighter than the darkest.

E

Edge blending – the process of lightly blurring the seam between two composited images in order to blend them more naturally.

Edge detection – a computer algorithm that finds and creates a matte of the edges found in an image.

Edge mask – a mask that demarks just the outlines of another mask.

Edge processing – computer algorithms that operate on the edge of a matte to either expand (dilate) or contract (erode) the perimeter.

Environment light – a 3D lighting technique that projects an image onto a surrounding sphere to add reflections of the environment to the surface of 3D objects.

Equirectilinear / Equirectangular – another name for a LatLong image used as a texture map around a sphere that is twice as wide as it is tall.

Erode – to shrink the outer perimeter of a white matte equally all around.

Exposure – the total amount of light over a period of time that falls onto a frame of film.

EXR File – a file format developed by ILM that features 16-bit floating-point high dynamic range image data without banding and multiple user-defined channels.

Extrapolate – to calculate a hypothetical new data point by projecting past the range of existing data points.

F

Fade – an optical effect where an image gradually transitions to black.

Falloff (or fall-off) – the change in density of a matte or key from its 100% solid core to its outermost black edge pixels.

FGsee foreground.

Field dominance – for interlaced video, whether field one or field two is displayed first.

Film recorder – a machine that exposes film using digital images rather than a live action scene.

Filter – to derive a new value for a pixel by evaluating it with its neighbors using one of several filtering algorithms.

Flare – the light from a bright light source that reflects and refracts through the camera lens and contaminates the image.

Flip – to swap the top and bottom of an image by mirroring it around its horizontal axis. See also flop.

Floating point – representing numbers with several degrees of decimal precision, such as 1.38725.

Flop – to swap the left and right sides of an image by mirroring it around its vertical axis. See also flip.

Flowgraph – the graphic representation of the sequence of operations in a compositing script which represents each image-processing operation by an icon with connecting lines to indicate their order of execution.

Focus pull – to change the focal point of the lens in order to re-focus on a second item of interest, such as from a person in the foreground to someone in the background.

Follow focus – to change the focal point of the lens in order to maintain focus on a moving target in a scene, such as someone walking away from the camera.

Footroom – the range of code values in the blacks of an image that encompass the variations in grain or noise.

Foreground – the element composited over a background. There is often more than one foreground element.

Fourier transforms – an image-processing operation that converts an image into spatial frequencies for analysis and/or manipulation.

Frame average – to average two or more frames together, as when doing speed changes.

Frustum – a camera’s rectangular viewing cone extending from the lens to infinity.

Full aperture – a 35mm film format that uses the entire frame of 4-perf film stock for the captured image.

G

G–matte – abbreviation for garbage matte.

Gain – the color correction operation that scales RGB values up or down centered on zero.

Gamma – a mathematical operation that calculates a value by raising the original value to some power. Used to characterize the response of a display device

Gamma correction – the opposing gamma operation applied to an image to compensate for the inherent gamma of the display device.

Gamma slamming – extreme increases and decreased in viewing gamma in order to reveal defects in keys and mattes.

Gamut – the full range of possible colors that may be created by an image display device or color space.

Garbage matte – a quick but imprecise matte that roughly isolates an item of interest from its background.

Gate weave – a horizontal wobble found in some 35mm film scans due to the film wobbling horizontally in the camera’s film gate.

Gaussian distribution – to perform a weighted average of a range of values using a distribution curve similar to the normal bell curve used by statistics.

Gaussian filter – uses a gaussian distribution of the pixel averages as it resamples an image. Does not perform edge sharpening.

Geometric deformations – to deform an image in x and y in non-linear ways, such as with mesh warps or image displacement.

Geometric primitives – simple 3D shapes such as sphere, cylinder, cube, and card.

Geometric transformation – an image-processing operation that changes the position, size, orientation, or shape of an image. Includes operations such as translate, scale, rotate, warp, and four-corner pinning.

Global Illumination – a CGI lighting model that simulates all of the bounce light in a scene.

Global shutter – a type of shutter on a digital camera that “dumps” all of the pixels simultaneously to avoid the “rolling shutter” effect.

Grain – the non-uniform “clumping” of silver halide particles and the resulting color dyes in film. The amount of grain differs between film stocks, levels of exposure, as well as between the three-color records of a film image.

Grayscale image – another term for a monochrome image.

Greenscreen – using green as the backing color for a bluescreen shot.

GUI – abbreviation for Graphical User Interface, the user interface to operate a computer program that uses graphics manipulated by a mouse or pad, as opposed to a command line interface where typed commands control the software.

H

Hard clip – to clip an image’s pixel values by simply limiting them to the clipping threshold value. Produces hard-edged flat spots in the resulting image. See also soft clip.

Hard light blending mode – an Adobe Photoshop blending mode that combines two images without a matte.

HDR images – High Dynamic Range images are floating point images containing image data in excess of 1.0 and are used for capturing scenes with extremes in brightness.

HDTV – abbreviation for High Definition Television, a collection of high-resolution video standards characterized by a 16:9 aspect ratio, different scan modes, various frame rates, and up to 1,080 scan lines.

Head-Mounted Display (HMD) – a display device worn on the head that is used to view virtual reality images.

HFR (High Frame Rate) – film frame rates greater than the standard 24 frames per second.

Hicon matte – short for “high contrast”, literally a black and white matte.

Highlight passsee specular pass.

Histogram – a statistical analysis of an image that plots the percentage of pixels at each level of brightness. Can be for each color channel or its overall luminance.

Histogram equalization – an image-processing operation that finds the darkest and lightest pixels, then adjusts the image contrast to fill the available data range.

HSL – a color metric for describing colors based on Hue, Saturation, and Lightness (brightness) rather than RGB values.

HSV – a color metric for describing colors based on Hue, Saturation, and Value (brightness) rather than RGB values.

Hue – the attribute of a color that describes what point on the color spectrum it represents, separate from its saturation or brightness.

I

ID (Identification) passes – masks rendered along with CGI that are used to isolate specific elements of the image for separate processing such as color correction.

IDT (Input Device Transform) – the start of the ACES color management pipeline where the camera characteristics are backed out of the captured images.

Image-based lighting – a CGI lighting technique that illuminates 3D objects in a scene using a photographed image that captured a 360-degree view of the original location or set.

Image displacement – to shift the position of the vertices of a 3D object based on the pixel values in an image. Can be used to build terrain and other 3D objects.

Image processing – using a computer to modify a digitized image.

IMAX – a 15-perf 70mm film format developed for special large screen theatres by the Imax corporation.

Impulse filter – a filter used to resample an image that simply selects the nearest appropriate pixel rather than calculating some kind of interpolated value. Very fast, but poor quality. Often used for quick motion tests.

Integer – a whole number without fractions or decimals.

Integer operations – math operations that use only integer numbers for input and output instead of floating-point numbers. Computationally much faster, but results are far less precise and can introduce artifacts.

Interactive lighting – the 3-dimensional lighting effects between light sources and objects, as well as from object to object within a scene. Usually refers to how the lighting on objects changes over time as they move within the shared light space.

Interaxial distance (IAD) – the distance between the optical axis (center) of the two lenses of a stereo camera rig.

Interlace flicker – the rapid flickering introduced into an interlaced video image due to a high contrast edge being precisely aligned with a video scan line.

Interlaced video – a video-scanning method that first scans the odd picture lines followed by the even picture lines, then merges them together to create a single video frame. See also progressive scan.

Interpolate – to calculate a hypothetical new data point between two existing data points.

Interpupillary distance (IPD) – the distance between the center of the pupils of human eyes, typically 68mm or 2.7 inches.

Inverse square law – the law of physics that describes the falloff of light with distance. The intensify of the light is reduced by the inverse of the square of the change in distance i.e. if the distance is doubled, the light intensity is ¼.

Invert – in a matte, to reverse the black and white regions. Mathematically, it is the complement (1–matte).

Iterate – to repeat or “loop” through some process.

J

Jaggies – slang for aliasing.

K

k – abbreviation for kilo, which is 1,000. In computerland it is 1,024 (210).

Kernel – short for convolution kernel.

Key – the video term for a matte. See also alpha, and mask.

Keyer – a compositing node that extracts the matte from a bluescreen, performs the despill, and produces the final composite. Usually a third-party “plug-in”.

Keylight – the name of a famous greenscreen and bluescreen keyer known for its excellent fine hair detail.

KeyMix – a compositing operation requiring three inputs – the foreground, the background, and a matte.

L

Lanczos – a filter used for resizing images that adds a bit of sharpening but avoids introducing too many sharpening artifacts.

LatLong image – an image that is twice as wide as it is tall so that it maps to a sphere 360 degrees around the equator and 180 degrees from pole to pole.

Lens flare – the colorful light pattern from a strong light source that becomes double exposed with the image due to light reflections and refractions within the lens elements.

Lift – a color-correcting operation that scales RGB values centered on 1.0 so that it raises the blacks most.

Light field cinematography – a new type of camera that simultaneously captures multiple angles of a scene using millions of microscopic “cameras” behind a single lens.

Light space – a general term for the three-dimensional light environment of a scene within which the objects in the scene are illuminated. Alludes to the location and intensity of light sources plus their secondary and even tertiary reflection and scattering effects.

Light wrap – to blend light from the background plate with the outer edges of a foreground element to enhance photorealism.

Lighten operation – the Adobe Photoshop term for a “maximum” image-blending operation.

Lighting pass – a 3D rendering technique where a 3D object is rendered multiple times with only one light turned on per pass. The lighting passes are then combined in compositing.

Linear data – an image-encoding scheme with a one-to-one relationship between the code values and the image brightness such that an increase in the code values causes a proportional increase in the image brightness.

Linear gradient – a gradient with pixel values that increase uniformly across the gradient.

Linear images – images that are represented by linear data as opposed to log data. Each step in the data represents a uniform change of actual brightness instead of a uniform change of apparent brightness to the eye.

Linear space – representing image color values in any one of several color spaces that use linear data.

Linearize – to remove the baked-in LUT of an image to convert it to linear data.

Locking point – in motion tracking, the point in the image that the tracked object needs to appear to be locked to.

Log – short for logarithmic.

Log data – large color values represented with logarithmic numbers.

Log images – images that are represented by log data as opposed to linear data. Each step in the data represents a uniform change of apparent brightness to the eye, instead of a uniform change of actual brightness.

Log space – representing image color values in a color space that uses log data.

Logarithmic – where the numbers represent exponents of numbers rather than the numbers themselves.

Look Up Table (LUT) – a table of numbers that uses the value of the input data as an index into the table to look up a second number for output. Commonly used to correct the gamma of a monitor or to assign the colors from a palette to a one-channel image.

Luma key – the matte extracted from an image based on its luminance (brightness) or the use of such a matte to isolate an object from its background for compositing or other treatment.

Luminance

Luminance image – the one-channel image created by mixing the RGB channels in the right proportions so as to maintain the same apparent brightness levels to the eye as the original color image.

Luminance matte – a matte created from a luminance image.

LUTsee Look Up Table.

M

Map projection – the process of fitting a texture map to 3D geometry.

Mask – a one-channel image used to restrict and/or attenuate an operation to a specific region of another image. See also key, alpha, and matte.

Match box – in motion tracking, the rectangular box around a tracking point that is used to compare to the initial match reference image to find a match.

Match reference – in motion tracking, the initial rectangular piece of an image from the first frame of the tracking process to which the subsequent frames are compared to find a match.

Match-moving – to process a live action plate with a program that calculates the original camera move and the location of reference points in three-dimensional space. This data is then used by 2D and 3D artists to add elements to the shot that match the moves of original camera.

Matte – a one-channel image used to define the transparent regions of the foreground and background elements of a composite. See also key, alpha, and mask.

Matte painting – a painting, usually photorealistic, that is typically used as the background for a composite.

Matte pass – a 3D rendering technique where a one-channel mask for one part of a 3D object is rendered to a file to be used for compositing effects such as color correcting.

Maximum operation – an image-processing operation that makes a pixel-by-pixel comparison between two images and selects the greater (maximum) pixel value between the two images as the output. Called “Lighten” in Adobe Photoshop.

Mesh warp – a limited image-warping method that lays a grid over an image so that deformations of the grid lines define the deformation of the image. See spline warp.

Metadata – data about an image embedded with the image such as its resolution, date of capture, ISO, etc.

Midtones – the middle brightness range of an image; not in the very darks or very brights.

Minimum operation – an image-processing operation that makes a pixel-by-pixel comparison between two images and selects the smaller (minimum) pixel value between the two images as the output. Called “Darken” in Adobe Photoshop.

Mitchell filter – performs moderate edge sharpening as it resamples an image. Less prone to edge artifacts than a bicubic filter.

Monochrome – a one-channel image. Often a luminance image, but can be any single-channel image such as the red channel.

Morph – an animation that blends one image into another by warping two images to each other while dissolving between them.

Motion blur – the smearing of an object’s image on film due to its movement while the shutter is open.

Motion estimation (aka optical flow) – a technology that assigns motion vectors to pixel regions of an image to be used for calculating in-between images.

Motion strobing – an unpleasant artifact where an object’s motion takes on a jerky and “staccato” appearance due to the lack of appropriate motion blur.

Motion tracking – tracking the location of a point of interest through a sequence of moving images. Also refers to using the motion-track data to animate a second element to make it appear to be locked to the point of interest.

Multi-pass compositing – the process of rendering and compositing the different render passes of a CGI object separately.

N

Nearest neighbor – method used by an impulse filter to simply choose the nearest pixel from the input image to use in the output image. No actual filtering or averaging.

Nodal point – the point inside a lens assembly where the light beams cross such that if the camera is rotated around this point there is no parallax introduced to the recorded clip.

Node – the image-processing operation represented by an icon within a compositing flowgraph.

Noise displacement – to shift the position of the vertices of a 3D object based on the values of a noise function. Can be used to build terrain and other 3D objects.

Non-linear – to not have a linear relationship between one parameter and another. If one parameter is changed the second one does not change proportionally. For example, when doubling the first parameter the second parameter quadruples.

Normalize – for data, to scale the range of a data set so that all of the data values fall between zero and one.

Normalssee surface normals.

Normals relighting – to relight a CGI image based on the interaction between its normals pass and lights added in 3D.

Null object – a three-dimensional pivot point, or axis, that with rotation, scale, and translation capabilities can be attached to 3D objects to manipulate them.

O

Occlusion pass – a white image with dark regions that indicates how the ambient light is blocked (occluded) on a 3D object, such as in cracks, corners, and crevices.

ODT (Output Device Transform) – the last step in the ACES color management pipeline where the image data is modified based on the characteristics of the display device (TV, digital projector, etc.) in order to maintain the original appearance.

Offset – a color correction operation that simply adds a constant value to every pixel in the image.

One-channel image – an image with pixels that have only one component (brightness), as opposed to the three RGB components of a color image.

Opacity – the attribute of an image that blocks the ability to see through it to the images behind it. The opposite of transparency.

Opaque – an object that cannot be seen through because it is impervious to the transmission of light. The opposite of transparent.

OpenColorIO (OCIO) – a color-management system for vfx studios that provides a pre-defined set of LUTs for vfx artists to consistently convert images into and out of the in-house color workspace.

OpenEXR filesee EXR file.

Optical flow (aka Motion estimation) – a technology that assigns motion vectors to pixel regions of an image to be used for calculating in-between images.

Orthogonal view – a “straight on” view of a 3D object from the front, top, or side that displays no camera perspective.

Overlay blending mode – an Adobe Photoshop blending mode that combines two images without a matte.

P

P3 (DCI/P3) – the official color space for Digital Cinema projection as defined by the Digital Cinema Initiative (DCI).

Pan – to rotate a camera about its vertical axis, which causes the scene to move horizontally in the opposite direction.

Pan and tile shot – a visual effects technique where the far background of a live action scene with a moving camera is replaced with images projected onto 3D cards (tiles).

Panosphere – a sphere within which multiple camera views are projected to completely cover the interior with a scene for VR.

Parallax – the shift in position of an object in a scene when viewed from different camera angles.

Parallax shift – shifts in the relative positions of objects in a scene due to perspective changes caused by a moving camera.

Parallel light – a 3D light source that simulates the parallel light rays from a light source at a great distance, such as the Sun.

Perforations (perfs) – the holes along the edges of film where a registration pin is inserted to precisely position each frame of film.

Perfs – short for perforations.

Perspective – the differing appearance in size and position of objects in a scene due to their different distances and locations from the camera.

Photorealism (photo-realism) – achieving the realism of a single photograph in a composite. Making a synthetic image look realistic.

Photogrammetry – the science of deriving geometric information about a scene from photographs.

Pillar box – a video format where a narrow image is fitted top-to-bottom inside a wider image leaving black bars on either side.

Pivot point – the point in an image around which a geometric transformation occurs, such as the center of rotation or scale.

Pixel – a digital image is composed of a two–dimensional array of many small squares or tiles. A pixel is one of these squares. In many applications the pixels are square, but they can be rectangular. Originated as a concatenation of “picture element”, it is the smallest piece of an image.

Planar tracker – a type of 2D image-tracker that uses large pieces of an image to match frame-to-frame to derive tracking data.

Point cloud – a large collection of 3D points that define an object.

Point light – a 3D light source located at a single point such as an infinitely small light bulb.

Point tracking – a type of 2D image-tracker that uses small pieces of an image to match frame-to-frame to derive tracking data. Often requires multiple points to be tracked for a solution.

Polygon – the three- or four-sided planes defined by vertices that are joined together to define the surface of a 3D object.

Polygonal mesh – a collection of polygons that define a surface.

Position pass – an AOV pass for a CG image that contains the xyz location of the polygonal surface for each pixel in the rendered image.

Premultiplied (pre-multiplied) – an image that has already been multiplied (scaled) by its matte channel to clear the surrounding pixels to zero black. Usually refers to a CGI image with its alpha channel.

Primary chromaticities – the actual color of the three primary colors that make up a color space as located on the CIE chromaticity chart.

Primatte – a third-party keyer available for some compositing packages that uses sophisticated 3D colorspace masking techniques to produce a high quality chroma key composite.

Procedural – to let the computer achieve an objective from a set of rules as opposed to performing the operations by hand. An example would be extracting a matte based on the brightness values of the pixels (a luma key) as opposed to drawing it by hand (rotoscope).

Progressive shutter – a shutter in a digital camera that scans the imaging sensor from top to bottom that can introduce the rolling shutter effect. See also global shutter.

Progressive video – a video-scanning method that scans the picture one line at a time sequentially from top to bottom to make a video frame. See also interlaced video.

Projection aperture – the portion of the exposed film frame that is projected in a theatre. See also camera aperture.

Proxies – small copies of the set of images for a shot that are used to save disk space and processing time while setting up the shot.

Q

Quantization artifact – any one of a number of artifacts that can be introduced into an image due to the fact that it has been digitized.

Quantize – to convert a continuously varying signal into discrete numbers. To digitize. Also, to convert image data from floating-point values to integers, such as converting normalized data (range 0.0–1.0) to 8-bit data (range 0–255).

R

Radians – a unit of measuring angles. 2p radians equals 360 degrees, or about 57.3 degrees per radian.

Ray tracing – a sophisticated rendering technique that correctly accounts for reflections, refractions, and shadows, for more-photorealistic CGI.

Real–time – to perform a computer operation at the same speed as the real world operation. For example, to display digital images on a monitor at 24 frames per second, the same speed as a movie projector.

Rec709 – the official color space for HDTV as defined by SMPTE (Society of Motion Picture and Television Engineers), the standards body for film and television.

Reference image – in motion tracking, the rectangular sample taken on the first frame to compare to the match box from subsequent frames to look for a match.

Reflection mapping – a CGI rendering technique where an image is reflected off the surfaces of geometry.

Reflection pass – a 3D render pass that contains only the reflections on a 3D object.

Refraction – the bending of light as it passes from one medium into another with a different index of refraction. The principle behind lenses.

Render – the computational operations that produce an image. Applies to 3D CGI as well as 2D compositing.

Render layers – to render each object in a 3D scene to a separate file.

Render passes – to render the different surface attributes of a 3D object into separate files to be combined during compositing.

Resample – to compute new pixel values based on some kind of averaging scheme with their neighbors. Necessary whenever an image goes through a geometric transformation. See also filter.

Resolutionsee spatial resolution, color resolution, or temporal resolution.

Retiming – to change the speed of a clip by interpolating new frames in between the original frames.

RGB – abbreviation for “Red, Green, and Blue”, the three color channels of a color image.

RGB scaling – scaling the color values of an image brighter or darker, as opposed to scaling the image size.

Ringing – an artifact introduced after some image-processing operations that artificially increases the contrast of the edges of objects in an image.

Ringing artifacts – excessively high-contrast edges resulting from a filtering operation such as scaling an image up or down in size.

Rolling shutter – a sometimes bizarre image distortion artifact that skews a moving element in a shot, which is caused by the progressive scan of a digital camera shutter.

Rotation – a geometric transformation of an image that changes the orientation of the image about an axis.

Roto – slang for rotoscope.

Rotoscope – to draw mattes frame-by-frame by hand.

RRT (Reference Rendering Transform) – converts the scene-referred data in the ACES color management workflow to the intended appearance of the finished color corrected image.

S

S curve – an “S”-shaped color LUT that increases the contrast of an image without introducing clipping.

Safe-to window – the working window of a film scan beyond which the vfx artist need not operate on since it wont be seen when projected.

Saturation – the attribute of a color that describes its “purity” or “intensity” (not brightness). As a color gets less saturated it becomes more pastel.

Scale – a geometric transformation that changes the size of an image horizontally, vertically, or both.

Scaled foreground – the foreground layer of a bluescreen composite that has been multiplied (scaled) by its matte channel to clear the surrounding pixels to black.

Scan line – a row of pixels the full width of a digital image.

Scene-referred data – imaging data that has had the characteristics of the capturing camera backed out so that the remaining data accurately represents the actual light of the original scene.

Scenegraphs – an interface used with Alembic geometry to select just the items of interest to load in from a 3D scene.

Screen correction – a green/bluescreen pre-processing operation that replaces uneven lighting with a uniform color in the backing region for better keying.

Screen operation – combining two images without a matte using the formula.

Screen space – referencing the location of an element based on its xy position on the viewing screen.

Search box – in motion tracking, a larger rectangular box outside of the match box that the computer will search to looking for a match between the match box and the initial match reference image.

Self-matted – an element such as smoke that is used to create its own matte, usually from its own luminance.

Set extension – a visual effects technique where a small stage set is photographed then extended using 3D animation to make it appear to be a much larger set.

Shader – 3D rendering algorithm that calculates the appearance of the surface of 3D objects based on the materials assigned to them and the light sources.

Shadow pass – a 3D render pass that contains only the shadows of a 3D object.

Sharpening – using one of several image-processing algorithms to increase the apparent sharpness of an image.

Shearsee skew.

Sinc filter – a filter used to resample an image that is optimized for scaling images down without introducing aliasing artifacts.

Skew – a geometric transformation where the opposite edges of an image are offset from each other but remain parallel.

Slice tool – an image analysis tool that, after first drawing a line (slice) across an image, plots the RGB values of the pixels under the line on a graph.

Slot gags – using two moving hicon mattes masked together to create an animated hicon matte. An old film optical term.

Soft clip – to clip the brightness level of an image gradually, as with a color curve instead of a single threshold value. Prevents the “flat spots” introduced by a hard clip.

Soft light blending mode – an Adobe Photoshop blending mode that combines two images without a matte.

Spatial resolution – a measure of how “finely” an image has been digitized horizontally and vertically. Indicated by the width and height of an image in pixels or the number of pixels per unit length, such as 300 ppi (pixels per inch).

Spatial transformation – to reformat an image between different geometric spaces such as rectilinear, spherical, and LatLong.

Specular highlight – a glint reflecting off a shiny surface. More technically, when the angle of reflection of the light is equal to the angle of incidence, as with a mirror.

Specular pass – a 3D render pass that contains only the specular highlights of a 3D object.

Specular shader – a shader that renders specular highlights (shiny spots) on a 3D object based on the surface’s reflectivity and the location and brightness of light sources.

Spherical aberration – a defect in camera lenses that cause a loss of focus on the outer edges of the image.

Spherical projection – a texture-mapping procedure to wrap an image around a sphere.

Spill – stray light that contaminates and discolors objects in a scene.

Spillmap – in the despill operation, a map of the backing color spill that has contaminated the foreground object.

Spline – a mathematically smooth “line” whose shape is determined by a few control points and their settings.

Spline warp – a sophisticated image-warping method that uses splines drawn over an image to define its deformation. See also mesh warp.

Spot light – a 3D light source that can simulate the lighting from a variable spotlight.

Squeezed – an anamorphic image that is horizontally scaled by 0.5 for Cinemascope. See also unsqueezed.

Squirming – in motion tracking, a poor “lock” between the tracked object and the background so that the tracked object appears to drift around over the length of the shot.

sRGB – the official color space for computer monitors that share the same primary chromaticities as rec709 for HDTV but a different gamma and contrast ratio.

Stabilizing – in motion tracking, using the tracking data to remove gate weave, camera jiggle, or other unwanted motions in a shot.

Stereosee stereoscopic.

Stereo compositing – to composite the left and right views of a stereo image pair.

Stereo pair – the left and right pair of images that make up one frame of a stereo movie.

Stereographer – the digital artist that adjusts the depth of the objects in a stereoscopic scene for best appearance.

Stereopsis – the perception of depth derived from two slightly different views of a scene.

Stereoscopic (stereo) – displaying three-dimensional depth in an image using two slightly different views of the scene. Commonly referred to as “3D images”.

Stereoscopy – the techniques of creation and display of stereo images.

Stochastic – a random process where the next data value is independent of any previous data values.

Stop – a doubling or halving of the exposure for an image.

Subpixel – literally, smaller than a pixel, refers to image-processing operations that compute pixel values to a fraction of a pixel with floating-point precision rather than just whole pixel integer values.

Super 35 – any one of several 35mm film formats characterized by using the full width of the film frame (the full aperture), with the intention of reformatting the film later for theatrical projection.

Superwhites – image brightness values greater than 1.0 typically found in light-emitting elements such as fire, lights, explosions, etc.

Surface normals – a vector, or mathematical “arrow”, perpendicular to the surface of a polygon that is used to calculate its angle to a light source and hence its brightness.

Syntax – the rules that govern the terms and symbols used to construct a command for a computer program or a language.

T

Telecine – a machine that scans film at video resolution for transfer to video tape.

Temporal – having to do with time.

Temporal resolution – the number of digital “samples” taken of an event over time, such as a film camera with a shutter speed of 24 frames per second.

Texture map – an image that is mapped, or placed, onto 3D geometry to give it color.

Theater space – refers to objects in stereo 3D films that appear to come off the screen into the audience.

Tiled Image – an image that is regularly repeated like a tiled ceramic surface. Often a small image designed so that it does not show a seam when tiled.

Timecode – running digital “clock” data embedded in the video signal off-screen to mark each frame of the video in terms of hours: minutes: seconds: frames. When printed in a window of each video frame it is called “visible time code” or a “window burn”.

Tone – the darker or lighter versions of a given color.

Track – using a computer to find the location of a target object frame-by-frame so that a second element may be animated to appear locked to it. Also the path itself that is created by the tracked object.

Tracking markers – objects deliberately placed in a scene to provide good tracking targets, usually removed later.

Tracking point – the on-screen graphic placed on an image by the operator to mark the feature to be motion-tracked by the computer.

Tracking targets – the features in a moving image marked by the tracking points that are to be motion-tracked by the computer.

Transfer function – one part of the definition of a color space that specifies the mathematical relationship between the incoming signal and the resulting image brightness.

Transformationsee geometric transformation.

Transforms – image-processing operations that alter the size, position, or shape of an image such as translation, rotation or skew.

Translation – a geometric transformation that simply moves an image horizontally, vertically, or both.

Transparency – the attribute of an image that allows images behind it to be visible to some degree. The opposite of opacity.

Transparent – the attribute of an object that permits light to pass through it so that objects behind it are visible. The opposite of opaque.

Turbulence – in computer graphics, chaotic changes in the variation of bright and dark regions of an image.

U

Uberkey – in keying, a single master key made up of the best parts of other keys.

Ultimatte – a keyer offered by the Ultimatte Corporation based on color-difference matte-extraction techniques that is often included as a third-party module in compositing software.

Unpremultiply – the mathematical operation of reversing the premultiply operation of a four-channel CGI image by dividing its RGB channels by its alpha channel.

Unsharp mask – an image-processing operation borrowed from conventional photography that sharpens an image by subtracting a blurred version of an image from itself.

Unsqueeze – to remove the anamorphic squeeze from an image by scaling it horizontally by two.

UV coordinates – the horizontal and vertical coordinates of an image that are used to map it to the surface of 3D geometry.

V

Value – the “V” in HSV colorspace, roughly translates to the brightness value of a color.

Veiling glare – the haze of light added over a photographed image due to non-parallel light rays, lens imperfections, internal reflections, and other optical anomalies.

Vertexsee vertices.

Vertices – the three-dimensional points that are connected to define the polygons that define a 3D object.

Vignetting – the darkening of the perimeter of an image due to some light being blocked by a camera’s lens housing.

VistaVision – a 35mm large-film format characterized by running the film through the camera horizontally rather than vertically so that the image height is the full width of the film and its width is 8-perfs. Also known as 8-perf.

Volumetric optical flow – in image processing, tracking the motion of pixel regions in three dimensions (xyz) rather than the usual two dimensions (xy).

VR (Virtual Reality) – the creation of a simulated 3D environment that viewers can explore and manipulate in an immersive way.

VR stitching – the processing of blending multiple camera views to create a 360-degree environment for viewing with a VR headset.

W

Warm – a general term to describe an image with a yellowish or reddish cast. See also cool.

Warp – a non–linear deformation of an image where any part of the image can be deformed as if it were a sheet of rubber.

Weighted screen – a variant of the screen operation that adds the use of a matte or mask to partially suppress the background layer to increase the apparent density of the screened object.

White point – the brightest part of an image that will still be present when viewed in its final display mode.

Window violation – in stereo 3D movies where the left and right views do not overlap on the left and right edges.

Witness points – markers or objects deliberately placed in a scene to be used later for 3D tracking references.

World space – in 3D systems, the coordinate system that references the systems’ origin at 0,0,0 as opposed to other references such as camera space or screen space.

X

X – the horizontal direction.

X-axis – the horizontal axis of a graph, a rotation, a scale, or display device such as monitor.

Y

Y – the vertical direction.

Y-axis – the vertical axis of a graph, a rotation, a scale, or display device such as monitor.

YCbCr – a color space that converts RGB values to Y (luminance) and Cb (blue-difference) and Cr (red-difference) chroma components.

Z

Z channel – an additional image plane similar to an alpha channel, but the data represents depth into the scene for each pixel. Used for 3D compositing.

Z-axis – in 3D animation, the axis perpendicular to the X and Y axis. By convention, the axis perpendicular to the viewing screen that is used to represent depth into the scene.

Zero black – the color black represented exactly by the number zero, as opposed to one that just appears black to the eye, which might be somewhat greater than zero.

Zoom – to simulate zooming a camera into (or out of) a scene by scaling an image larger (or smaller) while keeping it trimmed it to its original dimensions.