Glossary

Definitions of terms used in the sensor documentation. Follow the links for model-specific behavior and examples. Entries are alphabetical within each section; Scope identifies the relevant modality, or — for shared terms.

Requesting and authoring

Term	Scope	Meaning
AOV	—	Arbitrary output variable: a single-tensor output a RenderProduct requests through one RenderVar, such as LdrColor. A camera product delivers its images as AOVs. See Output Configuration.
Channel	—	One named tensor inside a composite output, with its own dtype, shape and units. Three other senses of the word are in use; refer to the note below.
Composite output	—	An output containing named tensors (channels) and metadata (params), such as PointCloud, SignalWays or Contributions. Also called an output container; either collection can be empty. See Sensor Outputs, C API, Python API.
Element	Generic wpm	One transmitter, receiver or ray within a pattern. An element index is distinct from the index selecting a stored pattern. See Patterns and selection.
Mapping page	Generic wpm	A stored collection of ordered receiver groups used by map connectivity. See Receiver mapping.
Pattern and family row	Generic wpm	A pattern is one stored alternative for tx, rx or ray. Its index selects a role’s patternTypes entry; its family row counts only preceding entries of that same family. See USD Attributes.
Profile	—	Attribute values that configure a generic sensor model to represent a particular device. Profiles are typically derived from specifications and checked against measurements or a known scene. omni:sensor:marketName identifies the device. See Lidar generic core model, Radar generic wpmDmat model, Camera single-view stereo depth model, tutorials.
Render var output	—	The data delivered for one RenderVar in a frame, keyed by its prim path. It is either a single tensor or a composite output. Mapping yields ovrtx_render_var_output_t in C or MappedRenderVar in Python. See Sensor Outputs.
RenderProduct	—	A USD prim that selects sensors through rel camera and requested outputs through rel orderedVars. Cameras, WPM lidars and the Generic wpm model support multiple sensors as tiles in one product. See Output Configuration.
RenderVar	—	A USD prim that requests one output through its sourceName, such as LdrColor, PointCloud, SignalWays or Contributions. See Output Configuration.
Runtime stage	—	The scene representation held by ovstage and read by the renderer. It can be populated from USD; runtime attribute updates write to this stage. See ovstage Integration, Runtime Parameter Updates.
Sensor introspection utility	—	A utility that validates a sensor definition against the model’s rules before rendering and reports rejected values. Available for both lidar models, radar, acoustic and Generic wpm sensors; unavailable for the camera models. See Sensor Definition Validation.
Sensor model	—	The simulation selected by a prim’s type and applied API schema. It defines attribute behavior and available outputs; identical attribute names in different models need not have identical meanings. See Sensor Configuration.
Sensor prim	—	A USD prim representing a sensor and its configuration: a Camera, OmniLidar, OmniRadar, OmniAcoustic or OmniGeneralSensor prim, with any applicable sensor API schemas. See Sensor Configuration.
Single-tensor output	—	An output that is one tensor and carries no params, such as the camera image LdrColor. It can be consumed directly as a DLPack tensor; in Python the mapping itself is the tensor. See Sensor Outputs.

The word channel has four uses:

• Output tensor: a named tensor in a composite output, selected by the RenderVar’s channels list where supported.

• Lidar detector label: the authored identifier reported by ChannelId. See the Lidar generic core model and the Lidar generic wpm model.

• Acoustic channel id: a firing-event label used with the receiver index to select a signal way. See acoustic channels.

• Image component: R, G, B or A. Camera documentation uses component for this meaning.

In WPM lidar, a receiver is the detector side of an emitter row, not a unique channel id. Two retained rows with the same channelId count as two receivers in receiversPerTile. See scan schedule.

Reading the output

Term	Scope	Meaning
Capacity	Point clouds	The allocated size of a per-point output tensor, shown as Nmax and reported by maxPoints. It can exceed the number of entries delivered in a frame. See Sensor Outputs.
Capture	—	One camera exposure or sensor scan. omni:sensor:frameRate schedules captures; supported models can deliver intermediate results through partialOutputs. See Sensor Scheduling.
Contribution slot	Generic wpm	One dense slot for a tile, transmitter, emitted ray, bounce and receiver slot. Read its payload only when Valid is nonzero. See Outputs.
Count	Point clouds	The delivered-entry count, Counts[0]. It bounds per-entry reads; the tensor capacity and the length of Counts do not. See Delivered-entry count, Sensor Outputs.
Detection	—	One delivered point-cloud entry. Radar detections are extracted from the detection matrix by CFAR. See the radar Outputs page, Detection processing.
DLPack	—	A tensor-exchange convention describing a data pointer, shape, data type and device. Compatible consumers can share tensor storage without copying. Strides are measured in elements, not bytes. See Sensor Outputs.
Frame	—	One delivery of a RenderProduct’s outputs. A step can return zero or more frames. Depending on the model and delivery settings, a frame may contain a partial scan or a complete scan. Also used for a coordinate frame. See Sensor Scheduling, Configure Lidar Output, the note below.
Frame rate	—	The scheduled capture rate, authored as rational Hz in omni:sensor:frameRate. Also called scan rate or capture rate; its reciprocal is the capture period. Internal firing and sampling rates can differ. See Frame rate, step and delivery, Sensor Scheduling.
Mapping	—	Making an output’s tensors readable. After a step, a render var output is mapped on the host (CPU) or on a CUDA device before application code reads its data; params are CPU-resident whatever the mapping. See Sensor Outputs.
Param	—	Named, typed output metadata such as frameId, timestampNs or coordsType. Params are CPU-resident DLPack tensors and can contain scalars, vectors or matrices. See Sensor Outputs.
Point cloud	—	A set of points and associated data. Lidar and radar deliver it as a composite PointCloud output containing coordinates and model-specific channels. See Sensor Outputs.
Receiver slot	Generic wpm	A position in an ordered receiver group, rather than a receiver’s global index. Repeated receiver indices retain distinct slots. See Outputs.
Signal way	Acoustic	The acoustic envelope and metadata for one receiving mount and channel id, delivered in SignalWays. Events from different transmitters can share a signal way. See the acoustic Outputs page.
Stable id	Camera, Lidar	The renderer’s 128-bit identifier of a prim or geometry instance, carried as four uint32 words that are read together: per pixel in StableIdSegmentation and SemanticInstanceSegmentation, per point in the Lidar generic core model’s ObjectId channel. StableIdMap and StableIdMapDeltas map each id to its prim path; NonStableInstanceSegmentation carries a runtime index instead, which can change as the scene changes. Stable ids are active only while some RenderProduct in the scene requests a stable-id output. See Identifier maps and deltas on Decoding camera outputs and to ObjectId on the Lidar generic core model’s Outputs page.
Step	—	One renderer call that advances simulation by delta_time and requests outputs for a set of RenderProducts. Include every active product on every step. See Step every RenderProduct on every step, Sensor Scheduling.
Step result	—	The outputs returned by a step: zero or more frames per requested RenderProduct, with one output per RenderVar in each frame. See Frame delivery, Sensor Outputs.
Tensor	—	A named multi-dimensional array carrying the bulk data of an output, described through DLPack. A variable-sized tensor’s shape is its capacity, and a scalar is a one-element tensor of shape [1]. See Sensor Outputs.

A coordinate frame defines the axes used to express a measurement:

• Sensor frame: the ISO 8855 convention used by composite sensor outputs.

• World or custom frame: selected through outputFrameOfReference and reported by the frameOfReference output param.

• Application frame: the application’s axes; modelToAppTransform maps the sensor frame into them.

See sensor coordinates and Sensor Configuration for the relationship to the sensor prim’s axes.

Scanning and propagation terms

Term	Scope	Meaning
Acquisition time	Radar	The instant at which the Radar generic wpmDmat model acquires a scan: its scan time plus that scan’s timeOffsetUsec delay, reported as the frame’s timestampNs. See the frame rate on the model’s USD attributes page.
Aliasing	Radar	The folding of a contribution whose range, angle or radial velocity lies outside a scan’s unambiguous interval back into that interval, so that the detection appears inside that interval at a plausible value other than the true one: a target beyond maxRangeM comes back at a shorter range. enableRangeAliasing and enAngAliasing switch it on for range and for angle in a FAST scan, which always folds radial velocity. See Aliasing and the unambiguous intervals on the Radar generic wpmDmat model’s USD attributes page. Camera anti-aliasing, the smoothing of jagged edges in a rendered image, is unrelated; refer to DLSS under Camera terms.
Beam divergence	Lidar	The opening angle of a beam in each axis: how fast its width grows with distance. You author either the divergence or the beam waist, and the model derives the other. See Divergence or beam waist on Tune Intensity and Point Drops.
Beam quality factor (M²)	Lidar	How far a real beam departs from an ideal Gaussian beam: 1 is the ideal, and a larger value spreads the beam faster for the same waist. Authored as Msquared. See Tune Intensity and Point Drops.
Beam waist	Lidar	The beam’s radius at its focus, in each axis: the narrowest point, from which the beam diverges. See Tune Intensity and Point Drops.
Boresight	Lidar, Radar	The reference direction at zero azimuth and elevation, used to measure angular limits and antenna or detector response. See Field of view, range and resolution, Detector and detection thresholds.
Chirp	Radar	One frequency-swept pulse of a WAVEFORM scan’s waveform, described by chirpDuration, chirpBandwidth and the selected chirpRepetitionTimeSequence entry. Refer to Waveform processing on the Radar generic wpmDmat model’s USD attributes page.
Contribution	—	Energy arriving at one receiver along one propagation path in a WPM model. A surface hit alone is not a contribution; paths with multiple interactions can produce contributions at different arrival times. See Detection processing. For the Generic wpm model’s raw path records, see Paths and contributions.
Detection floor	Lidar	The processed signal level below which the Lidar generic core model drops a return, set by a reflectance calibration pair on its USD attributes. See Tune Intensity and Point Drops.
Doppler	Radar	The frequency shift used to measure radial motion between sensor and target. RadialVelocityMs reports signed radial velocity; approaching targets have negative values. Motion perpendicular to the line of sight has no radial component. See RadialVelocityMs, Simulate Moving Targets.
Echo	Lidar, Acoustic	For the lidar models, one of several returns from the same firing, numbered by EchoId; refer to the Outputs pages of the Lidar generic core model and the Lidar generic wpm model. For the Acoustic generic wpm model, one arrival of the transmitted pulse at a receiving mount, recorded as amplitude over time in that mount’s signal way; the sampled series is the echo envelope, and an echo from a path that bounced more than once is a ghost echo. See Simulation on the model’s overview.
Emitter	Lidar	One laser in a firing pattern: offset from the tick in azimuth and in fire time, aimed at an absolute elevation. See Configure a Firing Pattern for the Lidar generic core model. The Lidar generic wpm model calls the same thing an emitter row of its firing pattern; refer to its counterpart.
Emitter state	Lidar	An instance of the core lidar emitter-state multi-apply API, containing per-emitter arrays for one application of the firing pattern. The WPM lidar equivalent is a firing-pattern instance. See Configure a Firing Pattern, its counterpart.
Field of view (FoV)	—	The observable region a sensor covers. The Lidar generic core model authors it as how near and how far a return may be plus, for a rotary sensor, which slice of the rotation counts; refer to Constrain the Field of View. The Radar generic wpmDmat model authors angular limits per scan configuration; refer to its USD attributes.
Firing event	Acoustic	An acoustic transmission defined by its time within the cycle, transmitting mount, receiver group and channel id. See Mounts, receiver groups and firing sequences, Lay Out a Transducer Array.
Firing pattern	Lidar	The directions and timing of lidar emissions. Core lidar combines scan and tick settings with emitter-state API instances. WPM lidar uses firing-pattern API instances, such as p001, whose emitter rows fire during one tick. See Configure a Firing Pattern, its counterpart.
Firing sequence	Acoustic	A repeating cycle of timed firing events, each naming a transmitting mount, a receiver group and a channel id. See Mounts, receiver groups and firing sequences on the model’s USD attributes page.
Half-angle	Lidar, Radar	An angular limit measured from boresight to one edge of the field of view, so that the total field of view is twice the value: the Radar generic wpmDmat model’s maxAzAngDeg and maxElAngDeg are half-angles, and so are the limits of the Lidar generic wpm model’s detector field of view. See the half-angles on the Radar generic wpmDmat model’s USD attributes page and to Detector and detection thresholds on the Lidar generic wpm model’s USD attributes page.
Intensity processing mode	Lidar	The Lidar generic core model’s choice of what the reported Intensity number means: raw, normalized or corrected. See Intensity processing on Tune Intensity and Point Drops.
Membrane	Acoustic	The vibrating surface of an acoustic transducer. omni:sensor:WpmAcoustic:membraneDiameter specifies the receiving membrane diameter. See the model’s overview, Physical sensor.
Motion artifacts	—	Distortion caused by sensor or scene motion during a scan, because measurements are taken at different times. See Motion Artifacts.
Motion BVH	—	Renderer support for tracing against sensor and scene transforms that vary during a scan. The mode is selected when creating the renderer and cannot be changed afterwards. See Output Configuration.
Motion compensation	—	Also called deskewing: transforming measurements using their timestamps and sensor motion into a common reference pose. Core lidar selects it with outputMotionCompensationState = COMPENSATED; WPM lidar uses true. The output param motionCompensationState reports the applied mode. See Motion Artifacts, Lidar generic core model, Lidar generic wpm model.
Mount	Acoustic	An acoustic transducer with a position and rotation in the sensor frame. Firing sequences determine whether it transmits, receives or both. See Mounts, receiver groups and firing sequences, Lay Out a Transducer Array.
Multipath	—	Propagation along indirect reflected paths. These can produce ghost detections with different apparent ranges, angles, velocities or strengths. See the radar USD attributes.
Partial beam coverage	Lidar	A beam of finite width landing partly on an object and partly past it at an edge, so that one firing returns from two distances; only the beam ray types reproduce it. See Ray type on the Lidar generic core model’s USD attributes page.
Pending scan	Radar	The scan configuration whose turn is next in the Radar generic wpmDmat model’s cycle; a frame that fires no scan leaves it unchanged. See Scan configuration sequence.
Ray bundle	Generic wpm	An ordered ray pattern reused at each transmitter pose. See Patterns and selection.
Receiver group	Acoustic, Generic wpm	Acoustic: the set of mounts that listen together to a firing event, named by their mount indices. See Mounts, receiver groups and firing sequences on the model’s USD attributes page and Lay Out a Transducer Array. Generic wpm: an ordered list of indices into the active receiver pattern. See Receiver mapping.
Region of interest (ROI)	Lidar	The azimuth interval in which ROI-tagged emitter states of a multi-state firing pattern participate; it is not the field of view. See Constrain the Field of View and the model’s USD attributes.
Return	Lidar, Radar	For the lidar models, a detection: one point; a firing may produce several or none. See the Outputs pages of the Lidar generic core model and the Lidar generic wpm model. In radar prose a return is the reflected signal, whose strength and RCS a detection reports; the delivered unit there is a detection (the radar Outputs page).
Scan	—	One complete pass of a sensor pattern, scheduled at omni:sensor:frameRate. In radar configuration, also an applied scan-configuration instance such as s001. See Sensor Scheduling, Run More Than One Scan Pattern.
Scan time	Radar	One point of the Radar generic wpmDmat model’s scan grid: omni:sensor:startTime plus a whole number of scan periods. Each scan time serves one scan. See the frame rate.
Sweep	—	The same pass as a scan, named to stress that it takes time. See the note below.
Tick	Lidar	One application of the firing pattern, and the Lidar generic core model’s unit of both time and angle; refer to the model’s USD attributes. The Lidar generic wpm model’s tick is the same unit; refer to its scan schedule.
Tile	—	One of several sensors that share a RenderProduct through its rel camera list: a camera product splits its output image into tiles, one per camera, and a Lidar generic wpm model product lays every tile’s points out in one aggregate point cloud; refer to its Tile Lidars page. The Lidar generic core model does not tile: it takes one RenderProduct per sensor (one RenderProduct per sensor). See Output Configuration for independent products.
Trace tree depth	—	The maximum number of material interactions followed along a propagation path. Greater depth allows more indirect paths and increases tracing cost. See Radar generic wpmDmat model, Acoustic generic wpm model, Choose a Return-Processing Mode.

Scan and sweep refer to the same pass of a firing pattern. Sweep emphasizes measurements taken over time; an instantaneous scan, such as lidar with instantLidar enabled, does not sweep through the scene.

Camera terms

Term	Scope	Meaning
Alignment	Camera	Which bit of the output integer a companded code’s most significant bit occupies. See Choose the alignment on the companding how-to.
Bayer pattern	Camera	A 2×2 color filter array with two green, one red and one blue filter; RGGB, GRBG and BGGR name where each sits. The Camera generic core model’s default weights form a GRBG pattern. See Author a Color Filter Array.
Beauty output	Camera	The final shaded image a RenderProduct delivers, as HdrColor and LdrColor; a debug view replaces it. See the camera output tables.
Camera responsivity	Camera	Spectral sensitivity curves for the camera color components. PhysicalCameraResponsivityAPI uses these curves in place of the CIE standard observer during spectral rendering. See Spectral Rendering.
CIE standard observer	Camera	The color-matching functions that turn a spectrum into CIE XYZ, and so into an image, in spectral rendering. The renderer offers the 1931 2-degree observer, the default, and the 1964 10-degree observer. See Spectral Rendering.
Color filter array (CFA)	Camera	Optical filters over sensor photosites, typically in a repeating 2×2 pattern. Each photosite records one filtered value, producing a single-component mosaic. The Camera generic core model specifies the pattern with four RGB weight triples. See Author a Color Filter Array.
Companding	Camera	Compressing internal sensor values into a narrower RAW transport format using a piecewise-linear curve, then expanding them in the ISP. The curve controls how output codes are distributed across the signal range. See Compand to a RAW Format and Run the ISP, Companding.
Dark shot noise	Camera	Noise from thermally generated charge carriers, present even without incident light. See Sensor noise.
Debug view	Camera	A diagnostic quantity, G-buffer data or a heat map, routed into the beauty output in place of the shaded image; selected on the RenderProduct with omni:rtx:debug:view:target. See Outputs.
Demosaicing	Camera	Reconstructing three color components per pixel from the one component the mosaic recorded at each photosite; a step of the ISP. See Compand to a RAW Format and Run the ISP.
Denoiser	Camera	A filter that removes the sampling noise of a path-traced image. Real-Time Path Tracing denoises every frame, Path Tracing has its own denoising settings, and in Minimal mode DLSS Ray Reconstruction denoises. See Render Modes.
Disparity	Camera	The horizontal pixel shift between two imagers’ views of one point, which a stereo camera turns into distance; the Camera single-view stereo depth model computes it from rendered distance and applies its noise and quantization to it. See the model’s overview.
Distortion coefficients	Camera	The numbers that describe how a real lens bends the ideal pinhole mapping; which coefficients each OpenCV schema takes is on Lens Calibration.
DLSS	Camera	Deep Learning Super Sampling, the image reconstruction Minimal mode offers: DLSS Super Resolution (DLSS-SR) upscales and anti-aliases, DLSS Ray Reconstruction (DLSS-RR) upscales, anti-aliases and denoises, and DLAA anti-aliases at native resolution without upscaling. See Minimal.
Firefly	Camera	A bright, isolated pixel produced by a rare high-energy light path. The firefly filter clamps per-sample intensity to suppress them, at some cost in energy. See Real-Time Path Tracing.
Fixed-pattern noise	Camera	Per-photosite offsets and gains that do not change from frame to frame: photo-response non-uniformity (PRNU) in gain and dark-signal non-uniformity (DSNU) in offset. The Camera generic core model does not simulate them; refer to Sensor noise on its USD attributes page.
G-buffer	Camera	The per-pixel geometry and material data (depth, normal, position, texture coordinates and the like) the renderer keeps before shading; debug views expose it. See Outputs.
Global illumination	Camera	Light that reaches a surface indirectly, after bouncing off other surfaces, in addition to direct light; the path-tracing modes compute it. See Render Modes.
Global shutter	Camera	A shutter that exposes all photosites over the same time interval. See Rolling Shutter.
HDR and LDR	Camera	High dynamic range: linear color with a range wider than a display can show, as the float16 HdrColor output. Low dynamic range: display-ready color, as the tone-mapped 8-bit sRGB LdrColor output. See the camera output tables.
HDR zone	Camera	One of the gain ranges an HDR image sensor captures and stitches into one code space; the Camera generic core model’s noise follows the zones. See Sensor noise on its USD attributes page.
Illuminant	Camera	A light source’s spectrum. In spectral rendering, physical:illuminant selects the rendering-space white, a blackbody, CIE daylight or custom spectral samples. See Spectral Rendering.
Image signal processor (ISP)	Camera	The processing chain that turns RAW sensor codes into a viewable image: decompanding, demosaicing, denoising and conversion to a working color space. In the Camera generic core model it runs a program you supply. See Compand to a RAW Format and Run the ISP.
Interpupillary distance (IPD)	Camera	The separation of the two eyes of a stereo camera, in meters; the omni-directional stereo projection renders one eye at a time. See Omni-directional stereo on Lens Calibration.
Intrinsics	Camera	The parameters of a camera’s projection: the focal lengths fx and fy and the principal point cx and cy, in pixels of the calibration imageSize, together with the distortion coefficients. See Lens Calibration.
Knee point	Camera	One (x, y) pair of a companding curve: an internal value and the output code it maps to. See Compand to a RAW Format and Run the ISP.
Light path expression (LPE)	Camera	An expression that selects which light-transport paths accumulate into a RenderVar, splitting a Path Tracing image into passes such as diffuse reflection, glossy reflection, emission or caustics. See Light Path Expressions.
Lookup-table (LUT) projection	Camera	A non-parametric lens model defined by two lookup textures: one gives the ray direction for every image position, the other the image-plane position a ray direction is emitted from. See Generalized Projection.
Mosaic	Camera	The single-component image a color filter array produces, one filtered value per photosite; defined under Color filter array (CFA).
Omni-directional stereo (ODS)	Camera	A 360-degree stereo projection rendered one eye at a time, with separate left- and right-eye cameras. See Omni-directional stereo on Lens Calibration.
Path tracing	Camera	Rendering by tracing light paths through the scene with random sampling and accumulating samples per pixel until the image converges. Path Tracing mode does it progressively toward a sample limit; Real-Time Path Tracing adds caching and denoising for real-time rates. See Render Modes.
Pedestal	Camera	A constant offset added before or after the companding curve so that noise around black does not clip at zero. See Compand to a RAW Format and Run the ISP.
Photon shot noise	Camera	Noise caused by random photon arrivals. Absolute noise increases with signal level, while noise relative to the signal decreases. See Sensor noise.
Primary output and tap	Camera	Camera generic core output categories: two primary images, at least one of which each product must request, and five optional diagnostic outputs from intermediate pipeline stages. See Primary outputs, Diagnostic taps, Outputs.
Principal point	Camera	The pixel where the optical axis meets the image: cx and cy in calibration pixels. See Lens Calibration.
RAW8, RAW10, RAW12, RAW14	Camera	The transport formats of an image sensor, named by bits per sample. The companding curve maps the internal range onto one of them, and the alignment says where the code sits in the output integer. The Camera generic core model’s ispSmodelSensorRawFormat also names RAW20 and RAW24, which do not fit the 16-bit unsigned integer its ISP reads. See Compand to a RAW Format and Run the ISP.
RCCB, RCCC, CCCC	Camera	Color filter arrays with clear, unfiltered positions. Red-Clear-Clear-Blue trades color separation for low-light sensitivity; RCCC and CCCC go further, toward a monochrome sensor. See Author a Color Filter Array.
Rolling shutter	Camera	A shutter that offsets the sample time across the image, one row or column at a time, so moving straight edges appear curved or skewed. Refer to Rolling Shutter.
Samples per pixel	Camera	The number of light paths accumulated for each pixel. Path Tracing accumulates toward omni:rtx:pt:samplesPerPixel before returning output. See Path Tracing.
Spectral rendering	Camera	Light transport sampled over wavelengths and converted to RGB using an observer or camera responsivity. This accounts for spectral variation in lighting and sensor response. See Spectral Rendering.
Subsurface scattering (SSS)	Camera	Light entering a translucent material, scattering inside it and leaving at a different point; the maxVolumeBounces settings bound its bounces. See Path Tracing.
Tone mapping	Camera	Compressing high-dynamic-range color into the display range; it produces the sRGB LdrColor image from the linear scene radiance. See Outputs.

Abbreviations

Short	Full term	Scope	Meaning
AOV	Arbitrary output variable	—	Defined under Requesting and authoring.
CFA	Color filter array	Camera	Defined under Camera terms.
CFAR	Constant false alarm rate	Radar	The algorithm that extracts detections from the detection matrix while holding the false-alarm rate constant, by comparing a candidate cell against neighboring test cells and excluding guard cells around it. See the radar USD attributes.
dBsm	Decibel square meters	Radar	The logarithmic unit radar cross section is reported in.
DLAA, DLSS, DLSS-RR, DLSS-SR	Deep Learning Anti-Aliasing; Deep Learning Super Sampling and its Ray Reconstruction and Super Resolution variants	Camera	Defined under Camera terms (DLSS).
DMAT	Detection matrix	Radar	The discretized representation of radar returns: a range, velocity, azimuth and elevation datacube populated with ray-tracing contributions, from which CFAR extracts detections. See the Radar generic wpmDmat model.
DSNU	Dark-signal non-uniformity	Camera	Defined under Camera terms (Fixed-pattern noise).
FMCW	Frequency-modulated continuous wave	Radar	A waveform whose frequency varies over time; the Generic wpm model does not perform FMCW processing. See model limits.
FoV	Field of view	—	Defined under Scanning and propagation terms.
HDR, LDR	High dynamic range, low dynamic range	Camera	Defined under Camera terms.
IPD	Interpupillary distance	Camera	Defined under Camera terms.
ISO 8855	The ISO road-vehicle coordinate convention	—	The coordinate convention used as the sensor frame of the composite outputs. See The sensor frame on Sensor Outputs for the statement and the figure.
I/Q	In-phase and quadrature	Signals	Two components representing a coherent signal. Generic wpm scalar contributions do not supply coherent phase. See model limits.
ISP	Image signal processor	Camera	Defined under Camera terms.
LPE	Light path expression	Camera	Defined under Camera terms.
LUT	Lookup table	Camera	Defined under Camera terms (Lookup-table projection).
M²	Beam quality factor	Lidar	Defined under Scanning and propagation terms.
MIMO	Multiple input, multiple output	Radar	A radar with several transmitting and receiving elements: a real aperture of 1 Tx and N Rx, or a virtual aperture of M Tx and N Rx that behaves as 1 Tx and M × N Rx. See the receiving aperture.
ODS	Omni-directional stereo	Camera	Defined under Camera terms.
PRNU	Photo-response non-uniformity	Camera	Defined under Camera terms (Fixed-pattern noise).
RCS	Radar cross section	Radar	A measure of how detectable an object is by radar, reported per detection in dBsm. See the radar Outputs page.
ROI	Region of interest	Lidar	Defined under Scanning and propagation terms.
RTPT	Real-Time Path Tracing	Camera	The default render mode: path tracing with caching and denoising for real-time rates. See Real-Time Path Tracing.
Rx	Receiver	—	The receiving side of a sensor: in radar, the receiving antenna system; in the Acoustic generic wpm model, the mounts a receiver group names in its receiverIndices; in the Lidar generic wpm model, the detector side of one emitter row of the firing schedule (refer to the channel note under Requesting and authoring).
SPG	Sensor Processing Graph	Camera	Processing graph used by camera sensor pipelines.
SSS	Subsurface scattering	Camera	Defined under Camera terms.
Tx	Transmitter	—	The transmitting side of a sensor: in radar, the transmitting antenna system; in the Acoustic generic wpm model, the mount a firing event names in txSensorId.
USD	Universal Scene Description	—	The scene description the sensor prim, its API schemas and its RenderProduct are authored in.
USS	Ultrasonic sensor	Acoustic	An acoustic sensor operating at ultrasonic frequencies. See the overview of the Acoustic generic wpm model.
WPM	Wave propagation model	—	A physics-based wave-propagation approach to sensor simulation, named in the Lidar generic wpm model, the Radar generic wpmDmat model, the Acoustic generic wpm model and the Generic wpm model; each model’s pages describe what it does to the signal.

Related

• Output Configuration — RenderProducts, RenderVars, channels and the renderer options sensors depend on.

• Sensor Outputs — mapping, tensors, params, capacity and count, and the sensor frame.

• Sensor Configuration — prims, applied schemas and the shared attributes.

• Sensor Scheduling — frame rate, captures, steps and the delivery policy.

• C API and Python API — the C and Python definition of the output container.

• Sensor API Schemas — every attribute of every sensor schema, with its type, default and doc string.

• Symptom Index — diagnosing sensor output problems.