Node reference

Node reference

Every node in JOS Compositor, by the node picker's categories: what it does, its inputs, and each setting with its default and range. These pages are written from the nodes themselves, so they match the version they come with. A setting's range is the slider's; you can always type a value past it.

3D

• Camera: A camera
• CameraTracker: Follows features over the shot and solves the moving camera (and the points' places), for RenderScene, Project3D and the 3D viewport
• Card: An image on a plane in a scene (Nuke's Card), for set extensions and 2
• Cone: A cone in a scene
• Cube: A box in a scene (Nuke's Cube)
• Cylinder: A cylinder in a scene (Nuke's Cylinder)
• DepthToPoints: Points from a depth pass seen through a camera (Nuke's DepthToPoints)
• FocalFinder: Finds a camera (focal length, rotation, place) from lines drawn along two perpendicular directions of the set, like fSpy
• Light: Adds a light (UsdLux
• Material: An OpenPBR Surface material (MaterialX) for the objects whose material input takes it
• PointCloudGenerator: A CameraTracker's solved points as a point cloud, for the 3D viewport and WriteUSD (sparse
• PositionToPoints: Points from a position pass (Nuke's PositionToPoints)
• Projection: Projects an image through a camera onto a rendered scene (RenderScene's position pass)
• ReLight: A lighting pass from the normal and position channels and Light3D nodes
• Read Splats: Gaussian splats from a file (
• Read USD: Opens a USD scene (
• Read VDB: An OpenVDB volume from a
• RenderScene: Renders a USD scene through a camera at the project format (Embree
• Scene: Gathers objects into a scene
• Sphere: A sphere in a scene (Nuke's Sphere)
• Torus: A torus (a ring) in a scene
• Write USD: Saves the scene with its lights, materials and Outliner settings, and the camera, as a USD file over the source (Render, as a Write)

Analysis

• CurveTool: Measures the picture over frames (average, brightest pixel, crop box) into keyed knobs

Blender Filters

• Brick: Rows of bricks with mortar (Blender's Brick Texture)
• Combine: Builds an image from channels
• Despeckle: Replaces isolated pixels unlike their neighbours by the neighbours' average (Blender's)
• DoubleEdgeMask: A gradient from an inner mask out to an outer mask's edge (Blender's Double Edge Mask)
• Kuwahara: Painterly smoothing that keeps edges (Blender's Kuwahara
• MaskToSDF: A signed distance field of a mask
• Pixelate: Blocky mosaic
• Voronoi: Cells around random points (Blender's Voronoi Texture)
• Wave: Bands or rings (Blender's Wave Texture)

Channel

• ChannelSelector: Shows one channel in every channel (or the luminance, or the matte over the picture)
• Contact Sheet: Every layer of the input side by side in a grid
• Remove Channels: Removes channels, or keeps only the ones listed

Color

• Channel Crosstalk: Each colour channel gets curves of the other two added
• ColorCorrect: Saturation, contrast, gamma, gain and offset for the whole image (master) and for its shadows, midtones and highlights (ranges from the luminance)
• Colour Curves: Curves on the channels' values
• ColorMatrix: R, G, B through a 3 x 3 matrix
• ColorTransfer: Moves the source's colours to the reference's (each channel's mean and spread)
• ColorTransferWrapper: Moves the source's colours to the reference's (each channel's mean and spread)
• Colorspace: Converts R, G and B between transfer curves (Nuke's list
• Dither: Noise of a few code values to break banding
• Exposure: Exposure per channel, in stops, densities, printer lights or Cineon code values (Nuke's Exposure node)
• Expression: Computes channels per pixel from expressions of the pixel's position and colour
• Grade: Nuke-style grade
• HSV Adjust: Picks colours by hue, saturation and brightness ranges and changes them
• Equalize: Histogram equalisation
• Hue Curves: Corrections by hue
• Hue Rotate: Turns colours round the grey axis, with saturation overall and along one colour
• Log ↔ Lin: Cineon log to linear (or back) with the film's black and white points and gamma
• Grade Match: Matches the source's grade to the target's as a CDL
• Min Colour: Adds pixeldelta to the channels (Nuke finds it by analysis
• CDL: An ASC CDL
• Colour Space: Converts R, G and B from one OCIO colour space to another and tags the image with the new space
• Display Transform: Bakes an OCIO display and view (e
• LUT: Applies a LUT or colour correction file (
• Log Convert: Between the OCIO config's scene-linear and log (compositing_log) spaces
• Look: Applies a look from the OCIO config (a grade it defines) between two colour spaces
• Named Transform: One of the OCIO config's named transforms, forward or inverse
• Log ↔ Lin (Pines): Josh Pines' log curve, log to linear or back
• Posterize: Reduces each channel to a number of levels
• Soft Contrast: Contrast around a centre, rolled off softly at the top
• Sampler: Curves on red, green and blue
• Saturation: More or less colour
• Highlight Rolloff: Compresses values above a threshold smoothly towards 1 instead of clipping them
• Toe: Lifts the blacks without moving the whites
• Unmultiply Colour: Takes a background colour out
• Apply 3D LUT: A 3D LUT file applied (from colorspaceIn, back to colorspaceOut)
• Vignette: Darkens or brightens the frame towards its edges, in stops (or towards a colour), with a soft ellipse or a real lens's fall-off

Deep

• Deep Channel Blanker: Sets one channel of every deep sample to 0
• Deep Clip Z: Keeps the deep samples in front of the second input's depth (ours
• Deep Color Correct: ColorCorrect on deep samples, optionally only in a depth range
• Deep Crop: Keeps the deep samples inside a box and a depth range (or outside them)
• Deep De-Overlap: Splits overlapping volume samples where they meet, merging the pieces
• Deep Expression: Expressions computing deep samples' channels from their colour, alpha and depths
• Deep From Image: A flat image as a deep one
• Deep Holdout: The main deep image cut out by the holdout one in front of it, flattened
• Deep Merge: Combines deep images by depth (B and every A), or holds B out by them
• Deep Omit: Keeps the deep samples by a depth keyer (0 outside its ramps drops them)
• Deep Read: Reads a deep OpenEXR file or sequence (#### or %04d for the frame number)
• Deep Recolor: Gives a deep image's samples the colour of a flat image (a render's beauty on its deep samples)
• Deep Reformat: A deep image to a new format or scale (each new pixel takes the input pixel's samples under it)
• Deep Sample: Lists the deep samples under a point (its handle in the viewer), front to back
• Deep Shift: Moves deep samples across the frame and in depth (z + translate z) / zscale
• Deep To Image: Flattens a deep image
• Deep Transform: Moves deep samples across the frame and in depth, and scales their depth
• Deep Volume Maker: A deep volume over the format
• Deep Write: Writes a deep image to deep OpenEXR files (Render, as a Write)
• TimeToDepth: The picture as deep volume slices from zmin to zmax, its alpha split between them

Depth

• DepthGenerator: Depth from the camera's motion (ours, approximate)
• DepthToPosition: Each pixel's position in the world from its depth and the camera

Draw

• DustBust: Removes dust spots by taking each from a neighbouring frame
• Fill: Sets channels to a colour
• Flare: Lens-flare elements (glowing rings or polygons) added to the image
• FloodFill: Fills the connected area around a point with a value
• Grid: Grid lines over the picture
• Inpaint: Fills a region (an alpha or luminance) from what surrounds it
• Noise: Nuke's noise (fBm or turbulence) drawn over the input
• Radial: An ellipse filling a box, soft edged, over the input
• Ramp: A gradient from p0 (nothing) to p1 (the colour), over the input
• Rectangle: A box, over the input
• Paint Roto: Bezier shapes with feathered edges (a matte), drawn in the viewer
• ScannedGrain: Grain from a scanned grain plate, added to the input in proportion to its brightness
• Sparkles: Rays of light from a point (a star burst), added to the image
• Text: Text over the input, in any script (shaped
• Vector Graphics: Bezier shapes and lines with fills and outlines (SVG files import into it), drawn in the viewer
• VolumeRays: Light streaks away from a point through the bright parts of the picture (ours, fitted to Nuke's)

Filter

• Bilateral: An edge-keeping blur
• Blocky: The picture in blocks, each its average
• Blur: Gaussian blur
• Bokeh: Lens defocus from a depth channel
• Convolve: Convolves the image with a filter image (a bokeh or any kernel shape)
• Defocus: A lens blur
• DegrainBlue: Softens the blue channel's grain (blurs its colour, keeps its brightness)
• DegrainSimple: Smooths grain, keeping detail
• Denoise: Removes grain and noise
• Dilate: Grows (positive size) or shrinks (negative) bright areas and mattes by a square of pixels
• Directional Blur: Directional blur (Nuke's DirBlur)
• DropShadow: A shadow of the input's alpha, moved, softened and coloured, under the input
• EdgeBlur: Blurs the image along the edges of a matte (the alpha by default), leaving the rest sharp
• EdgeDetect: The luminance's edges (simple, sobel, prewitt) into a channel
• Edge Detect (Advanced): The edges of the chosen channels, as a matte (alpha by default)
• EdgeExtend: Pushes the colours inside a matte out past its edge (before keying, despill or blurs)
• EdgeScatter: Light from bright areas scattered over their darker edges
• Emboss: Relief
• Erode: Shrinks (or grows) what is in the channels
• FFT: The Fourier transform
• FFTMultiply: Two spectra multiplied (convolution) or divided (deconvolve)
• Erode (Filtered): Erodes (positive size) or dilates (negative) with a soft kernel
• Glint: Star-shaped streaks from the bright parts of the image
• Glow: A glow
• GodRays: Streaks of light
• Grain: Film grain
• InvFFT: Back from the Fourier transform (the fft layer
• Laplacian: Edges and fine detail
• LevelSet: The signed distance to a channel's threshold crossing, its gradient, or a matte
• Remove Markers: Rebuilds what the mask covers (tracking markers) from the pixels around it
• Matrix: A 3 x 3 convolution (sharpen, edges, emboss
• Median: Each pixel becomes the median of the square around it
• Sharpen: Sharpens
• Soften: A blur added back by an amount where values lie between a minimum and a maximum
• VectorBlur: Motion blur from motion vectors
• Depth Blur: Blurs more the further a pixel's depth is from the focus plane
• Depth Defocus: Depth of field from a depth channel
• Depth Defocus (Fast): Defocus by depth

Image

• CheckerBoard: A checkerboard in the project format, with optional lines between squares and through the middle
• ColorBars: Test colour bars in the project format
• ColorWheel: A wheel of hues
• Constant: A solid colour of a given size, in the working colour space
• Read: Reads an image file or sequence
• Write: Writes its input to a file when rendered

Keyer

• ChromaKeyer: A basic chroma key (Keith Jack's)
• Cryptomatte: Mattes from a render's Cryptomatte layers
• Despill: Takes a green (or blue) screen's spill out
• Difference: A matte of where two images differ (e
• HueKeyer: A key from hue
• Plate Key Colour: A clean plate of the screen for IBKGizmo
• Plate Keyer: Image-based keyer
• Keyer: A key from luminance, a channel, saturation, min or max, or a green, blue or red screen, shaped by a range (A rises to B, C falls to D)
• Qualifier: An HSL qualifier
• ScreenKeyer: Screen keyer
• Depth Slice: Keeps what is at a depth

Math

• Add: Adds a value to each channel (per channel for R, G, B, A)
• Clamp: Keeps values between a minimum and a maximum
• Gamma: Raises positive values to 1/gamma
• Invert: 1 - value
• Multiply: Multiplies each channel by a value (per channel for R, G, B, A)

Merge

• AddMix: A added over B, weighted by curves of A's alpha (by default a premultiplied over)
• Blend: The inputs averaged by their weights
• Contact Sheet (Inputs): The inputs side by side in a grid, each scaled to fit its cell
• Copy Bounds: A's image with B's bounding box
• CopyRectangle: A's channels inside a rectangle over B, its edge softened inwards
• Dissolve: Blends between its images
• DualBlend: From the second input (amount 0) to the first (amount 1)
• Mask Mix: A where the mask is white, B where it is black (no merge maths)
• LightWrap: The background's light wrapped onto the foreground's edges (A over B's light)
• Merge: Combines A onto B with a Nuke merge operation (premultiplied maths)
• MergeExpression: Computes B's channels per pixel from expressions of A's and B's colours
• PSDMerge: Photoshop's blend modes
• Premult: Multiplies colour by alpha (straight to premultiplied)
• Switch: Passes on one of its inputs
• Unpremult: Divides colour by alpha where it isn't 0 (premultiplied to straight)
• Depth Comp: A soft depth merge of opaque images
• Depth Merge: Per pixel, the input nearer the camera (smaller Z
• Depth Merge (Reverse): A mixed over B by B's depth

MetaData

• View Metadata: Passes its input through and lists its metadata at the current frame (key and value), as Nuke's

Other

• AddTimeCode: Adds timecode and frame-rate metadata
• Assert: Stops the render with a message when its expression is false
• AudioRead: A sound (WAV, AIFF, MP3, AAC, FLAC, Ogg, or a movie's sound), played with the timeline and drawn in it
• Backdrop: A coloured frame for organising nodes
• Copy Metadata: The Image input's pixels with metadata from the Meta input (and its own)
• DiskCache: Caches the branch above it on disk
• Dot: A routing point for tidying connections
• Edit Metadata: Sets, removes or renames metadata keys
• Nest: A group of nodes shown as one (like Nuke's Group, or Blender's node groups)
• Pass Through: Does nothing
• Thumbnail: Shows its input as a thumbnail in the node graph
• Precomp: Another script's result (a
• StickyNote: A note in the node editor

Time

• AppendClip: The inputs' clips one after another, with dissolves and fades
• DeInterlace: The odd lines rebuilt from the even ones around them
• FieldSelect: One field (odd or even lines), its lines doubled
• FrameBlend: Averages several frames
• FrameHold: Holds the input at a frame (firstFrame)
• FrameRange: Sets the clip's frame range (what follows uses it)
• MotionBlur: Blurs the picture along its own motion (found by optical flow), as a camera shutter would
• 2D Motion Blur: Motion vectors from the 2D transform upstream, over the shutter, for VectorBlur
• Whole Frames: Upstream rendered at whole frames
• OpticalFlow: Retimes the input (slow or fast motion, or any frame), building the frames in between by optical flow (Nuke's Kronos and OFlow2)
• Retime: Plays the input range over the output range at a speed (2
• RollingShutter: Straightens a rolling shutter's lean (phones, most CMOS cameras)
• Detail Vectors: Vectors for VectorDistort
• TemporalMedian: Per pixel the median of the frames before, at and after
• TimeBlend: The average of a few frames ending at this one
• TimeBlur: Motion blur from time
• TimeClip: The input's frames first to last as a clip
• TimeDissolve: Input 1 until in, input 2 from out, a dissolve between
• TimeEcho: Merges the current frame with those before it (plus, max or average)
• TimeOffset: Moves the input in time
• TimeShift: Nuke's older retime
• TimeWarp: Shows the input at the time a curve gives
• VectorDistort: Carries a fix painted on the reference frame through the shot along the plate's motion (src
• Motion Vectors: Motion vectors
• VectorToMotion: Forward vectors into the motion layer, scaled

Transform

• Add UV Map: Adds an identity st-map (layer
• Adjust Bounds: Grows the bounding box (repeating the edge pixels) or shrinks it (negative), e
• BlackOutside: The picture inside the format, with a black edge around it
• CameraShake: A random wobble of translate, rotation and scale (fractal noise)
• Card Transform: The input on a card in 3D, seen through the camera input (or the card's own lens)
• Corner Pin: Moves the four corners of the "from" shape onto the "to" corners (a perspective warp), as Nuke's CornerPin2D
• CornerPinTracker: A Tracker with four tracks, the corners 1 bottom-left, 2 bottom-right, 3 top-right, 4 top-left, that also pins
• Crop: Keeps the pixels inside a box
• GridWarp: Bends the image from one grid to another (Bezier grids, edited in the viewer), or morphs two images
• UV Displace: Moves pixels by a uv pair of channels (a motion or distortion pass)
• LensDistortion: Removes a lens's distortion (Undistort), puts it back (Redistort), or renders both as STMaps, from a lens model with Nuke's knobs, measured against Nuke 16 (radial, asymmetric, tangential, anamorphic squeeze)
• MakeLatLongMap: Six cube faces (right, left, top, bottom, front, back) into a lat-long
• Match Move: Follows two points through the shot and moves the image on its "image" input with them (with nothing connected there the plate passes unchanged)
• Merge Tracks: Merges trackers (Tracker, Stabilize, Match Move) on its yellow inputs into one motion, frame by frame
• Mirror: Flips the image upside down (flip) or left to right (flop)
• Perspective Distort: A rectilinear picture of a field of view as an equidistant one, or back
• PlanarTracker: Follows a flat surface (a screen, a wall, a sign) through the shot from points inside it
• Reformat: Resizes the image to a new format (width x height) or by a scale factor
• UV Remap: Moves pixels by a map
• SphereToLatLongMap: A mirror ball turned into a lat-long
• SphericalMap: A lat-long turned into a mirror ball
• SphericalTransform: An environment image from one projection to another (lat-long, mirror ball, light probes, fisheye
• SplineWarp: Warps the picture with curve pairs
• StabTrack: A Tracker's motion (named) applied
• Stabilize: Follows two points through the shot and undoes their motion on the plate
• Tile: Rows x columns smaller copies of the image in its format, optionally mirrored every other one
• Tracker: Follows points through the shot (tracks)
• Transform: Translates, rotates, scales and skews the image (Nuke's order
• TransformMasked: A Transform of some channels, limited by a mask
• Twist: Turns the picture about a centre, most at the centre, nothing at the edge of scale
• UVTile: A uv image's values moved into one tile (UDIM or u/v), scaled, translated, wrapped
• Upscale: Scales the picture up around the origin, the format kept

Viewer

• Viewer: Shows its input in the viewer

258 nodes.