OGEX is now Kinematic. Graphs are saved as .kin. macOS will ask for camera and microphone access again. On Windows, remove the old OGEX entry from Installed apps yourself.
18releases
170changes listed
19in 0.5.0
0.5.0
Now called Kinematic. Model, sculpt, paint, and edit photos and sound in the viewport.
New project timeline: play, pause, scrub, loop ranges, and markers on every graph. Generators run on Play with no Startup wiring, simulations scrub from their cache, and the timeline can sync to MIDI timecode, LTC, or MIDI clock. New Timeline, TimelineGate, and UITransport nodes.
3D modeling in the viewport: pick points, edges, or faces and edit them directly, and each edit is added to the graph for you. Extrude, inset, bevel, loop cut, knife, slide, poke, bridge, booleans, edge marks, snapping, typed values with units, and mesh tools that work on only the picked parts.
New sculpting nodes: 3DGeoSculpt and 3DGeoDynamicSculpt. Pen pressure and tilt, symmetry, layers, masks, face sets, alphas and stencils, voxel remesh, stroke history, 3DGeoExtractShell, and 3DGeoTrim. Sculpted files reopen without replaying every stroke.
Also in 0.5.0 15 more changes
- Retopology, UV, and baking:
3DGeoRetopo(draw a low-poly mesh over a sculpt),3DGeoQuadRemesh,3DGeoTrisToQuads, automatic unwrap, UV relax / stitch / align / stack / straighten, UDIM tiles, and3DGeoBakeMaps(normal, AO, curvature, and height from a high mesh onto a low one). - Texture painting with
3DGeoTexturePaint(color, roughness, metallic, emissive, and height, with layers and masks), plus3DGeoDecaland3DGeoPlacePointsfor placing pictures and copies by hand on a surface. - Instance picking in the viewport: select, move, or delete single copies of an instanced object.
- Image editing in the viewport: crop, develop, curves, selections (subject, sky, color range, refine edge), paint, text, layers, clone, heal, liquify, and looks.
- New image nodes:
ImageDevelop,ImageSelectShape,ImageMaskRefine,ImageMaskMix,ImagePaint,ImageClone,ImageHeal,ImageLiquify,ImageHueVsCurves,ImageAdjustChain,ImageDetailRecombine,ImageLightWrap,ImageEdgeDecontaminate,ImageGrainMatch,ImageExposureMerge,ImageBracketAlign,ImageStabilize,ImageDeflicker,ImageFrameStack,ImageLutLattice,ImagePlanarTrack, andOutputLutFile. - Audio editing in the viewport: trim, cut, fade, ride the volume, EQ, clean noise, and normalize a range of a clip, with markers and recorded takes. New nodes:
AudioTake,AudioClipPlayer,AudioEditRange,AudioGainEnvelope,AudioRangeMix,AudioDenoise,AudioMarkers, andAudioOnsetList. - New project resolution setting that picture-making nodes can follow, updating a running graph when it changes.
- New Objects browser: a sidebar list of everything the graph makes, with live thumbnails.
- Accessibility support: screen readers, full keyboard control of the canvas and viewport, and high contrast and reduce motion options.
ShaderMaterialupdates: up to 8 texture inputs, scene shadows and environment lighting, and time inputs.ImageShaderwrites 32-bit float on every output, andPbrMaterialgets a height map input.- Send to menus across data types: drive DMX from channels or audio, play channels as MIDI notes, send tables over OSC, project a picture through a light, etc.
- Viewport settings cards replace the old slider strips, with a redone right-click menu and a popup for mapping skeleton or pose fields onto a viewport handle.
- Other new nodes:
3DGeoSequenceRead,3DGeoDimension,3DGeoRigGoalPoints,3DGeoRigPoseBank,PhysicsBodiesFromPieces,StudioActionRequest, and Null nodes for SVG, volume, NURBS, MIDI, DMX, JSON, and OSC. - Python updates: extensions can add items to the viewport Add menus and ask permission before using the GPU, audio, MIDI, or serial ports.
- A wide variety of application fixes and quality of life improvements.
0.4.6
Edit and light splats, import and export characters, and capture performers live.
Gaussian Splat selection and editing in the viewport: click, select, paint, and measure splats. New LocalAISplatSegmentGroup picks out an object in a splat capture from a few clicks, and 3DGeoAttribPaint paints color, opacity, size, numbers, or groups onto points with a brush.
Lit splats: splats now receive scene lights and shadows at render time, cast accurate shadows from every light type, and can be mirrored with 3DGeoMirror. 3DGeoSplatPbrRelight now reads the scene's lights and adds shadows.
3DGeoSplatKernel updates: read up to two images, write spherical harmonics, and know where the camera is. New splat helper library for shader nodes, and helper completion in the kernel editor.
Character import/export: import every take from a file, FBX character export (skeleton, skin, blend shapes, and animation), USD skeleton export, VRM import and export, and BVH / FBX motion export with 3DGeoRigClipExport. Weight painting and a joint table now live in the viewport.
Also in 0.4.6 18 more changes
- Animator tools: a character's skeleton, skins and takes now travel together on one wire (
3DGeoRigCharacterPack,Unpack,AddClip,Merge), grabbable controls with custom shapes (3DGeoRigControls,3DGeoRigControlShapes,3DGeoRigCharacterPose), and rotation keyframe tracks that no longer flip. Plus new IK / FK switching, space switching, twist joints, pose readers and pose drivers, clip sets, and foot / hand contacts with plant locking so feet stop sliding. - Live performance capture:
3DGeoRigMocapReceiveOperation/Interfacetake body and face capture from the network (VMC, Live Link Face, iFacialMocap, VTube Studio, VRChat and custom OSC), with Python adapters for OptiTrack, Rokoko, Xsens, Qualisys, Noitom, and Vicon.3DGeoRigAutoRetargetputs any performer on any character. - Optical mocap:
3DGeoRigMarkersFromFilereads C3D files, and3DGeoRigMarkerSolveturns labelled markers into a skeleton. New joint name presets and Unreal-ready FBX export. - New local AI node:
LocalAIImageInfill. Paints a masked person or object out of a picture. - New local AI node:
LocalAIImageToComposedScene. Turns one photo into a 3D scene, with depth, a camera, placed 3D objects and a background built from the photo. - New local AI node:
LocalAIVideoUpscale. Makes a live video or camera feed 4 times larger, with detail that stays steady from frame to frame. - New 3D geometry nodes:
3DGeoScenePick(what's under a screen position in a render),3DGeoAttribField,3DGeoAttribPhaser(staggered reveals and cascades),3DGeoAttribAccumulate,3DGeoAttribHistogram,3DGeoAttribToImage,3DGeoLookupChannelData,3DGeoLineMetrics,3DGeoLineBreak,3DGeoReconnect,3DGeoPattern,3DGeoSurfaceDeform,3DGeoCameraFromScreenQuad(a camera that looks through a screen like a window), and3DGeoSequenceWrite(numbered ply, obj, stl, glb, or spz files). - New
GeometryComputeShadernode: write your own compute shader that builds new geometry, changing point and face counts as it goes. - New image nodes:
ImageSpectrum(remove moire and screen grain by painting on the frequency view),ImageInverseToneMap(brings SDR video and stills up to sit next to HDR),ImageQrCode,ImageBarcode,InputImageSequence, andMediaFileInfo. - New channel data, DMX and audio nodes:
ChannelDataClock(time of day and countdowns),ChannelDataPhaser(staggered fades across a row of lights),ChannelDataTransport,DmxPanTilt(aims a rig of moving heads at a target),3DGeoToDmxFixtures(points to DMX fixtures), andOutputAudioFile(wav, aiff, flac, mp3, and ogg). - New serial port nodes for talking to Arduinos, dev boards, and other serial devices:
SerialOperation,SerialSendInterface,SerialReceiveInterface,SerialRequest, andSerialDevices. - Color management updates: new
ImageOpenColorIOnode (ACES 2.0 and the standard OCIO configs built in) andImageColorTag. Loaders now read the color space of photos and video (Display P3 photos, correct HD video color, 10-bit video and 16-bit images kept at full precision), and saves write color info back out. HDR output windows on macOS, per-window calibration LUTs for projectors, and AgX, PBR Neutral, and ACES 2.0 tone mapping on the 3D render nodes. - Python extension updates: extensions can add node types that run their own GPU shaders or geometry code, draw and add tools in the viewport, and add panels, sidebar tabs, status bar items, and buttons to existing panels and menus.
- OpenPBR material fixes on all CPU/GPU render paths.
- Updated and redesigned app preferences window.
- Viewport gizmo updates and fixes across the 3D geometry, image, and channel data node families.
- Additional performance optimizations for the localAI node family.
- A wide variety of application fixes and quality of life improvements.
0.4.5
Live head and hand meshes from a camera, rigged splats, and faster booleans.
New local AI node: LocalAIImageToHeadMesh. Fits a live 3D head mesh to each face in a camera image, with skin, eyes, teeth and tongue following head shape, pose, and expression.
New local AI node: LocalAIImageToHandMesh. Live 3D hand meshes and hand skeletons from a camera image.
LocalAIImageToBodyMesh updates: new head refine mode that replaces the body's head with a fitted head (face shape, expression, eyes, teeth, and tongue). Hand refine now works much better, running a second close-up pass on each visible hand for more accurate finger pose detection.
New local AI node: LocalAIImageToAnySkeleton. Poses any rig from a single camera, for any skeleton type, human or not. Wire in a bird, animal, or prop skeleton and the node drives it joint by joint.
Also in 0.4.5 13 more changes
- New local AI node:
LocalAIImageRestore. One-pass photo restoration that cleans up grain, haze, rain, snow, blur, and dark exposure. - New local AI node:
LocalAIImageToVisualRelations. Describes how detected objects relate to each other ("person holding umbrella", "cup on table"). - Rigging for gaussian splat captures:
3DGeoSplatRigFitplaces a skeleton inside a splat scan of a person, and3DGeoRigCaptureBiharmonicbuilds smooth skin weights through a tet cage. Splat skinning now also runs on the GPU. - Rig node updates: soft IK and stretch, swing-twist limits, multi-target constraints, additive clip blending, secondary motion colliders, ragdoll motors, and combination blend shapes.
- Splat updates: binary USD (.usdc) read and write, group support on the splat deformers, per-splat rotation in
3DGeoSplatKernel, and compressed PLY export. - Rewritten
3DGeoBooleanfor much faster and more reliable union, subtract, and intersect on complex meshes. - Drag a channel data or typed data output port onto a number parameter to drive it from that value.
- Custom shader uniforms can now be driven by expressions (channel values, time, other node parameters), including single components of a vector.
- Python and JavaScript script nodes now receive tables and JSON as data instead of text. New
JsonHoldnode for holding a JSON value until triggered. - Images now carry a color tag (display color, linear, or data), so saves, streams, and video outputs convert linear and HDR images correctly. New warnings for mixed or mismatched images.
- Viewport gizmos extended across the splat, image, 3D geometry, rig, NURBS, SVG, audio, channel data, MIDI, DMX, localAI, and Python node families, plus camera frustum handles, attribute and group overlays on meshes, and point editing for 3D cages.
- Performance optimization pass across the splat, rig, 3D geometry, NURBS, SVG, image, audio, and channel data nodes.
- A wide variety of application fixes and quality of life improvements.
0.4.4
A full-screen Perform mode, one undo history, and a viewer for every node.
Perform mode updates: the main window shows the graph's output with the editor hidden, and a kiosk mode puts projector and control surface windows on the monitors you pick. New Outputs window for managing Perform mode settings and show state.
Keyboard, mouse, and touch input now reach the graph from every Kinematic window. New MultiTouchIn and InputIdle nodes, working gamepad support in GamepadInput, and wheel / per-button outputs on MouseInput.
Expansion of the kinematic command line for headless and scripted runs: run and render modes, a virtual clock for repeatable frame-accurate renders, control socket for driving a running graph, a health port and watchdog, device listing, etc.
Unified undo/redo: one timeline across the node canvas, every editor window (DMX, keyframe, piano roll, compositor, etc), and panel layout changes, with named undo entries and a new Undo History panel. Node drags, node flag toggles, driver edits, and MIDI/DMX/OSC mapping edits are now undoable.
Also in 0.4.4 7 more changes
- New node view in the primary viewport: every node now has a viewer showing its ports and wiring, live input/output values, state keys, and recent errors, so nodes that produce no picture are no longer blank in the viewport.
- New image nodes:
ImageTrace(image to line geometry),ImageDespill(green / blue screen spill removal),ImageStarGlow(star / cross-screen filter), andImageMaskCombine(combine up to 16 masks).ImageCompositenow supports per-layer masks and up to 64 layers. CPU and GPU optimized implementations for each. - New 3D geometry nodes:
3DGeoTriangulatePolylines(fills closed outlines into faces),3DGeoAttribTimeFilter(smooths attributes over time per point),3DGeoTransformByAttrib(moves points by per-point transform attributes), and3DGeoCurveFrame(orientation frames along a curve). Plus hit attributes and probe mode on3DGeoRay, and matrix / quaternion attribute support in3DGeoAttribMath,3DGeoAttribInterpolate, and3DGeoAttribTransfer. - Object holdout mattes and a new
ShadowCatcherMaterialnode on all CPU/GPU render paths, for compositing 3D renders over live video and photos. Fixed per-object light linking on the GPU. - Camera capture updates: manual exposure, gain, white balance and auto-exposure lock now work on macOS, new automatic exposure assist for cameras that stay dark, correct Rec.709 color decode, proper resolution / framerate negotiation, and stable camera identity across relaunches. Fixed low frame rates on some USB cameras.
- Auto exposure support on the 3D scene render nodes, and whole-scene glTF export (object hierarchy, materials with textures, instancing, cameras, and lights).
- A wide variety of application fixes and quality of life improvements.
0.4.3
Dock panels where you like, build components, and undo while the graph runs.
Docking updates: arrange, place, show and hide panels however you like.
Big updates to components: wrap part of a graph into one node, choose which settings it shows, pass data in and out, and clone it.
Added SSR (screen space reflection) support to the GPU and CPU rendering engines.
New localAI scene text reading node, for reading signs / printed text visible in a camera frame.
Also in 0.4.3 4 more changes
- Live undo/redo while a Kinematic graph is running (previously graph had to be stopped to undo/redo).
- Wiring two different kinds of data together now adds the converter node for you (
SvgToGeometry,GeometryToImage, etc). - Installers for Windows, Linux and macOS now work on more machines, with better Vulkan and CUDA support.
- A wide variety of application fixes and quality of life improvements.
0.4.2
Scrub and meter sound, export tables, and fly or walk through a scene.
Revamped audio and channel data viewport: waveform, scrub, spectrum, and value views, a timeline ruler with draggable playhead, transport keys, zoom and scrollbars, peak envelopes, dBFS meters with clip latch, listen back from the playhead, and WAV, CSV, JSON, and clipboard export.
Audio viewport now supports live monitor mode with volume control.
New typed data viewport tools: column show and hide, single value cards, recording and scrubbing of published tables, CSV, TSV, JSON, and clipboard export.
New viewport navigation: fly and walk modes, navigation presets with a first run picker, orbit, zoom, and fly speed settings, zoom to cursor, drag inversion, and three button mouse emulation.
Also in 0.4.2 6 more changes
- Viewport gizmos greatly extended in the image, NURBS, audio, and 3D geometry node families.
- New
ImageTimeMachinenode: per-pixel temporal displacement, reading each output pixel from a moment in the recent past chosen by a control image. CPU and GPU. - New
3DGeoParticleChainImageForcenode: pushes particles with an image, reading red and green as a 2d force vector. - Audio and channel data generator nodes (oscillators/etc) now support continuous generation and pumped generation mode.
- Removed extra control inputs across the app; you can now drive the setting itself instead.
- Various application fixes and quality of life improvements.
0.4.0
The center viewport arrives, with handles, scopes, and relightable splats.
New app viewport: resizable viewing and visual editing pane in the center of the application, above the node graph. Displays 3D scenes and geometry, images, volumes, NURBS, SVG, channel data, audio, MIDI, DMX, and tables, following the current selection or pinned to a chosen node.
Parameter gizmos in the primary viewport covering transforms, centers, radii, angles, directions, look-at targets, cones, planes, boxes, and corner pins.
New viewport tools and navigation: move, rotate, scale, and universal handles with numeric entry, snapping to scene geometry, multi-object transforms, custom pivots and orientations, free camera and trackpad navigation, view bookmarks, isolate mode, and a mini scene outliner.
New 2d inspection tools: waveform and histogram scopes, zebra clipping overlays, loupe, line profile, eyedropper and swatches, A/B compare with wipe, before and after peek, background and matte options, and flipbook recording with scrubbing and export.
Also in 0.4.0 5 more changes
- New kinematic.viewport Python module for registering custom gizmos from scripts.
- Gaussian splat relighting (27 new splat nodes): estimates and removes the light a capture was shot under, then relights it with PBR materials, light probes, or an irradiance grid. Splats now cast and receive shadows, sit correctly in fog, work with shadow catchers and holdout mattes, and write depth, normals, and motion vectors so depth-based effects can see them.
- Live per-splat modulation on the GPU: audio, MIDI, OSC, and DMX drive per-splat position, scale, color, and opacity at render time. Includes a custom WGSL kernel node, advection through particle and fluid velocity fields, and conversion between splats and particles.
- New splat deformation and cleanup nodes: per-splat skinning, cage deformation, and mesh-driven warps for rigged and animated captures, plus floater removal, feature and histogram selection, color LUTs, point-based alignment, mesh and volume conversion, and static and dynamic layer splitting for 4D sequences.
- Various application fixes and quality of life improvements.
0.3.0 to 0.2
Earlier releases.
0.3.0
Liquids that pour and foam, OpenPBR materials, and a full character rigging kit.
12 changes
New liquid simulator (FLIP, PIC, APIC, AFLIP), with full GPU/CPU implementations (21 new nodes) for creating water, oceans, and liquids. Supports whitewater foam, spray and bubbles, anisotropic surface meshing, ocean wave and water boundary coupling, rigid body interaction, viscosity, surface tension, and wet maps.
OpenPBR Surface v1.1.1 material support (41 parameters) on all CPU/GPU render paths: IOR based specular, diffuse roughness, coat, fuzz, thin film iridescence, transmission, and subsurface scattering.
New material map slots for fuzz, coat normal, thin film, and transmission.
glTF import now reads the full set of KHR material extensions, plus punctual light falloff, second UV sets, authored tangents, GPU instancing, quantized meshes, and Meshopt compression.
Also in 0.3.0 8 more changes
- New character rigging and animation system (54 new nodes) for rigging, skinning, and animating characters. Supports forward and inverse kinematics, constraints and joint limits, linear blend and dual quaternion skinning on CPU and GPU, automatic skin weights, blend shapes, animation clips with blending and retiming, retargeting, secondary motion, and ragdoll physics.
- Character and animation clip import from glTF, FBX, BVH, USD, and VRM, with glTF and USD export.
- New kinematic.rig Python module for building and posing rigs from scripts.
- Added specular occlusion support to all CPU/GPU render paths.
- MSAA and TAA performance and quality improvements added to all CPU/GPU render paths.
- New local AI node: 3D object splat generation. Creates a 3D gaussian splat of an object from a photo, with controls for detail, splat count, and seed.
- Fixed vertically flipped glTF textures, and various import and render fixes validated against the Khronos glTF sample asset suite.
- Added two new higher quality model options to the background removal node, with sharper cutouts around hair, fur, and fine edges.
0.2.9
See-through materials, up to eight cameras, stereo output, and linear color.
16 changes
New transparency modes for 3D materials: order independent transparency.
New sheen and clearcoat material parameters on all PBR materials.
Per-instance custom attributes: send up to 4 vec4 attributes per instance from geometry into custom shader materials.
Multi-camera rendering: new 3DGeoSceneAddCamera node plus a camera_index setting that selects any of up to 8 cameras, with per-cam TAA, motion vectors, culling, and light masks.
Also in 0.2.9 12 more changes
- Stereo rendering: side-by-side and stacked packed stereo output with eye spacing and convergence controls.
- New camera nodes:
3DGeoLookAt(for looking at objects), and3DGeoCameraPath. - Per-object light linking and draw priority support added.
- Integrated support for per-map UV transforms.
- Color management overhaul of the 3D render pipeline: shading now happens in linear light with correct sRGB handling for texture maps, material colors, and vertex colors. Float/HDR renders stay linear.
- New
InputImageLutFilenode for loading .cube LUTs. - New local AI node: 3D object mesh reconstruction and texture baking from RGB image input.
- New 3dgeo node: body mapping and alignment - maps a localAI body tracking mesh to a 3D object mesh to enable puppeting a textured 3D reconstruction from a live pose.
- Metal shader optimization work for most localAI nodes, various localAI quality fixes.
- Various application fixes and quality of life improvements.
- Improved mouse and keyboard navigation support in all in-app 3D viewports.
- Vertical video file support (auto flipping) in video input nodes and video clip manager.
0.2.8
Windows support, faster local AI on CUDA, and live 3D capture from cameras.
8 changes
Added Windows support. Now supporting macOS (Apple Silicon), Linux Ubuntu (x64), Windows, and Raspberry Pi (still experimental). Code signed Windows installers will be available in the future.
Added heavily optimized fused CUDA shader implementations for 25 LocalAI nodes, so most vision nodes run much faster.
New local AI node: Volumetric 4D capture. Use 1 to 4 uncalibrated cameras to create a volumetric splat of live input.
New local AI node: Live 3D scene reconstruction from a monocular RGB camera input. Creates and reconstructs a 3D point cloud of a scene (indoors or outdoors) based on a moving camera input.
Also in 0.2.8 4 more changes
- New local AI node: whole body pose tracking (body + face + hands combined into a single node).
- Added new LocalAI nodes for 3D body mesh reconstruction from rgb, camera intrinsics identification, 6dof rigid object tracking, 3D camera trajectory tracking, and 3D point tracking.
- Model auto-downloading improvements and progress bar modals added for all localAI models.
- Various Raspberry Pi, Linux, and application fixes.
0.2.7
Over twenty new tracking and detection nodes, plus video and image file output.
10 changes
Added new nodes for hand gesture recognition, 3D body pose tracking, face-part segmentation, body-part segmentation, animal pose tracking, rgb-to-surface-normal, gaze detection, point-based-tracking, mask-based-tracking, neural flow field generation, fiducial marker tracking, face blend shape generation, face expression detection, inappropriate visual content blurring, camera salience region detection, neural edge detection, low light/gamma correction, crowd heatmap generation / counting, accessibility required person detection, vector line detection, speech emotion detection, audio chord detection, and audio sound classification.
Added OutputVideoFile and OutputImageFile output nodes for writing images and video to disk.
Significant performance optimization effort on CPU and GPU (CUDA + Metal) for all 233 Image processing nodes and all LocalAI nodes.
Added Moonshine model support to transcribe audio node.
Also in 0.2.7 6 more changes
- Updated the models behind body pose tracking, prompted blob segmentation, and the animal pose detector stage to a new model family with better keypoint accuracy at the same speed.
- Accuracy improvements to hand tracking and face tracking, and body pose tracking now supports crowds of up to 50 people.
- New
DepthColorAlignnode for automatically aligning depth camera output with RGB camera feeds. - New point cloud processing nodes: depthmap refinement and hole filling, masked point cloud generation, scene parsed/labeled pointcloud generation, depthmap augmented 3D body pose tracking, and room wall/ceiling/floor plane detection.
- New command line options for scripted runs, running until idle, time limits, and a render mode for writing video and other files. Includes new experimental "perform mode" for Kinematic.
- Added preliminary Raspberry Pi support. Now supporting macOS, Linux Ubuntu, and Raspberry Pi. NOTE: Raspberry Pi support should still be considered experimental.
0.2.6
Linux support, HDR rendering, and faster image nodes across the board.
8 changes
Added Linux support. Now supporting macOS and Linux Ubuntu.
Added f32/HDR rendering support on CPU/GPU to the 3D rendering subsystem (simple, scene, cubemap capture, multipass).
Updates to ImageHomography and ImageCornerPin nodes and new editor UI for adjusting corner points.
Updates and functionality improvements to ImageEdgeBlend node.
Also in 0.2.6 4 more changes
- Various improvements and fixes to line rendering and line materials in the 3D rendering subsystem.
- Significant performance optimizations to 100 different image processing nodes on both CPU and GPU.
- Additional performance optimizations to body pose tracking, face tracking, hand tracking, rgb-to-depth, background segmentation, upscaling, and speech transcription on both CPU and GPU.
- Various live-connect and quality of life fixes to the node editor.
0.2.5
Fire and smoke on a sparse voxel grid, running on GPU or CPU.
7 changes
Grid fluid simulator (84 new nodes, many microsolvers, full GPU/CPU implementations with dense and sparse voxel grid support) for creating simulations of fire and smoke.
Stability and speed work on the particle and soft body engines.
Updates and reliability fixes to projection mapping calibration node and the radiometric calibration node.
Added support for the Orbbec Astra 3D depth camera.
Also in 0.2.5 3 more changes
- New
ImageGridWarpinteractive editor GUI and a number of improvements to Projection Mapper editor. - Bug fixes, GPU/CPU equivalency and performance updates in the image and 3dgeo node subsystems.
- Performance optimizations to the nodes for body pose tracking, hand pose tracking, face pose tracking, background segmentation, captioning, rgb to depth, blob segmentation, and image classification.
0.2.4
Float images in 200+ nodes, GPU math you write, and masks on image nodes.
4 changes
More stability and bug fixes on the way to the beta release. Large number of bug fixes across all subsystems.
Native f32 image support added to engine and 200+ image nodes on CPU and GPU.
3DGeoAttribMath and ImageMath nodes with dedicated editor for custom math expressions that execute on the GPU.
Mask support and significant parameter/feature updates for around 80 image nodes to improve the functionality for building procedural textures in Kinematic.
0.2.3
Stability fixes across every part of the app.
1 change
Stability / bug fix release. Large number of fixes across all subsystems.
0.2.2
Particle and terrain updates, plus two node fixes.
2 changes
Updates to the 3D particle engine and terrain simulation subsystem.
3DGeoAttribRandomize and RampEvaluator fixes.
0.2.1
Video streaming, Syphon, and better OSC, MIDI, DMX, and Ableton.
5 changes
RTMP, HLS and RTSP video streaming, plus Syphon improvements.
Significant updates to OSC, MIDI, DMX, and Ableton subsystems.
GPU/Vulkan fixes and hardening.
Also in 0.2.1 2 more changes
- Various minor bugfixes.
- Various minor node updates.
0.2
The first release, where it all started.
1 change
Initial release.
Free and open. Yours, forever.
No account, no license key. macOS, Windows, Linux and Raspberry Pi.