Full-Car exports use the active session's default or imported GLB/STL exterior and See-through body setting. A detached export owns its captured visual selection even if the interactive body is later restored or replaced. Imported geometry stays local and does not alter completed numerical evidence. Reimport after browser reload.
Full-Car Spatial Engineering Video
Complete a Full-Car Spatial road traversal and select Export Video from the results area, including Response. Procedural Flat, Raised cosine and Sine roads and successfully imported supported-subset OpenCRG roads are supported. Prescribed corner inputs / Rig have no video export. Full-Car video uses Drive over Road; choose Engineering 3/4, Chase, Side tracking, Front tracking or Fixed.
Physics-linked suspension motion uses solved Double Wishbone hardpoints and independent spring/damper axes. Constant illustrative damper housings expose internal pistons driven by mount separation. Geometry-dependent motion ratios enter the existing passive force model. Review standalone kinematics verification and its limits.
View distance Near / Normal / Wide applies target-relative camera offset factors 0.75 / 1 / 1.35 with the same fixed 42-degree field of view. Normal retains the existing camera transforms. Fixed stays fixed in world space; road stays stationary while the vehicle moves through the completed traversal.
Publishing retains a fixed completed-evidence envelope fit, disclosed on each playback frame. There is no adjustable vertical display scale. Physical heave, pitch, roll and wheel readouts remain unscaled. Interactive Drive over Road uses uniform physical proportions, orbit, pan and zoom; those temporary camera changes and road appearance never affect video. Publishing owns a separate camera and uses the Engineering road material.
Export settings are explicit and independent of interactive camera, scrub time and play/pause. Presentation edits do not stale results or run simulation. The complete record and settings are captured before asynchronous export. OpenCRG videos use the normalized completed RoadDefinition without reopening a file, parsing CRG or importing again. Cards show compact source/grid/domain context and truncated Road/Traversal SHA identifiers. Full identities remain in JSON and Engineering Report. Container description metadata carries compact engineering provenance.
Full-Car Drive-over-Road video requires WebGL and has no Model fallback. Context loss, cancellation or failure releases owned resources and preserves completed evidence for retry. Encoding stays local with MP4/H.264 preference and VP9/WebM fallback. Detached scenes use explicit publishing-region dimensions and the deterministic frame clock below.
Linear passive 7-DOF model; straight kinematic traversal with prescribed speed. No steering, yaw, lateral or longitudinal dynamics; no tire slip, contact-patch, enveloping or contact-loss model. Road mesh and vehicle graphics are contextual presentation. Imported OpenCRG supports the documented subset only, implies no ASAM certification and does not establish measurement validity. No Full-Car Rig, Compare, Explore or Frequency video. project.v1 is unchanged; presentation preferences and encoded videos are not persisted.
Create a Quarter-Car or Half-Car video
- Complete a Quarter-Car or Half-Car Transient Study and open Motion.
- Select Export Video. Choose Engineering Wide (16:9) or Engineering Vertical (9:16), Model or Vehicle, and 0.5×, 1× or 2× speed. Vehicle offers 3/4 and Side cameras.
- Set an optional title/subtitle (up to 80 characters each), opening card, configuration card and road-event labels. Generate the video. Progress counts rendered/encoded frames; Cancel stops further work. Keep the tab open until the download finishes.
Engineering Wide is 1920 × 1080; Engineering Vertical is 1080 × 1920. Both use 30 fps with physical metrics, original road trace(s), simulation time and a fixed “Generated with VehicleLab” watermark. No audio is added. The maximum output duration including cards is 120 seconds. Vertical reorganizes the visualization and readouts; it does not crop Wide.
Completed evidence and deterministic timing
Export reads the last completed Study, even after input edits. The panel identifies stale evidence; restoring the completed definition exactly restores current status. Export captures its source and settings when generation starts. Changing drafts or interactive playback cannot replace that source.
For physical duration D and speed s, engineering playback duration is D/s. Frame n starts at n/30 seconds within playback and samples physical time s·n/30. There are ceil(30·D/s) playback frames. The final interval is shortened if needed. The exact endpoint closes the final interval; it is not duplicated as an extra frame. A 5 s study at 1× yields 150 playback frames and 5 s of engineering playback. At 2× it yields 75 playback frames and 2.5 s. The optional opening adds 45 frames (1.5 s), the optional configuration card adds 60 frames (2 s), and the closing adds 30 frames (1 s). With both cards enabled, the 2× example totals 210 frames and 7 s. Cards have no engineering time. Interpolation is for display only; physics is never rerun and stored channels remain unchanged.
Publishing context
Opening and closing cards identify Study, Transient and the model. The configuration card shows four curated physical parameters from the completed definition, with units. See the Engineering Report for the full configuration. Road-event labels derive onset, bump peak/end, or step-rise completion from the explicit road definitions. Half-Car front and rear timing use each stored axle road; no event or settling conclusion is inferred from animation.
Titles are user-provided presentation text, not generated engineering conclusions. Settings stay in the mounted publishing panel for the session and are not saved in project files. Filenames include the sanitized project name when available, model, Study and orientation.
Read the engineering context
Readouts show physical acceleration, suspension travel and tire deflection for Quarter-Car; Half-Car shows body acceleration, pitch and front/rear travel. Units are always displayed. Vertical motion is amplified with the same completed-data envelope used by Motion. The gain is disclosed; it does not scale numerical readouts. Half-Car visible pitch follows amplified attachment geometry. Quarter-Car Vehicle wheels mirror one vertical corner and do not imply pitch or roll dynamics.
Front/rear road traces and pads represent prescribed axle inputs, not a spatial road reconstruction. Generic models are not measured-vehicle correlation or engineering recommendations.
Browser support and local processing
The browser must provide WebCodecs VideoEncoder and VideoFrame in a secure context (HTTPS or localhost) with a codec supporting the selected output dimensions. VehicleLab probes actual support, preferring MP4/H.264 and otherwise WebM/VP9. The panel names the selected format before generation. Codec availability depends on browser, OS and hardware; unsupported environments receive an explicit reason. There is no real-time recording fallback.
Vehicle export additionally requires WebGL. A missing module, unavailable context or context loss stops that export explicitly; select Model to use Canvas without WebGL. Response, projects and reports remain available. Encoding failures do not alter completed evidence.
Rendering, encoding and download happen locally. Nothing is uploaded. The encoder/muxer loads only when opening the video panel; Three.js loads only for Vehicle or Full-Car Drive over Road. Export owns its canvases and scene, awaits encoder backpressure and releases resources afterward. Final container writing may finish internally after cancellation, but the canceled artifact is not downloaded.
Publishing limits
Quarter/Half Transient Study and Full-Car Spatial Study only. Compare Motion remains available separately; faithful Compare video needs a dedicated source-selection and shared-scale contract and is deferred. Frequency video, square output, live response plots, free-form editing, custom branding and audio are deferred. Encoding time varies by device and software/hardware codec. Compressed bytes and font/GPU rasterization can differ across machines; explicit timestamps and sampled engineering values remain deterministic. Downloads are buffered in memory; the duration limit bounds normal usage, not a guarantee against device memory exhaustion.
Mediabunny 1.56.2 provides browser-local encoding orchestration and container writing under MPL-2.0. Its unmodified source is available in the corresponding source package.