Job Types

Each job in the queue has a type that determines which MetaHuman Animator pipeline it runs, what assets it requires, and which per-job options are available.

Mono

Mono Video

Processes a single color camera recording. This is the standard path for footage captured with a head-mounted camera or any single-lens setup. It is the fastest job type and the right default for most productions.

Required assets

Capture Data UFootageCaptureData Your video footage import asset

Identity UMetaHumanIdentity Auto-discovered if under a standard folder structure

Pipeline stages

Validate Input

→

Process Performance Tracking + Fitting

→

Export Outputs

→

Save Assets

job-type-mono-row.webp Queue row with Job Type set to "Mono" — Capture Data and Identity cells filled, status showing "Ready"

A Mono job row in the queue — Identity and Capture Data assigned, ready to process

Mono-specific options

Head Stabilization Corrects for head movement drift during the solve. Recommended when the subject moves significantly.

Tongue Solve Enables tongue tracking. Requires UE 5.7+ and appropriate capture quality. Disabled by default.

Body Tracking (UE 5.8+)

Mono jobs can optionally solve body motion alongside the face, using the SMPLX-based body tracker. Requires the MetaHumanBodyTracker plugin. Choose Face only, Body only, or Face + Body per job.

Body Height, Detection/Tracking Confidence Auto-infers subject height from footage by default (145–190cm recommended range), with tunable acquire and maintain-lock thresholds.

Foot Locking, Head Rotation Limits Stabilizes foot contact and defines yaw/pitch limits beyond which a frame falls back to a neutral head pose (or is left unsolved).

Body Export Export the solved body onto its own performer-shaped mesh, or retarget it onto an existing skeleton via an auto-generated IK Retargeter (+ Create Retargeter... in the job panel).

Visualization Blueprint Body-tracked jobs need an Actor Blueprint (not a plain mesh) for the A|B viewport. Use Create Blueprint... in the job panel to auto-build one that leader-poses Face to Body.

See Per-Job Options for the full Body Tracking, Body Export, and Visualization option references.

Stereo

Stereo HMC

Processes footage from a multi-camera head-mounted capture rig. Stereo jobs provide depth information that improves solve accuracy, particularly for facial geometry that is ambiguous in monocular footage. Depth sequences can be pre-generated or produced automatically by the plugin.

Required assets

Capture Data UFootageCaptureData Must reference both camera image sequences

Identity UMetaHumanIdentity Auto-discovered if under a standard folder structure

Camera Calibration UCameraCalibrationData Required for depth generation; auto-discovered if available

Pipeline stages

Validate Input

→

Generate Depth Stereo only

→

Clean Depth Optional

→

Process Performance

→

Export + Save

job-type-stereo-depth-gen.webp A Stereo job mid-process at the "Generate Depth" stage — progress bar visible, stage label highlighted

A Stereo job during depth generation — the stage before the performance solve

Depth generation is serial. Even when Max Concurrent Jobs is set above 1, depth generation for all stereo jobs completes before any job enters the concurrent processing pool.

Stereo-specific options

Solve Type Selects the stereo solver variant. Leave at default unless directed by Epic support.

Head Movement Reference Frame Frame number used as the reference pose for head movement correction.

Neutral Pose Calibration Uses a neutral expression frame for calibrating the face geometry baseline.

Face Coverage Threshold Minimum proportion of the face that must be visible in depth for a frame to be used.

Face Width Threshold Minimum face width in depth pixels required to include a frame.

Baseline / Scale Tolerance Acceptable deviation between the stereo baseline in calibration and the live footage.

Enable Depth Cleaning Runs NNE hole-filling on the depth sequence before the performance solve. Preserves GrayF EXR format.

Depth Clamp Min / Max Clips depth values outside this range before cleaning. Useful for eliminating background noise.

Depth Clean-Up Pass

Depth cleaning is a dedicated Clean Depth pipeline stage that runs between depth generation and the performance solve. It is not a blur or a global denoise — it distinguishes background from the subject by flood-filling inward from the image border, then only patches genuine interior gaps (dropouts inside the face/body silhouette) using a distance-weighted average of the surrounding valid depth pixels. Background and out-of-range pixels outside the clamp range are left alone rather than smeared.

Non-destructive Writes a new _Cleaned copy of the depth sequence on disk and as a new UImgMediaSource asset — the original capture data is never modified.

Runs asynchronously EXR I/O and pixel processing run on the thread pool, not the game thread — the editor stays responsive, and the pass is cooperatively cancellable mid-run.

Idempotent Re-running a job that already has a cleaned sequence re-cleans from the newest source instead of stacking cleaning passes on top of each other.

Timed separately Depth cleaning time is tracked independently from solve time (DepthCleaningTimeSecs) and reported in the job log alongside frames-cleaned and sequence counts.

See Per-Job Options for Enable Depth Cleaning and the Depth Clamp Min/Max parameters that control it.

Audio

Audio Only

Drives facial animation from an audio track alone — no video footage required. Useful for generating lip-sync and expression animation from dialogue recordings when no performance capture footage is available.

Required assets

Capture Data UFootageCaptureData Must contain an audio track; video tracks are ignored

job-type-audio-row.webp Queue showing an Audio job row — only Capture Data assigned, Identity cell is greyed out / not required

An Audio job requires only Capture Data — Identity and Calibration columns are inactive

Audio jobs do not require a MetaHuman Identity or Camera Calibration asset. They are the simplest job type to configure.

Audio-specific options

Solver Mode Realtime (faster, lower quality) or Offline (slower, higher quality). Use Offline for final output.

Mood Preset Sets the emotional tone used to color the generated animation — Neutral, Happy, Sad, Angry, etc.

Blink Generation Auto-generates blinks when no blink data is present in the audio solve result.

Audio Channel Which channel from a multi-track audio asset to use for the solve (0-indexed).

Lookahead Window Number of frames the solver looks ahead in the audio signal for anticipation cues.

Universal Options

These options appear on every job regardless of type.

Frame Exclusion Ranges One or more frame ranges to skip during the solve. Use to cut out blinks, obstructions, or bad frames.

Show Frames When enabled, the solver renders preview frames during processing. Useful for debugging; slower in production.

Capture Frame Rate Override the frame rate detected from the capture data. Leave blank to use auto-detection.