Level Sequence Preview

Reviewing a solve normally means building a shot: drop a MetaHuman in a level, wire the animation, light it, aim a camera. Turn on Level Sequence export and each job writes that shot for you — framed, lit, and playable in whatever level you happen to have open.

What gets exported

Export LevelSequence is on by default in Global Settings, so every job adds a Level Sequence to its outputs beside the Performance and AnimSequence. Three things make it reviewable on its own:

Framed camera The camera is placed to frame the face and given a 50mm portrait lens with focus set to the subject. Without footage to match, the engine falls back to a 10mm lens, which reads as a fisheye on a face at conversational distance.

Pinned exposure MetaHuman exposure settings are written onto the camera itself, so the face is correctly exposed no matter which level the sequence is opened in.

Baked-in lighting The chosen lighting rig's lights are copied into the sequence as spawnables. The sequence lights its own face instead of borrowing whatever the current level has.

Face binding

The engine binds a face to an exported Level Sequence only for stereo takes — where it uses the take's own rigged identity. An audio sequence would otherwise play dialogue over an empty frame, and a mono one its plate with nothing animated in front of it, so for those two the plugin binds a face itself, in order of preference:

A MetaHuman actor Blueprint you nominate in settings — the better target, because the engine spawns it and bakes an editable Face Control Rig into the sequence.
A face mesh, bound to a plain animation track playing the AnimSequence the job already exported. Leave it empty and a skeletal mesh built on that AnimSequence's own skeleton is found automatically.

An explicit face mesh on a different skeleton is rejected rather than bound. The shipped face archetypes carry identical bones under different Skeleton assets, so a mismatched pair looks correct and evaluates to nothing — a silent, still face is worse than a clear skip.

A mono sequence that gets a face drops the footage plate: the engine adds it as a standalone plane in front of the take's own camera, and the bound face is reframed at the origin, so the plate would only sit in shot as a stray billboard. Only a face-bound sequence gets the lighting rig — there is nothing for it to light otherwise.

Choosing the lighting rig

The rig is picked by name from Epic's MetaHuman lighting environments — Portrait by default, with Studio, Split, Fireside, Moonlight, Tungsten, RedLantern, and TextureBooth alongside it. The list is discovered from the installed content rather than hardcoded, so environments a future engine adds show up on their own.

Two more sources join the list: the Custom Light Presets registered in MetaHuman Character, shown under their preset names, and every map under the Custom Lighting Environment Folder (/Game/LightingEnvironments by default). A custom level follows Epic's convention - the subject stands at the origin facing +Y with the head at (0, -1, 147) - and only its lights, static meshes and post process volumes are copied into the sequence.

The same list lights the Performance editor viewport, chosen from the Lighting section of the viewport's View menu (see Expression Limits). That path copies nothing: it streams the rig's own map into the preview world, the way the MetaHuman Character editor does, and unloads it again on switch. Streaming only works for a plain level, so a World Partition map is refused there with a warning even though it exports into a sequence normally.

The rigs' shared companion level (the backdrop and its post-process grade) is deliberately left out of the sequence: the preview action opens the sequence against that level anyway, and a second copy inside the sequence would double it up. A project without the MetaHuman Character plugin content exports an unlit sequence rather than failing — the camera's own exposure pin still applies.

Preview in Studio

Right-click any finished job with a Level Sequence artifact and choose Preview in Studio. The sequence opens in Sequencer, ready to scrub.

With Load Studio Backdrop When Previewing LevelSequence on (the default), the editor loads Epic's clean MetaHuman backdrop level first for a controlled background. Loading it replaces the current level, so the editor offers unsaved work up for saving first — declining that prompt cancels the preview rather than half-performing it. Turn the setting off to open the sequence in whatever level is already loaded, with no map switch. A project without the MetaHuman Character plugin content always opens in the current level, and the job says so.

Because the lighting travels inside the sequence, the backdrop is only ever a background. The face stays lit either way.

Wwise dialogue

A Wwise-sourced job's sequence carries a Wwise event track instead of a sound wave track, so the dialogue plays through the sound engine. See Wwise Audio — including the audio routing setting that silences sound wave tracks entirely.

Driving it from a script

preview_level_sequence(job_id) performs the same action from an MCP client, including the backdrop load and its save prompt. See MCP Toolset.