Video Capture Guidelines

The quality of your MetaHuman animation depends on your footage and performer setup. This page serves as a collection of general best practices for capturing facial and body performance video for MetaHuman Animator, as well as a troubleshooting guide. 

These guidelines apply to all mono video capture workflows: face-only detail capture, body-only capture, and combined face and body capture. Best practices that differ between capture types are reflected as subsections. 

If you're not getting the results you want, try the suggestions here to identify and address possible issues. These are recommendations based on what has worked well. Get started by trying out your own capture and reviewing the results. 

To learn more about mono video capture workflows, including ingesting and processing your footage, see Animation from Mono Video. 

Performer Framing

The following examples show a performer's position that works well while moving within a given aspect ratio.

Face

A slightly lower position with an upward-facing angle is usually preferable. It’s important that the lip seal and the inside of the upper eyelid are visible as often as possible. A tighter framing is recommended for facial-only capture.

Below are examples of framing that may reduce the quality of your results.  

Performer out of frame	Performer out of frame	Performer too low	Performer too high	Off-axis	Performer too close

Body

The performer's full body should be visible from head to feet with a clear margin on all sides. The camera should be positioned straight on at a neutral height so the entire figure is captured.

Below are examples of a performer's and a camera's positions within a given aspect ratio that may reduce the quality of your results. 

Performer out of frame	Performer out of frame	Camera angle too low	Camera angle too high	Off-axis	Performer too close

Stability

A stable camera throughout a recording is essential for all capture types. Camera movement during a take can introduce noise, reducing the solve quality. Use a fixed setup, such as a tripod or stand, to reduce movement and produce consistent data.

Face

For a head-mounted camera (HMC), the performer can move freely around their environment. However, the camera must remain stable relative to the face throughout the recording. This stability achievement requires a well-fitted mount. 

The video below is an example of excessive camera motion that could impact the quality of the results. 

Lighting and Quality

Use even, diffused lighting that fully covers the subject. Avoid harsh directional light, strong shadows, blur, and camera shake. 

Record in HD or 4K at 60fps using H.264 at CRF 18 or better, with a bitrate of approximately 25 Mbps for HD. The performer's face should occupy at least 1/30th of the image width and height, and the full body should be clearly visible with sufficient detail across the limbs. 

A face below 40×80 pixels is likely to fail detection entirely. Body detail can be lost due to heavy compression or low resolution. 

Face

Ideal lighting is flat with no shadows. If ambient lighting is insufficient, additional lights can be attached to the tripod to provide frontal lighting that illuminates the correct portions of the face.

Below are examples of lighting that may reduce the quality of your results. 

Over exposed	Under exposed	Lit from below	Lit from side	Back lit	Shadows	Motion Blur

Body

Ideal lighting is even and diffused, covering the full body from head to feet with no harsh shadows. Consistent ambient lighting helps the solve process distinguish the performer clearly from the background and ground.

Below are examples of lighting that may reduce the quality of your results. 

Over exposed	Under exposed	Lit from below	Lit from side	Back lit	Shadows	Motion Blur

Environment

Face

A background that clearly contrasts with the performer's face works best; a uniform background helps where possible. Nothing in the environment or rig should block any part of the performer's face.

Body

The background for body capture can vary in location and complexity. A uniform background is not required to capture body performance; an environment like a fully white studio, where the performer blends into the surroundings, may reduce solve quality.

Use the table below for guidance to improve your capture if you are getting unexpected results. 

Clothing

Form-fitting or neutral clothing works best. Avoid uniformly colored clothing that matches the background or ground.

Occlusion

Nothing should block any part of the performer's body during the take, including props, furniture, and other people.

Additional Head Mounted Setups

The following are additional guidelines for facial capture when using an HMC. 

Focus

Focus each lens on the nasolabial area of the face (the cheek surface to the side of the nostrils). Depending on the depth of field, you may not be able to keep the entire face in sharp focus.

The following image shows an out-of-focus example.

Lighting (visible on-board lights)

When using an HMC, consider both ambient and on-board lighting. Optimal lighting evenly illuminates the face with minimal shadows and no patches of overexposure.

Below are examples of sub-optimal lighting conditions that can cause quality issues in the output animation data.

Over exposed	Under exposed	Side lit	Room lights visible	Shadows

Next Up

• 

  Depth Data Capture Guidelines

  Set up a stereo head-mounted camera (HMC) to capture depth data for MetaHuman Animator in Unreal Engine.