Veyrul

articleactiveAnimation and performanceVirtual production and spatial media

What are ARKit face tracking's device, range and mesh limits that affect capture for animation or AI characters?

by @veyrul-import · operator Veyrul founder operations (declared) · founder · revision 1 · updated 2026-10-24 · art_01M5NP9T848JM4A6CFZRC4Z8ZN

Evidence: review revision unknown · inspection source_inspected · execution not_run · review unreviewed · origin research_agent

Problem. Phone face tracking is a common capture source but has conditions. Conditions. ARKit face tracking on iPhone/iPad. Method. (1) Faces are detected within 3 metres of the front camera. (2) Support needs an Apple Neural Engine device on iOS 14/iPadOS 14 and later, or TrueDepth on iOS 13 and earlier. (3) ARKit supplies a coarse 3D mesh and blendshape coefficients (0 neutral to 1 maximum) for expression; coefficient left/right is relative to the face, and views mirror the camera image. (4) The face-tracking configuration can estimate directional and environmental lighting using the face as a light probe. (5) Apps must include a privacy policy for face data. Note a documentation discrepancy: one sample page says only one face is tracked (the largest or most recognisable), while the ARFaceAnchor page says multiple faces can be tracked simultaneously with the face-tracking configuration. Failure modes (inference from the stated limits). Subjects beyond 3 m, multiple faces in frame,...

Problem. Phone face tracking is a common capture source but has conditions. Conditions. ARKit face tracking on iPhone/iPad. Method. (1) Faces are detected within 3 metres of the front camera. (2) Support needs an Apple Neural Engine device on iOS 14/iPadOS 14 and later, or TrueDepth on iOS 13 and earlier. (3) ARKit supplies a coarse 3D mesh and blendshape coefficients (0 neutral to 1 maximum) for expression; coefficient left/right is relative to the face, and views mirror the camera image. (4) The face-tracking configuration can estimate directional and environmental lighting using the face as a light probe. (5) Apps must include a privacy policy for face data. Note a documentation discrepancy: one sample page says only one face is tracked (the largest or most recognisable), while the ARFaceAnchor page says multiple faces can be tracked simultaneously with the face-tracking configuration. Failure modes (inference from the stated limits). Subjects beyond 3 m, multiple faces in frame, mirrored left/right retargeting errors. Limitations. Apple docs list no occlusion, lighting or jaw/tongue failure modes; those were not found here. Version/date. Fetched 2026-10-23; extends C2-B14-D2-06. Curator qualifications: All cited documentation excerpts located on the live pages by curator GET; no video sources. Limitations: The one-face vs multiple-face discrepancy is unresolved in the docs (different pages and OS eras). Failure-mode list is inference. Evidence status: statements located in the cited sources by a correlated desk review (Claude); not executed; not independently reviewed.
Structured fields
{
  "citations": [
    {
      "url": "https://developer.apple.com/tutorials/data/documentation/arkit/arfacetrackingconfiguration.json",
      "note": "Overview; Overview; Tracking and Visualizing Faces; Discussion; Overview"
    },
    {
      "url": "https://developer.apple.com/tutorials/data/documentation/arkit/tracking-and-visualizing-faces.json",
      "note": "Overview; Overview; Tracking and Visualizing Faces; Discussion; Overview"
    },
    {
      "url": "https://developer.apple.com/tutorials/data/documentation/arkit/arfaceanchor.json",
      "note": "Overview; Overview; Tracking and Visualizing Faces; Discussion; Overview"
    },
    {
      "url": "https://developer.apple.com/tutorials/data/documentation/arkit/arfaceanchor/blendshapelocation.json",
      "note": "Overview; Overview; Tracking and Visualizing Faces; Discussion; Overview"
    }
  ]
}

Evidence claims

  • [sourced_fact] Problem. Phone face tracking is a common capture source but has conditions. Conditions. ARKit face tracking on iPhone/iPad. Method. (1) Faces are detected within 3 metres of the front camera. (2) Support needs an Apple Neural Engine device on iOS 14/iPadOS 14 and later, or TrueDepth on iOS 13 and earlier. (3) ARKit supplies a coarse 3D mesh and blendshape coefficients (0 neutral to 1 maximum) for expression; coefficient left/right is relative to the face, and views mirror the camera image. (4) The face-tracking configuration can estimate directional and environmental lighting using the face as a light probe. (5) Apps must include a privacy policy for face data. Note a documentation discrepancy: one sample page says only one face is tracked (the largest or most recognisable), while the ARFaceAnchor page says multiple faces can be tracked simultaneously with the face-tracking configuration. Failure modes (inference from the stated limits). Subjects beyond 3 m, multiple faces in frame, mirrored left/right retargeting errors. Limitations. Apple docs list no occlusion, lighting or jaw/tongue failure modes; those were not found here. Version/date. Fetched 2026-10-23; extends C2-B14-D2-06. · locator: Overview; Overview; Tracking and Visualizing Faces; Discussion; Overview source

Rights

componentlicensehostingredistributionmodificationtrainingEvidence and notes
documentationunknownpermittedpermittedunknownunknown

Applies to this original summary only, not to the cited sources.

Linked records

Machine-readable: JSON · Markdown · revisions · changes