{
  "schemaVersion": "1.0",
  "id": "deepswapai-face-swap-output-scorecard-v1",
  "version": "1.0.0",
  "title": "DeepSwapAI face swap output quality scorecard",
  "description": "A transparent human-review rubric for evaluating one photo, video, or GIF face swap output without claiming biometric accuracy or model superiority.",
  "canonicalUrl": "https://deepswapai.com/face-swap-quality-scorecard",
  "datePublished": "2026-07-18",
  "dateModified": "2026-07-18",
  "creator": "DeepSwapAI Product Team",
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "attribution": "DeepSwapAI Face Swap Output Quality Scorecard v1, https://deepswapai.com/face-swap-quality-scorecard",
  "scope": "Human review of one specific face swap output using visible evidence in the sampled media.",
  "mediaTypes": [
    "photo",
    "video",
    "gif"
  ],
  "ratingScale": [
    {
      "value": 0,
      "label": "Unusable",
      "definition": "A failure or severe defect prevents the intended use."
    },
    {
      "value": 1,
      "label": "Major defect",
      "definition": "A clearly visible defect dominates normal viewing."
    },
    {
      "value": 2,
      "label": "Noticeable defect",
      "definition": "A defect remains easy to notice and needs targeted correction."
    },
    {
      "value": 3,
      "label": "Minor defect",
      "definition": "A small defect is visible on review but does not dominate normal viewing."
    },
    {
      "value": 4,
      "label": "No material defect observed",
      "definition": "No material defect was observed in the sampled output for this criterion."
    }
  ],
  "criticalGates": [
    {
      "id": "permission-confirmed",
      "label": "Permission confirmed",
      "definition": "The reviewer has confirmed permission from every real person depicted and the intended use is allowed."
    },
    {
      "id": "identity-mapping-verified",
      "label": "Identity mapping verified",
      "definition": "The output shows the intended identity on the intended person or target throughout the reviewed sample."
    },
    {
      "id": "disclosure-reviewed",
      "label": "Disclosure reviewed",
      "definition": "The reviewer has considered whether the output needs an AI-generated-content disclosure before publication."
    }
  ],
  "criteria": [
    {
      "id": "identity-preservation",
      "label": "Identity preservation",
      "weight": 24,
      "mediaTypes": [
        "photo",
        "video",
        "gif"
      ],
      "reviewQuestion": "Do the visible eyes, brows, nose, mouth, jaw, age cues, and overall identity remain coherent with the intended reference?",
      "sampleGuidance": "Review at normal viewing size, then inspect the face during the hardest visible pose or frame."
    },
    {
      "id": "boundary-blend",
      "label": "Face boundary and blend",
      "weight": 16,
      "mediaTypes": [
        "photo",
        "video",
        "gif"
      ],
      "reviewQuestion": "Do skin texture, face edges, ears, jaw, hairline, and neck transition into the target scene without a pasted-on boundary?",
      "sampleGuidance": "Inspect high-contrast edges and the full face perimeter rather than only the center of the face."
    },
    {
      "id": "pose-expression",
      "label": "Pose and expression coherence",
      "weight": 14,
      "mediaTypes": [
        "photo",
        "video",
        "gif"
      ],
      "reviewQuestion": "Does facial geometry remain coherent with the target head angle, gaze, eye state, mouth shape, and expression?",
      "sampleGuidance": "Prioritize profile turns, closed eyes, open mouths, smiles, and steep head tilts."
    },
    {
      "id": "lighting-color",
      "label": "Lighting and color continuity",
      "weight": 14,
      "mediaTypes": [
        "photo",
        "video",
        "gif"
      ],
      "reviewQuestion": "Do exposure, color, skin shading, highlights, and shadows remain consistent with the target scene?",
      "sampleGuidance": "Check directional light, colored light, shadow boundaries, and transitions between bright and dark regions."
    },
    {
      "id": "occlusion-accessories",
      "label": "Occlusion and accessory continuity",
      "weight": 12,
      "mediaTypes": [
        "photo",
        "video",
        "gif"
      ],
      "reviewQuestion": "Do hair, hands, glasses, masks, microphones, foreground objects, and other occlusions stay in the correct visual order?",
      "sampleGuidance": "Review every point where an object crosses the face or where the face moves behind another object."
    },
    {
      "id": "technical-integrity",
      "label": "Technical integrity",
      "weight": 10,
      "mediaTypes": [
        "photo",
        "video",
        "gif"
      ],
      "reviewQuestion": "Is the output free from material blur, ringing, block artifacts, tearing, duplicate features, abrupt texture changes, or damaged frames?",
      "sampleGuidance": "Inspect at the intended delivery size and avoid treating deliberate motion blur or depth of field as a defect by itself."
    },
    {
      "id": "temporal-stability",
      "label": "Temporal stability",
      "weight": 10,
      "mediaTypes": [
        "video",
        "gif"
      ],
      "reviewQuestion": "Across motion, does identity remain stable without flicker, drift, face loss, sudden geometry changes, or a visible loop seam?",
      "sampleGuidance": "Review the full clip at normal speed, then replay the hardest turn, occlusion, cut, and loop boundary frame by frame."
    }
  ],
  "calculation": {
    "formula": "sum((criterionScore / 4) * criterionWeight) / sum(applicableCriterionWeights) * 100",
    "rounding": "nearest whole number",
    "photoTemporalHandling": "Temporal stability is excluded from photo scores, so the applicable weight denominator is 90.",
    "motionDenominator": 100,
    "photoDenominator": 90,
    "completionRule": "Every applicable criterion must be rated before a numerical score is shown.",
    "gateRule": "The numerical score never overrides an unresolved critical gate."
  },
  "decisionBands": [
    {
      "minimum": 85,
      "maximum": 100,
      "label": "Candidate for acceptance",
      "action": "Complete the critical gates and a final use-specific review before publication."
    },
    {
      "minimum": 70,
      "maximum": 84,
      "label": "Targeted review",
      "action": "Correct or retest the lowest-scoring criteria before publication."
    },
    {
      "minimum": 50,
      "maximum": 69,
      "label": "Regenerate or edit",
      "action": "Change one input or processing variable, then score a new output independently."
    },
    {
      "minimum": 0,
      "maximum": 49,
      "label": "Reject this output",
      "action": "Do not use this output for the intended purpose without a materially different result."
    }
  ],
  "reproducibilityRequirements": [
    "Use the same permitted source and target media when comparing systems or settings.",
    "Record media type, sample identifier, reviewer, date, criterion scores, and criterion notes.",
    "Review motion outputs at normal speed and inspect difficult events frame by frame.",
    "Use multiple independent outputs and multiple reviewers before making a provider-level claim.",
    "Report the number of reviewers, samples, runs, viewing conditions, missing ratings, and dispersion rather than publishing one score alone."
  ],
  "evidenceBoundary": [
    "The score is a structured human observation of one reviewed output, not a biometric identity measurement.",
    "The rubric does not establish model accuracy, average quality, safety, speed, reliability, or provider superiority.",
    "A high score does not replace consent, disclosure, legal, accessibility, or audience-specific review.",
    "A cross-provider claim requires the same inputs, repeated runs, independent reviewers, documented viewing conditions, and statistical reporting."
  ],
  "references": [
    {
      "name": "ITU-R BT.500-15: Methodologies for the subjective assessment of the quality of television images",
      "url": "https://www.itu.int/rec/R-REC-BT.500-15-202305-I",
      "relationship": "General principles for documented subjective image assessment; the DeepSwapAI rubric is not an ITU laboratory test."
    },
    {
      "name": "ITU-T P.910 (10/2023): Subjective video quality assessment methods for multimedia applications",
      "url": "https://www.itu.int/rec/T-REC-P.910/en",
      "relationship": "General principles for documented subjective multimedia assessment; the DeepSwapAI rubric is a practical single-review workflow."
    }
  ]
}
