CNF CERTIFICATION HANDBOOK

1. Purpose & Scope

This handbook defines the operational steps required to certify a scan, assign CNF labels, and expose results through the API. It is intended for analysts, QA engineers, and denim-fit teams evaluating CNF compliance.

• Applies to all scans submitted through the CNF pipeline.
• Defines minimum criteria for CNF Seat Grade, CNF Contour Stability, CNF Tension Readability, CNF Levi’s FitBand, and the Mesh-Ready flag.
• Uses JSON schemas, API rules, and provenance metadata as audit artifacts.

2. Required Inputs

Every certification run begins with a validated scan package:

• Input type: single RGB image.
• Minimum resolution: 720p.
• Coverage: waist through mid-thigh, rear or lateral view.
• Acceptable poses: standing, seated, neutral stance.
• Metadata required: submission ID, consent status, timestamp, analyst ID.

3. Certification Workflow

1. Step 1 — Scan Ingestion
   Upload or capture an image, assign a cnfcaseid, store the raw image in object storage, and trigger the CV orchestrator.
2. Step 2 — CV Processing
   Run segmentation, pose lifting, SMPL-X fitting, cloth mesh reconstruction, and optional depth/normal refinement. Reject if critical stages fail.
3. Step 3 — Metric Extraction
   Compute seat curvature sampling, projection vector, cloth-body distance fields, tension zone classification, and FitBand mapping.
4. Step 4 — Fit-Grade Report Generation
   Assign CNF grades A/B/C, evaluate Levi’s compatibility, and set the Mesh-Ready flag based on reconstruction accuracy and silhouette agreement.
5. Step 5 — Certification Label Creation
   Emit the cert label containing seat grade, contour stability, tension readability, FitBand, pass/fail flags, and Master Mold candidate flag.
6. Step 6 — Storage & API Exposure
   Store all artifacts in the DB and object store, then expose the full CNF package via /api/cnf/scan/{id}.
7. Step 7 — Master Mold Integration
   High-grade cases may be promoted to template updates, subject to analyst review and provenance logging.

4. CNF Metrics & Outputs

All CNF outputs must include the following artifacts:

• cnffitgrade_report — Seat grade, tier, score breakdown.
• cnfmeshspec — Mesh validity, vertex count, reconstruction accuracy.
• cnftensionheatmap — Tension zone intensities and classification.
• cnfcertlabel — Final label with pass/fail flags and seal ID.

5. Compliance Rules

A system, scan, or model is CNF-compliant only if:

• All required metrics are computed and present.
• All output JSON schemas validate successfully.
• Silhouette reconstruction accuracy meets the minimum threshold for the requested tier.
• Provenance metadata is retained for audit and traceability.

6. Master Mold Integration

Certified scans with high-grade CNF metrics may be promoted to Master Mold candidates. Promotion requires:

• Minimum CNF-Elite or CNF-Optimized tier.
• Passed analyst review.
• Consent and provenance verified.
• Logged via /api/master-mold/update with versioned delta.

7. Analyst Responsibilities

Analysts issuing CNF certifications must:

• Verify scan quality and consent before processing.
• Review CV outputs for segmentation or fitting failures.
• Document exceptions and override decisions in scan metadata.
• Maintain confidentiality of scan artifacts and model data.