Cultural Heritage AI Dataset · Verified · v2.0

The Celestial Church
of Christ Garments
Knowledge Base

A specialized, structured dataset focused on rank-specific CCC ecclesiastical garments from the Celestial Church of Christ — curated by an active CCC domain expert at Ondo South Regional HQ to address AI representation gaps regarding West African religious attire. Now evolving toward an AI benchmark.

50
Documented ranks across 5 lines — knowledge graph
36
Metadata fields per image — v2.0 with AI error taxonomy
3
Products: Dataset · 3D · AI Evaluation
✓
V2
AI
+
Built for AI labs, universities & cultural institutions
Domain verified at Ondo South Regional HQ, Okitipupa, Nigeria · Master template: Cape Prophetess
Cape Prophetess — Verified CCC Garment Render — Master Template

Cape Prophetess — Master Template

White Sutana · Square capelet · Royal-blue sash · 3 gold crosses · Fan crown

MASTER TEMPLATE
Domain Verified · Ondo South HQ · v2.0
Knowledge Graph
Female → Prophetess → Cape Elder
female-p · Batch 2 · 5 views
AI Benchmark v2
9 error fields → evaluation tests
✓ Corrective instruction for AI
A specialized dataset focused on rank-specific CCC ecclesiastical garments — built to address AI representation gaps
AI Research Labs
University AI
Cultural Heritage
3D & Fashion AI
AI Benchmark

AI struggles to distinguish rank.
We demonstrate — with exact garments.

Our testing identified recurring errors in AI-generated representations of CCC garments and rank-specific details — current models trained on general web data conflate rank-specific ecclesiastical dress with generic choir robes, Catholic vestments, or plain white gowns.

The Celestial Church of Christ has a precise, rank-encoded dress system where sash colour, cross count, lace trim, cape presence, and loin arrangement are not decorative — they are rank identifiers with theological significance.

This dataset was built by an active CCC member and domain expert at Ondo South Regional Headquarters, Okitipupa, Nigeria — with physical access to authentic garments, church documentation, and senior custodians who can verify rank-specific details.

Instead of claiming "no dataset like this exists anywhere", we describe this as a specialized, structured dataset focused on rank-specific CCC ecclesiastical garments — a claim we can demonstrate through our AI Error Benchmark table and evaluation system.

"The blue sash with gold crosses is not decoration — it is rank, it is authority, it is theology. AI must learn this exactly."

— Domain Expert, Ondo South Regional HQ, Okitipupa
Professional Framing — Per Assessor Toye
Instead of: "None of the AI models have the correct garment and ranks"
We use: "Our testing identified recurring errors in AI-generated representations of CCC garments and rank-specific details" — stronger professionally because we can demonstrate it.
Cape Prophetess — Master Template
Generic AI: white robe, no capelet, wrong blue, 1 cross, no fan crown
Verified: white sutana + square capelet + royal-blue sash front bow + 3 gold crosses (1L+2S) + pleated fan crown — exact mapping
AI Error Taxonomy v2.0 — New
Old: single field ai_common_mistake
New: 9 fields — ai_wrong_garment_type, colour, sash, insignia, headwear, cut, component + corrective_instruction → enables AI benchmark evaluation
Lace-Elder Sister
AI: Catholic nun habit or generic female sutana
Verified: cape finished in lace over the sister sutana, yellow loin, 1 cross — custodian verified, Batch 1 position 4
Most Superior Evangelist
AI: Catholic bishop mitre and cope
Verified: light-blue cassock, long blue robe, deep-blue cross-bar, lace on cape and cassock, blue loin with white Altar cross — Batch 5 position 11, not Catholic

CCC Garment Knowledge Graph
50 ranks across 5 lines

The custodian's standard ranks, with one rule that runs through the whole structure: the prophet ladders collapse into the normal ranking. A prophetess does not climb a separate ladder to the top — after Lace Senior Prophetess she continues in the sisters line from Superior Elder Sister. A prophet continues in the clergy ladder from Assistant Evangelist after Superior Wolider. The merge ranks are stored once, so no rank is counted twice. The ladder then runs to its true top: Venerable Most Superior Evangelist → Supreme Evangelist → Pastor, C.C.C. Worldwide, whose garment is restricted — recorded, never rendered without the church's authorisation.

Female

2 lines · 12 ranks · Batches 1 & 2 · Master template batch
Sisters (Standard Line) Batch 1 — 8 ranks7 documented
1 Sister — Plain white sister sutana, entry level, no markers
2 Elder Sister — White sutana with minimal trim, one small cross on the loin
3 Senior Elder Sister — Cape without lace, round neck, yellow loin, 1 cross
4 Lace-Elder Sister — Cape finished in lace over the sister sutana, yellow loin, 1 cross
5 Superior Elder Sister — Round-neck cape with lace, yellow top without lace, yellow loin, 3 crosses, white…
6 Lace-Superior Elder Sister — Round-neck cape with lace, yellow top WITH lace, yellow loin, 3 crosses, white strands…
7 Assistant Mother Celestial — regalia not yet documented — field capture outstandingto capture
8 Mother Celestial — Apex of the sisters line — reached from either the sisters ladder or the prophetess…
Prophetess Line Batch 2 — 4 ranks2 documented
1 Prophetess — Female sutana with 4-corner neck, royal-blue sash, 3 gold crosses at the bow
2 Cape Prophetess — White sutana + square capelet + royal-blue sash front bow + 3 gold crosses (1 large + 2…master template
3 Senior Cape Prophetess — regalia not yet documented — field capture outstandingto capture
4 Lace Senior Prophetess — regalia not yet documented — field capture outstandingto capture
After Lace Senior Prophetess this ladder collapses into the standard ranking — it continues at Superior Elder Sister. The merged ranks are stored once, in female-np, so no rank is counted twice.

Male

3 lines · 38 ranks · Batches 3, 4 & 5 · The ladder runs to the restricted apex
Non-Clergy Leadership Batch 3 — 18 ranks12 documented
1 Brother — Brother sutana — plain white, no markers
2 Onibode — Anointed brother robe — the same garment as the Akawe; the loin is what separates thempartly documentedshared robe
3 Akawe — Anointed brother robe — the same garment as the Onibode; the loin is what separates thempartly documentedshared robe
4 Elder Brother — Brother sutana with taped white loin, 1 cross
5 Cape-Elder Brother — Cape without lace, white loin, 1 cross
6 Lace-Elder Brother — Cape finished in lace over the brother sutana, white loin, 1 cross
7 Superior Senior Elder Brother — Cape with lace, white loin, 1 cross — senior brother grade
8 Assistant Leader — Leader sutana, unlaced white top, 1 cross
9 Leader — Leader sutana, white loin
10 Senior Leader — Leader sutana, blue loin, 3 crosses
11 Superior Senior Leader — Leader sutana laced, grey top, blue cross-bar with yellow strand, 3 crosses, yellow…
12 Assistant Honorary Evangelist — Leader sutana, long yellow robe, 3 crosses, yellow strands
13 Honorary Evangelist — White cassock, long yellow robe, blue-yellow stranded cross-bar, 3 crosses, yellow…
14 Honorary Senior Evangelist — White cassock, long green robe, yellow-white stranded cross-bar, 3 crosses, yellow…
15 Special Most Senior Evangelist — regalia not yet documented — field capture outstandingto capture
16 Special Venerable Most Senior Evangelist — regalia not yet documented — field capture outstandingto capture
17 Special Superior Evangelist — regalia not yet documented — field capture outstandingto capture
18 Special Assistant Venerable Superior Evangelist — regalia not yet documented — field capture outstandingto capture
Onibode and Akawe wear the same robe — the loin is the only visual difference, so the field capture must photograph and describe it.
Prophet Line Batch 4 — 6 ranks3 documented
1 Prophet — Male sutana with 4-corner neck
2 Cape Prophet — regalia not yet documented — field capture outstandingto capture
3 Lace Senior Prophet — regalia not yet documented — field capture outstandingto capture
4 Wolider — Wolider regalia — prophet line
5 Senior Wolider — regalia not yet documented — field capture outstandingto capture
6 Superior Wolider — Superior Wolider regalia — top of the prophet ladder before it continues into the…
After Superior Wolider this ladder collapses into the standard ranking — it continues at Assistant Evangelist. The merged ranks are stored once, in male-c, so no rank is counted twice.
Clergy Batch 5 — 14 ranks10 documented
1 Assistant Evangelist — Leader sutana, long grey robe, 3 crosses, yellow strands
2 Evangelist — White cassock, long grey robe, blue-yellow stranded cross-bar, 3 crosses, yellow strands
3 Senior Evangelist — White cassock, long blue robe, yellow-white stranded cross-bar, 3 crosses, yellow…
4 Most Senior Evangelist — White or pink cassock, long pink robe, green-purple stranded cross-bar, purple strands
5 Venerable Most Senior Evangelist — regalia not yet documented — field capture outstandingto capture
6 Assistant Superior Evangelist — regalia not yet documented — field capture outstandingto capture
7 Superior Evangelist — White or purple cassock, long purple robe, yellow-white stranded cross-bar, yellow…
8 Assistant Venerable Superior Evangelist — Progression rank — reached 10 years after Superior Evangelist (2023 pastoral directive)
9 Venerable Superior Evangelist — Progression rank — 15 years after Assistant Venerable Superior Evangelist
10 Assistant Most Superior Evangelist — Progression rank — 20 years after Venerable Superior Evangelist
11 Most Superior Evangelist — Light blue cassock, long blue robe, deep blue cross-bar, lace on cape and cassock, blue…
12 Venerable Most Superior Evangelist — regalia not yet documented — field capture outstandingto capture
13 Supreme Evangelist — Deep blue cassock, long deep blue robe, purple cross-bar, purple strands
14 Pastor, C.C.C. Worldwide (Spiritual Head) — regalia not yet documented — field capture outstandingto capturerestricted
One list, one source. Every rank above is generated from ranks-full.json — the same file the collector's rank picker and the reviewer's correction desk read, so the three can never disagree. 50 ranks across 5 lines: 34 documented, 2 partly documented, 14 still to capture. A rank that is not yet photographed is shown as to capture rather than quietly omitted — the gap is part of the record.
View knowledge-graph.json → View ranks-full.json →

Raw photo you sent.
Verified render exactly as built.

Master template pipeline: RAW PHOTO → verified garment-only render → 5 standardized views → metadata v2 → AI error taxonomy → verification → 3D component spec. Left: raw reference 20260913_134015 - Copy.jpg (3.5MB outdoor). Right: verified render 1052844521_1789671011208736 - Copy.jpg (71KB pedestal). Both files in public/ exactly as uploaded — zero AI generation. This is Batch 1/2 master template.

Exact raw reference — CAPE-ELDER PROPHETESS
RAW REFERENCE — GROUND TRUTH
Raw Reference — As Worn
Outdoor, pavement, palm trees, candle · Human present · Layer 01
20260913_*.jpg
What AI gets wrong here — v2 taxonomy
ai_wrong_sash: omits side-tie bow, confuses position
ai_wrong_headwear: misses fan crown pleats
ai_wrong_garment_type: generic white gown
✓ Our dataset: royal-blue side-tie bow, pleated fan crown, full-length sutana preserved
Exact verified 3D pedestal render — master template
VERIFIED RENDER — MASTER TEMPLATE
Verified Render — Training Ready
Stone pedestal, transparent mannequin, no human · Layer 02 · 5 views
MASTER TEMPLATE
Exact mapping — 8 components separated for 3D
✓ Sutana — white, full-length, long sleeves
✓ Capelet — white, square cut, shoulder
✓ Sash — royal-blue satin, front bow, 2 tails
✓ Crosses — 3 gold (1L+2S) on tails
✓ Head — pleated fan crown, halo
✓ Loin — matching sash, pleated
CORRECTIVE_INSTRUCTION — FOR AI BENCHMARK
Generate white full-length sutana with square white capelet covering shoulders, royal-blue satin sash tied at front waist with large bow and two tails, 3 gold embroidered crosses (1 large centered + 2 small) on sash tails, white pleated fan crown with circular halo...

One garment perfect.
Then scale to all female.

Per Toye: Finalize one female garment completely before producing hundreds of renders. Cape Prophetess is ideal master template. Establish visual standard, freeze design, then process all female garments using exactly same camera, lighting, background, mannequin and scale. Only garment changes.

📸
Raw Photo
Original as worn, human present, outdoor context, ground truth
🎨
Garment-Only Render
BG removed, verified, person-free, pedestal + transparent mannequin
📷
5 Standardized Views
Front, Back, Left, Right, Detail — frozen camera/lighting
📋
Metadata v2
42 fields, 9 AI error fields, distinctive features
🧪
AI Error Taxonomy
8 wrong_* fields + corrective_instruction → benchmark tests
✓
Verification
Custodian sign-off, Ondo South HQ, master template status
🔷
3D Component Spec
robe, sash, loin, embroidery, cross, lace trim, head covering, capelet → .glb

Visual Standard — Frozen

  • Camera: Fixed 50mm, distance 2.5m, height 1.6m, tripod locked — same for all female garments
  • Lighting: Softbox 3-point, 5600K, 1:1.2 key-fill ratio, no harsh shadows
  • Background: Neutral #F5F0E8 seamless, 10% vignette
  • Mannequin: Transparent female form, size M, no head features, matte
  • Scale: Garment occupies 80% frame height, centered, consistent
  • Resolution: 2048x2560 minimum, PNG + JPG, sRGB color checker calibrated

5 Standardized Views — Cape Prophetess

  • Front: Full garment, bow centered, 3 crosses visible, fan crown front
  • Back: Capelet coverage, sutana drape, sash tail back
  • Left: Left side, sash tail drape, capelet side profile
  • Right: Right side, sash tail drape, bow side
  • Detail: Close-up bow, crosses, fan crown pleats, capelet edge, embroidery
  • Bonus: 4-view grid composite for quick reference (garment-4view.jpg)
1

Establish Visual Standard

Take one garment — Cape Prophetess sample ideal. Create Front, Back, Left, Right, Detail. Freeze camera, lighting, background, mannequin, scale. This is master template.

FRONTBACKLEFTRIGHTDETAIL
2

Female Batch

Process all female garments (Batch 1 & 2, 12 ranks) using exactly same visual standard. Only garment changes. Ensures dataset consistency for AI training.

Batch 1: 8 ranks × 5 views = 40Batch 2: 4 ranks × 5 views = 20
3

True 3D — .glb

Once 2D standardized renders approved, create actual .glb with component separation: robe, sash, loin, embroidery, cross, lace trim, head covering, capelet. Avoids building 3D for incorrect render.

.glbrobesashloincrosslaceheadcapelet

Three layers.
One authoritative record · Private by design.

📸

Raw Reference

Original photographs with permission documentation. Human-present images showing authentic garments as worn in real CCC contexts. Ground truth evidence layer. Private/licensed per commercial access system.

Private — licensed
🎨

Garment Render

Standardised 5-view renders on neutral mannequin — Front, Back, Left, Right, Detail. No identifiable person. AI-training optimised. Same visual standard across batch. Private/licensed.

Private — licensed — 5 views
🔷

3D Asset

True 3D garment models (.glb) with component separation — robe, sash, loin, embroidery, cross, lace trim, head covering, capelet. Unlimited camera angles, exact rank mapping. Phase 3.

Exclusive licensing — Phase 3
CCC-GARMENTS-DATASET/ — v2.0 — Master Template Pipeline
├── 01_RAW_REFERENCE/  ← Original photographs + permissions (human-present) — Layer 01
├── 02_GARMENT_ONLY/   ← BG-removed + mannequin renders — your exact samples — Layer 02 base
├── 03_3D_RENDERS/     ← 5-view standardised: FRONT, BACK, LEFT, RIGHT, DETAIL — frozen visual standard
├── 04_3D_MODELS/      ← .glb component assets — robe, sash, loin, cross, lace, head covering, capelet — Phase 3
├── 05_METADATA/       ← metadata.jsonl per batch — 42 fields v2 — includes 9 AI error fields + corrective_instruction
├── 06_VERIFICATION/   ← Custodian sign-off documents — Ondo South HQ — master template verification
├── 07_DOCUMENTATION/  ← Rank taxonomy + field guide + knowledge graph — 50 ranks — overlap warnings
├── 08_BENCHMARK/      ← AI Error Benchmark — 7 tests — generic AI vs verified — evaluation reports — NEW v2
└── 09_KNOWLEDGE_GRAPH/ ← knowledge-graph.json — 5 lines — Mother Celestial* overlap handling — NEW v2
Commercial Access System — Mature Architecture
Portal already has: access requests, approved deals, access tokens, batch release, Firebase auth, Firestore-controlled visibility, negotiated batch licensing, exclusive licensing, attribution, expert consultation. More mature than simple Hugging Face repo. Raw reference, garment-only and 3D layers can be private/licensed, while sample can be public. Agree with that architecture — do not rush to make entire thing public.

Five batches.
Master template first, then scale.

Batch 2 (Prophetess line) carries the master template — Cape Prophetess — finalise it completely before hundreds of renders. Batch 1 (Sisters) is the standard line every prophetess eventually continues into. Once template perfect, apply to rest of female garments in batches rather than redesigning system each time.

BatchLineRanksViewsStatusProgress
Batch 1Female — Sisters (standard line)Sister → Mother Celestial (8 ranks) — every prophetess arrives here too40 views (8×5)Ready
Photos gathering — establishing visual standard
Batch 2 SAMPLE — MASTER GARMENTFemale — Prophetess LineProphetess → Mother Celestial* (7 ranks) — includes Cape Prophetess master garment — 5 views frozen35 views (7×5)Building — 5 views ready
Verified renders captured — 4-view ready — master template in progress
Batch 3Male — Non-ClergyBrother → Special Superior Evangelist (15 ranks) — systematic75 views (15×5)Upcoming
Planning — knowledge graph defined
Batch 4Male — Prophetic LineProphet → Superior Senior Woleader (6 ranks)30 views (6×5)Upcoming
Scheduled — after female batches
Batch 5Male — ClergySuperior Leader → Most Superior Evangelist (12 ranks)60 views (12×5)Upcoming
Scheduled — senior clergy

Batch status & progress managed from Admin panel → Firestore batches collection. Table auto-syncs. Master template principle: finalize one garment (Cape Prophetess) completely, then apply same pipeline to rest.

From training data
toward AI benchmark — 9 error fields.

Per assessor: Instead of one field ai_common_mistake, we now have 9 fields enabling AI evaluation tests. Generate Cape Prophetess → evaluation system checks: Is there cape? Is sash correct? Is sash position correct? Are crosses correct? Is head covering correct? Is garment length correct? This moves project from training data toward benchmark.

FieldCategoryDescriptionValue / Example — Master Template
rankidentityExact CCC rank titleCape Prophetess — master template
rank_lineidentityWhich of the 5 rank lines — knowledge graphfemale-p · Batch 2
sash_colourgarmentSash colour — rank-specific, exactRoyal Blue #4169E1 satin
sash_positiongarmentHow/where sash arrangedFront waist, large bow, two tails mid-calf
cross_count_or_insigniagarmentNumber/type of crosses3 gold embroidered (1 large + 2 small) on tails
head_coveringgarmentHead covering typeWhite pleated fan crown with circular halo
distinctive_featuresgarmentWhat uniquely identifies this rankSquare white capelet shoulder covering distinctive for Cape Elder
viewimageStandardized view — 5 viewsfront / back / left / right / detail
ai_common_mistakeai — legacyOverall: What AI gets wrong — keep for compatibilityAI generates generic white robe with no cape, wrong sash, no crosses
ai_wrong_garment_typeai — NEW v2NEW: AI confuses garment type with...AI generates Catholic nun habit or generic choir robe instead of CCC sutana with capelet
ai_wrong_colourai — NEW v2NEW: Specific colour errorsAI uses light blue or navy instead of exact royal-blue #4169E1
ai_wrong_sashai — NEW v2NEW: Sash errors — colour, position, bow, tailsAI omits front bow, places sash on shoulder like Catholic stole
ai_wrong_insigniaai — NEW v2NEW: Cross count, insignia errorsAI generates 1 generic cross or 5+ crosses instead of exact 3 (1L+2S)
ai_wrong_headwearai — NEW v2NEW: Head covering errorsAI generates uncovered head or Catholic veil instead of fan crown
ai_wrong_cutai — NEW v2NEW: Garment cut errorsAI generates short dress or sleeveless, omits square capelet
ai_wrong_componentai — NEW v2NEW: Missing componentsAI omits capelet, loin, embroidery — plain white gown
corrective_instructionai — NEW v2 — KEYNEW: Exact instruction to correct AI — used for benchmarkGenerate white full-length sutana with square white capelet, royal-blue sash front bow, 3 gold crosses...
Before — v1.0 — 28 fields
ai_common_mistake — single free text field. Valuable but not evaluable. Cannot automatically check if AI got sash vs cross vs headwear wrong.
After — v2.0 — 42 fields — 8 new AI fields
9 fields total: ai_common_mistake (legacy) + 7 wrong_* + corrective_instruction. Enables: Generate Cape Prophetess → Check: Is there cape? Sash correct? Position? Crosses? Head covering? Length? → AI benchmark.
View metadata-schema-v2.json (42 fields) → View ranks-full.json with v2 taxonomy →

From assertion to demonstration —
AI Error Benchmark table.

Instead of "None of the AI models have the correct garment", we publish benchmark showing recurring errors. This is stronger professionally because we can demonstrate it. Each test checks verified reference vs generic AI output.

TestGeneric AI OutputVerified Reference — Master TemplateStatusEvaluation Method
Sash colourLight blue or white sash (incorrect)Royal Blue #4169E1 satin — documented in sash_colourFAIL — Generic AIColor picker ±15 hue
Cape / CapeletMissing — generates plain robePresent — square white capelet, shoulder covering — distinctive for Cape ElderFAIL — Generic AIBinary present/absent + shape
Cross count1 generic cross or 5+ crosses3 gold embroidered (1 large + 2 small) on sash tails — exactFAIL — Generic AICount + position mapping
Head coveringNo covering or Catholic veilWhite pleated fan crown with circular halo, small emblemFAIL — Generic AIType classification
Loin arrangementMissing or Western skirtMatches sash, pleated, hidden under sash tailsFAIL — Generic AIPresence + colour + pleat
Garment lengthShort or inconsistentFull-length, floor touching, long sleeves — exact cutPARTIALLength ratio + sleeve
Garment typeGeneric white robe / Catholic albCCC sutana with square capelet — not confused with other attireFAIL — Generic AIClassifier: CCC vs Catholic vs choir
Overall — Generic Text-to-Image
1 / 7 tests passed
Demonstrates systematic hallucination, not minor error
Our Verified Reference — Master Template
7 / 7 tests passed
Exact mapping, custodian verified, corrective_instruction included
Try AI Evaluation →
LIVE — Existing Product

1. Dataset Licensing

Companies pay to access photographs/renders/metadata — 5 views, 42 fields, knowledge graph. Negotiated per batch, attribution required. Current portal supports this.
  • ✓ Raw Reference (Layer 01) — private
  • ✓ Garment Render (Layer 02) — 5 views — private
  • ✓ Metadata v2 — 42 fields — includes AI taxonomy
  • ✓ Knowledge Graph — 50 ranks — overlap handling
BUILDING — Phase 1-3

2. 3D Asset Licensing

Companies pay to use .glb garment assets with component separation. Phase 1: visual standard frozen (Cape Prophetess). Phase 2: female batch same standard. Phase 3: true .glb.
  • ✓ Phase 1: Visual standard — Front, Back, Left, Right, Detail — frozen
  • ✓ Phase 2: Female batch same camera/lighting/background/mannequin/scale
  • ○ Phase 3: .glb — robe, sash, loin, embroidery, cross, lace trim, head covering, capelet
View benchmark-tests.json → Open /benchmark.html →

Your verified model.
5 views frozen, exact rank mapping.

Using exactly your renders — not AI approximations. Master template: Cape Prophetess — white sutana with square capelet, royal-blue sash with gold crosses, white pleated head covering. Visual standard frozen: same camera, lighting, background, mannequin, scale for all female garments. Only garment changes.

Cape Prophetess — 4 views: Front, Back, Left Side, Right Side — Exact verified renders
Exact verified renders · 5-view standard · Batch 2 · Cape Prophetess · Master Template
FRONTBACKLEFTRIGHTDETAIL

Rank Selector — Knowledge Graph — 50 Ranks

Component Breakdown — 8 Components Separated for 3D .glb

✓ Sutana — white, full-length, long sleeves — robe component
✓ Capelet — white, square cut, shoulder covering, distinctive for Cape Elder — separate mesh
✓ Sash — royal-blue satin, front waist, large bow, two tails — sash component
✓ Crosses — 3 gold embroidered (1 large + 2 small) on sash tails — cross components
✓ Head Covering — white, pleated fan crown with circular halo, small emblem — head covering component
✓ Loin — matches sash, pleated, hidden under sash tails — loin component
✓ Embroidery — gold thread, cross motif — embroidery component
✓ Lace Trim — none for this rank, but component exists for Lace Elder Sister — lace trim component
Each component is separate mesh in .glb — enables unlimited camera angles, exact rank classification, correction of AI hallucinations. Phase 3: true 3D after 2D renders approved.
MASTER TEMPLATE PRINCIPLE — FREEZE DESIGN
We use exactly your verified renders — not generic AI approximations. Visual standard frozen: same camera, lighting, background, mannequin, scale. Only garment changes. Batch License includes 5-view renders like your 4-view grid. Full Exclusive includes .glb with component separation.

Three products.
You negotiate the terms.

Portal has mature commercial access system: access requests, approved deals, access tokens, batch release, Firebase auth, Firestore-controlled visibility, negotiated batch licensing, exclusive licensing, attribution, expert consultation. More mature than Hugging Face repo. Raw reference, garment-only and 3D layers can be private/licensed, while sample can be public.

Sample Access
Free · No license required · Public sample
  • Public Hugging Face dataset (sample batch) — includes master template metadata v2
  • metadata.jsonl preview (50 records) — 42 fields v2 — knowledge graph sample
  • Dataset structure documentation — 9 layers including benchmark
  • AI Error Benchmark sample — 7 tests — generic AI vs verified
  • Full batch downloads — licensed only
  • Raw reference photographs — private
  • 5-view standardized renders — licensed
  • 3D models (.glb) — exclusive only
View on Hugging Face
Full Exclusive + AI Eval
Strategic partnership · All batches · 3D .glb · AI Benchmark API
  • All 5 batches — all layers — 5 views + .glb — master template pipeline
  • 3D garment models (.glb) with 8-component separation — Phase 3
  • CCC Garment AI Benchmark API — evaluate AI models against verified reference
  • Ongoing dataset updates — new ranks = new verified record, not redesign
  • Domain expert consultation — direct line to Ondo South HQ
  • Co-authorship credit option — cultural AI research
  • First-mover exclusivity window — negotiated
  • Direct line to CCC custodians — verification
Open Partnership Discussion

Request dataset access

Tell us about your organisation and intended use. We review all requests personally and respond within 3 business days. Your licensed download will include exactly the verified renders you saw above — 5 standardized views, 42-field metadata v2, knowledge graph, AI benchmark — no AI approximations. Batch 1 is master template batch — finalize one garment completely, then scale.

✓ Request received — v2.0 master template pipeline. We will review and respond to your email within 3 business days with 5-view renders, 42-field metadata, knowledge graph, benchmark. — Rehoteq Technologies

By submitting, you agree that dataset use is subject to attribution requirements and terms negotiated with Rehoteq Technologies. This is not automatic approval — all requests reviewed. Licensed downloads include exact renders like Cape Prophetess master template — 5 views frozen, no AI hallucination, knowledge graph with Mother Celestial* overlap handling, AI error taxonomy v2.