# C01 downloadable case kit
C01 ชุดกล้องตรวจสภาพ: รายละเอียดที่เพิ่มใช้พื้นที่เท่าไร. Authored synthetic data, p4-toy-1, 2026-10-12.
ผู้ตรวจต้องเปิดทวนรายละเอียดของเป้าหมายสมมติพร้อมบริบทและรายการส่วนที่ยังขาด เพื่อเลือกงานตรวจต่อ

1. Read [requirements.md](requirements.md) for baseline, proposed change, role prerequisites, all 12 KSEC criteria and evidence. Read the [chapter](https://drone.pk-research.work/knowledge/drone-applications/cases/inspection-camera/) for the worked explanation.
2. Choose B (brief and selection), O (resources and hold conditions) or D (interfaces, tests and restoration). Complete the K/S prerequisite check; follow remedial links if missing. E means Ethics and C Character.
3. Download [interfaces.csv](interfaces.csv); mark unknown explicitly and request missing sources. Roles are proposed responsibilities, actual owner not_assigned. No real equipment claims.
4. Preserve v0 and use [change-record.md](change-record.md). Write expected before any run using [test-plan.md](test-plan.md). samples = 4 ผ่าน sampling; total_bytes = capacity ผ่าน storage; 0 GSD ถูกปฏิเสธ; null GSD เป็น unknown
5. Get all four [shared lab files](../lab/readme.md): [calculator.py](../lab/calculator.py), [fixture.json](../lab/fixture.json), [expected.csv](../lab/expected.csv), [readme.md](../lab/readme.md). Keep these together. Follow its bounded local run and capture steps.
6. [results.csv](results.csv) contains the author's host calculation for this case. The [raw results](../lab/raw-results.json) and [host record](../lab/host-results.json) document 17 expected matches, identical replay and three matching rollback pairs (Python 3.14.3, 2026-10-12 16:20:43 UTC). For your own work, capture a separate actual record with matching case IDs, disposition, evidence_level, JSON metrics, physical_status and learner_status. Do not paste expected as actual. Author runs do not assess learner skills.
7. Compare all expected metrics, rerun identical input for byte equality, then restore the saved baseline fixture and verify hash/output. The rollback fixture scenario alone is not evidence that your own edits were restored.
8. Submit role artifacts v0/v1 under the same APP-C01-role-K/S/E/C01 IDs, with feedback, response, reason, owner, missing evidence and retest. Keep NOT_RUN until work exists; use meets / needs_revision / insufficient_evidence for reviewed work. CLO/PLO blank pending_design.

Baseline: CAM01: 12 mm / 4 mm/px = 3 samples; 30 × 8388608 = 251658240 bytes (240 MiB), capacity 268435456 (256 MiB).
Proposal: CAM02: GSD 2 mm/px, 60 images => 6 samples but 503316480 bytes (480 MiB) exceed 256 MiB. CAM03 raises authored capacity to 536870912 bytes (512 MiB).
ไม่มีภาพกล้องจริง ความคม แสง การสั่น ตำแหน่ง เวลาและคู่ภาพที่ตรวจแล้ว ผลเห็น/ตรวจพบ NOT_RUN
numeric_fit means only authored numeric conditions. unknown retains null, outside_bounds requires revision, offline_hold/stale_plan/expired/deadline_missed are data-case dispositions. Physical/flight/RF/field NOT_RUN; permissions/receiver NOT_CONFIRMED; learners NOT_RUN. P4 pending_verification; P3 pending_field_review specialist not_assigned; P5 preparation only, no real learners.