Prepare v0.2 metadata and documentation release

Archive_Specifications.md

@@ -1,19 +1,59 @@
 # Archive Specifications
 
-## Scope
+## Scope and Protocol Version
 
-These specifications apply to recordings and drawings created beginning July 15, 2025, unless otherwise noted in the metadata.
+These specifications apply to recordings and drawings created beginning July
+15, 2025 unless a metadata row documents a deviation.
 
-Individual metadata records document recording-specific variables and any deviations from these standard specifications.
+The corresponding metadata identifier is:
 
-## Recording Configuration
+```text
+acquisition_protocol_version = standard-2025-07-15
+```
 
-- Orientation: Landscape
-- Video resolution: 2160p UHD (4K)
-- Image size: 4000 × 3000
+The identifier records the governing protocol. It does not imply that every
+published file has identical encoded dimensions, orientation, codec, audio, or
+post-processing history.
+
+## Capture Configuration
+
+The documented standard acquisition intent is:
+
+- Capture orientation: Landscape
+- Capture resolution class: 2160p UHD (4K)
+- Still-image size: 4000 × 3000
 - Exposure: Locked
 - White balance: Locked
 
+### Capture versus published encoding
+
+Capture orientation and stored pixel orientation are separate properties.
+`metadata.csv` fields `width_px`, `height_px`, and `orientation` describe the
+published encoded file. Many canonical files are stored as portrait video,
+including records that historical evidence identifies as rotations or crops of
+landscape sources.
+
+Do not infer a rotation, crop, or transcode solely from dimensions. Such a
+transformation should be asserted only when supported by provenance evidence.
+
+## Timing and Calibration Status
+
+The current protocol documentation does not yet provide a complete calibrated
+time base for motor-control or physical-trajectory measurement. In particular,
+the following may be unknown for a recording unless separately documented:
+
+- original frame-capture interval
+- real drawing-session duration
+- time-lapse acceleration factor
+- source-frame timestamps
+- removed pauses or edit decisions
+- dropped-frame history
+- camera model, lens, and mount geometry
+- pixel-to-physical coordinate mapping or homography
+
+`encoded_frame_rate` is playback rate only. It must not be treated as original
+drawing-time sampling.
+
 ## Drawing Surfaces
 
 ### 9 × 12 in
@@ -37,6 +77,9 @@ Individual metadata records document recording-specific variables and any deviat
 - 20-ply
 - White
 
+Other support sizes and materials may appear when identified in the canonical
+filename and metadata.
+
 ## Drawing Instruments and Ink
 
 ### Ballpoint
@@ -46,11 +89,25 @@ Individual metadata records document recording-specific variables and any deviat
 - Medium, 1.0 mm
 - Document-proof ink conforming to ISO 12757-2
 
-### Fountain Pen
+### Fountain pen
 
 - Pilot Custom 823
 - Aurora Black fountain pen ink
 
-## Exceptions
+Additional tools and media are recorded at the individual-record level.
+
+## Audio
+
+Audio presence is a stored-file property, not a required part of the acquisition
+protocol. `has_audio` records whether an audio stream exists. It does not
+describe the stream's contents or establish redistribution rights for music,
+speech, notifications, or other incidental sound.
+
+## Exceptions and Evidence
+
+Known departures from the protocol should be documented in the affected
+metadata row. A deviation claim should identify its evidence in `notes` and,
+where applicable, provenance fields.
 
-Any departure from these standard specifications is documented in the metadata for the affected recording.
+When a property is not documented, it must remain unknown rather than being
+reconstructed from visual appearance alone.

CHANGELOG.md

@@ -0,0 +1,65 @@
+# Changelog
+
+## v0.2.0 — Release candidate
+
+Base repository revision:
+`339e6149de130cbe2d001f7a0e5247872bc788ec`
+
+### Metadata contract
+
+- Added persistent `record_id` values for all 338 canonical records.
+- Added `collection_tier`, `is_canonical`, and
+  `acquisition_protocol_version`.
+- Normalized canonical multi-value `tool` and `medium` fields from pipes to
+  semicolons.
+- Defined tool and medium fields as independent ordered unique lists rather
+  than unsupported positional pairs.
+- Preserved every canonical file path, content hash, technical property, and
+  provenance value from the pinned source index.
+- Removed one pre-standard metadata row whose corresponding media is not
+  distributed because the required ownership and provenance threshold was not
+  met. No identifying local path or content hash is published for the excluded
+  holding.
+- Added complete canonical rows for three previously unindexed Standard files.
+  Their local inbox SHA-256 values exactly match the current Hub LFS identities;
+  stream metadata was extracted from those matching bytes.
+
+### Documentation
+
+- Distinguished archive scope from currently published video coverage.
+- Documented that the default Hub `train` split is not an official experiment
+  split.
+- Distinguished capture orientation from stored encoded orientation.
+- Added intended-use, unsupported-use, bias, privacy, incidental-rights, audio,
+  access, citation, and reproducibility guidance.
+- Documented missing temporal and physical calibration required for rigorous
+  motor-control measurement.
+
+### Validation
+
+- Added `tools/repair_metadata.py` for the reproducible v0.1-to-v0.2 transform.
+- Added `tools/validate_metadata.py` for identity, schema, technical-field,
+  orientation, duplicate, and Viewer-path checks.
+- Retained the 465-row pinned Dataset Viewer snapshot as a historical base
+  audit input. Final revised membership is validated against the PR repository
+  tree because the Viewer has not yet rebuilt from the amended indexes.
+- Reconciled the repository tree with both active indexes: 338 indexed
+  canonical records and 129 published pre-standard records, both with
+  one-to-one row/file coverage.
+- Added `tools/metadata_additions_v0.2.csv` so the 338-row index regenerates
+  deterministically from the pinned 335-row source plus 3 audited additions.
+
+### Media payload
+
+- No video was downloaded, moved, deleted, renamed, transcoded, or replaced.
+- No audio stream was modified.
+- No historical relationship was inferred or promoted.
+
+### Deferred
+
+- Formal Hub configurations separating `canonical` and `pre_standard` remain
+  pending a Viewer-build test because canonical media and provenance holdings
+  share a directory subtree.
+- Technical ingestion of 129 pre-standard records remains incomplete.
+- Public ingestion of the 2012–2016 livestream layer remains pending.
+- No lightweight proxy-video release is included in this metadata-only change.

Pre_Standard_Time_Lapses/metadata.csv

@@ -105,7 +105,6 @@
 "2025-03-01.1658.Pencil.Graphite.FountainPen.Ink.Paper.11x14.mp4","2025-03-01","16:58","Pencil;FountainPen","Graphite;Ink","Paper","11x14",""
 "2025-03-04.0907.Pencil.Graphite.FountainPen.Ink.Paper.11x14.mp4","2025-03-04","09:07","Pencil;FountainPen","Graphite;Ink","Paper","11x14",""
 "2025-03-05.0904.FountainPen.Ink.Paper.18x24.mp4","2025-03-05","09:04","FountainPen","Ink","Paper","18x24",""
-"2025-03-05-00.UnknownTime.Pencil.Graphite.FountainPen.Ink.IllustrationBoard.30x40.mp4","2025-03-05","","Pencil;FountainPen","Graphite;Ink","IllustrationBoard","30x40",""
 "2025-03-14.2338.BallpointPen.Ink.Paper.18x24.mp4","2025-03-14","23:38","BallpointPen","Ink","Paper","18x24",""
 "2025-03-15.0904.FountainPen.Ink.Paper.18x24.mp4","2025-03-15","09:04","FountainPen","Ink","Paper","18x24",""
 "2025-03-17.0621.Pencil.Graphite.FountainPen.Ink.Paper.11x14.mp4","2025-03-17","06:21","Pencil;FountainPen","Graphite;Ink","Paper","11x14",""

README.md

@@ -1,156 +1,289 @@
 ---
 license: cc-by-nc-4.0
+pretty_name: Time-Lapse Artifacts
+annotations_creators:
+- no-annotation
+source_datasets:
+- original
+size_categories:
+- n<1K
 tags:
 - video
 - time-series
 - longitudinal-study
 - computer-vision
-- motor-control
+- drawing-process
+- self-supervised-learning
 - archival
 ---
 
 # time-lapse-artifacts
 
-A 14-Year Longitudinal Archive of Analog Drawing Process (2012–Present)
+An active longitudinal archive of analog drawing process.
 
-**Status:** Active construction release. The canonical standardized collection is indexed and usable; pre-standard and legacy holdings remain under incremental review and ingestion.
+**Archive scope:** 2012–Present  
+**Currently published video coverage:** September 2024–Present  
+**Release status:** Active construction release
 
-## Overview
+The archive scope describes the underlying recording practice and holdings. It
+does not mean that recordings from every year have been published. In the
+audited 2026-08-18 snapshot, published video rows begin in September 2024; the
+2012–2016 livestream layer has not yet been ingested as public video data.
+
+## Snapshot and Coverage
+
+This documentation was audited against repository commit
+`339e6149de130cbe2d001f7a0e5247872bc788ec`.
+
+| Collection tier | Published MP4s | Indexed rows | Date coverage | Metadata state |
+| --- | ---: | ---: | --- | --- |
+| `standard` | 338 direct files | 338 | 2025-07-15–2026-08-10 | Complete core technical metadata for every direct Standard file |
+| `pre_standard` | 129 | 129 | 2024-09-17–2025-07-13, plus 9 unknown dates | Filename-level descriptive metadata; technical ingestion incomplete |
+| `legacy_livestream` | 0 public videos | 0 | Historical scope 2012–2016 | Placeholder layer; ingestion pending |
 
-`time-lapse-artifacts` is a longitudinal video dataset documenting analog drawing process across multiple recording periods and acquisition conditions.
+At the pinned base revision, the Hub Viewer exposes 465 rows: 335 standard
+rows and 130 pre-standard metadata rows. Audit found that one pre-standard row
+referred to material that is not distributed because the required ownership
+and provenance threshold was not met. This release removes that row from the
+active public index. This release also indexes three previously unindexed
+Standard files, leaving 338 standard and 129 pre-standard rows and an expected
+467 active Viewer rows after rebuild.
+
+The builder-generated split name `train` is an archive-loading default,
+**not** a recommended machine-learning train split. No official
+train/validation/test partition has been defined.
+
+## Overview
 
-The repository contains standardized time-lapse recordings, earlier pre-standard time-lapses, and historical livestream material. Raw video is the primary record; metadata is used to identify, organize, compare, and document the recordings without imposing aesthetic classifications on the drawings.
+`time-lapse-artifacts` documents analog drawing processes across different
+recording periods and acquisition conditions. Raw video is the primary record.
+Metadata identifies, organizes, compares, and documents recordings without
+imposing aesthetic scores or stylistic labels.
 
-The project scope is 2012–Present. This describes the span of the underlying recording practice and archive; it should not be interpreted to mean that material from every year in that span has already been published here.
+The collection is a single-creator longitudinal archive. This makes it useful
+for studying within-practice change, but it does not represent artists,
+materials, studios, or drawing practices broadly.
 
 ## Repository Structure
 
 ### `Standard_Time_Lapses/`
 
-The canonical standardized records are the MP4 files stored directly in `Standard_Time_Lapses/`. The nested `Provenance/` subtree contains supporting records and retained historical exceptions and is not part of the canonical record set.
+The directory contains 338 direct MP4 files. `metadata.csv` is their complete
+machine-readable canonical index: every direct file has one validated row with
+core technical metadata. Direct placement alone is not sufficient for future
+additions; metadata and release validation are required.
 
-Files use a structured filename convention encoding documented session information such as date, finish time when known, tools, media, support, and dimensions.
+Each canonical metadata row now includes:
 
-`metadata.csv` is the machine-readable index for the published canonical records.
+- a persistent `record_id`
+- `collection_tier = standard`
+- `is_canonical = true`
+- `acquisition_protocol_version = standard-2025-07-15`
+- descriptive session fields
+- encoded video properties
+- repository content hash and size
+- provenance fields where documented
 
-Historical source relationships should not be inferred from filenames alone.
+Files use a structured filename convention described in `SCHEMA.md`. A
+filename is descriptive, but it is not the stable identity of a record;
+external references should use `record_id`.
 
 ### `Standard_Time_Lapses/Provenance/`
 
-Supporting records documenting the reconstruction and verification of historical source relationships.
+This subtree contains supporting historical records and retained exceptions.
+It is not part of the canonical record set.
 
-`2026-07-31/` preserves the completed provenance release, including the consolidated mapping between historical `Series_9x12` filenames and canonical standardized records.
+`2026-07-31/` preserves the completed provenance release, including mappings
+between historical `Series_9x12` filenames and canonical records. The former
+full duplicate video snapshot is not retained. Relationships are preserved
+through mappings, hashes, metadata, and repository history.
 
-The former full `Series_9x12` video snapshot is not retained as a second copy of the canonical dataset. Its verified relationships are preserved through mapping records, hashes, metadata, and repository history.
-
-`Unresolved_Series_9x12/` contains a very small number of distinct historical recordings retained because they are not duplicates of canonical files but currently lack sufficient documentation for promotion into the standardized collection. They should not be treated as canonical `Standard_Time_Lapses` records.
+`Unresolved_Series_9x12/` contains distinct historical recordings that do not
+yet have enough documentation for promotion into the canonical collection.
 
 ### `Pre_Standard_Time_Lapses/`
 
-Earlier time-lapse recordings that predate the standard acquisition specifications.
-
-These recordings are kept separate from `Standard_Time_Lapses/` because their capture conditions and available documentation may differ from the later standardized records.
+These 129 recordings predate the standard acquisition protocol. They are
+published for archival continuity but are not interchangeable with canonical
+standard records. Their technical metadata remains incomplete. Missing values
+must not be interpreted as zero, false, or evidence that a property was absent.
 
-Where metadata is incomplete, uncertainty is preserved rather than reconstructed from unsupported assumptions.
+Material that does not meet the archive's ownership and provenance threshold
+is excluded from both the published media and the active metadata index.
 
 ### `Legacy_Livestreams_2012_2016/`
 
-Historical real-time/livestream material from the earlier recording period.
-
-This material is maintained as a distinct historical layer rather than being mixed invisibly with the standardized time-lapse collection. Organization and ingestion of this material may remain incomplete while the repository is under construction.
+This directory reserves the historical livestream layer. At the audited
+snapshot it contains no public video data. References to 2012–2016 therefore
+describe archive scope, not current downloadable coverage.
 
 ## Metadata and Provenance
 
-The metadata is intended to distinguish between information directly documented by the recording, information established through historical source comparison, and information that remains unresolved.
-
-Depending on the record, provenance evidence may include:
+The metadata distinguishes documented facts, established historical
+relationships, and unresolved information. Evidence may include:
 
 - exact file/blob identity
-- original filenames
-- historical repository paths
+- original filenames and repository paths
 - file sizes and hashes
 - duration comparison
 - visual continuity across recordings
 - documented physical references
 - repository commit history
 
-Uncertain relationships are not promoted to verified relationships solely because filenames or dates appear similar.
+Uncertain relationships are not promoted to verified relationships solely
+because dates or filenames look similar.
 
-Historical filenames are preserved in metadata where relevant even when the canonical filename has been standardized.
+Multiple `tool` and `medium` values use semicolons. The two fields are
+independent observed-value lists; equal list lengths and positional pairing are
+not guaranteed. See `SCHEMA.md`.
 
-The canonical metadata also includes automatically extracted technical properties such as:
+`encoded_frame_rate` is encoded playback rate. It is not the original temporal
+sampling rate of the drawing process.
 
-- playback duration
-- encoded width and height
-- orientation
-- encoded frame rate
-- video codec
-- pixel format
-- container
-- audio presence
-- repository content hash
+## Acquisition and Stored-Video Orientation
 
-`encoded_frame_rate` describes the encoded video playback rate. It should not be interpreted as the original temporal sampling rate of the drawing process.
+The acquisition protocol effective July 15, 2025 is documented in
+`Archive_Specifications.md`.
 
-## Standard Acquisition Specifications
+The protocol's capture orientation and a file's stored pixel orientation are
+different concepts. The `orientation`, `width_px`, and `height_px` metadata
+describe the published encoded file. Historical canonicalization may include a
+rotation, crop, or transcode; those transformations are documented only where
+evidence currently exists and must not be inferred from dimensions alone.
 
-**Effective: July 15, 2025**
+## Annotation State
 
-Beginning July 15, 2025, recordings and drawings follow the standard acquisition specifications documented in `Archive_Specifications.md`, unless otherwise noted in the metadata.
+The videos are primarily raw and unannotated. The release does not provide:
 
-Individual metadata records document variables specific to each recording and any known deviations from those standard specifications.
+- frame-level bounding boxes or segmentation masks
+- hand, pen-tip, pose, or stroke labels
+- aesthetic or quality scores
+- imposed stylistic classifications
+- calibrated physical trajectories
+- original capture timestamps for individual time-lapse frames
 
-The date and time encoded in standardized filenames represent the documented **finish date and finish time** of the drawing session, not the recording start time.
+Researchers may derive annotations, but derived labels should identify the
+source `record_id` and pinned repository revision.
 
-Where no reliable finish time is available, the time remains explicitly unknown rather than being inferred.
+## Intended and Unsupported Uses
 
-## Canonical Status
+### Reasonable uses in the current release
 
-The MP4 files stored directly in `Standard_Time_Lapses/` constitute the canonical standardized collection. Files within `Standard_Time_Lapses/Provenance/` are supporting provenance material and are not canonical records.
+- archival and drawing-process research
+- qualitative within-creator longitudinal analysis over published coverage
+- self-supervised video representation experiments
+- frame-change, scene, and workspace analysis
+- development of annotation and tracking methods
 
-Historical source files, provenance releases, unresolved historical recordings, and pre-standard recordings are retained for documentation and comparison but are not interchangeable with canonical standardized records.
+### Uses requiring additional validation or calibration
 
-Repository organization may continue to evolve as historical material is reviewed. Reorganization does not alter the underlying video evidence, and provenance records are retained where they are necessary to explain earlier repository states.
+- hand or pen-tip tracking
+- optical-flow comparison across acquisition eras
+- temporal behavior analysis
+- real-world trajectory, velocity, acceleration, or motor-control measurement
 
-## Ownership
+The release does not currently provide capture intervals, acceleration factors,
+source-frame timestamps, camera calibration, or pixel-to-physical coordinate
+mapping. Encoded time-lapse playback must not be treated as real drawing time.
 
-The creator has manually reviewed the published collection and confirms ownership of all files and depicted drawings. A previously identified exception was removed from the repository and is not part of the published collection.
+### Not validated
 
-## Annotation State
+- population-level claims about artists or drawing behavior
+- clinical, biometric, authorship, or identity inference
+- a leakage-safe training or evaluation benchmark
+- commercial use outside the CC BY-NC 4.0 terms
 
-The video collection is primarily raw and unannotated.
+## Limitations and Sources of Bias
 
-There are no frame-level bounding boxes, segmentation masks, pose labels, aesthetic scores, or imposed stylistic classifications. Researchers may derive such annotations independently for their own applications.
+- The archive documents one creator and one evolving practice.
+- Current public video coverage begins in 2024, not 2012.
+- The standard and pre-standard tiers have substantially different metadata
+  completeness and acquisition conditions.
+- The workspace can contain hands, tools, phones, tablets, reference material,
+  and other objects that may create visual shortcuts or occlusion.
+- Stored resolution and frame rate do not establish physical or temporal
+  calibration.
+- Large files make exhaustive inspection and replication expensive.
+- No official leakage-safe split exists.
 
-The archive does not select recordings on the basis of perceived drawing quality. Recording integrity and documentation are treated separately from aesthetic evaluation.
+## Rights, Privacy, and Audio
 
-## Potential Uses
+The creator confirms ownership of the published video files and depicted
+creator-owned drawings. Material that cannot pass the ownership and provenance
+review is not part of the active public index and is not distributed by this
+release. This statement does not independently grant rights in incidental
+third-party material that may appear on screens, in references, or in captured
+audio.
 
-The dataset may support research involving, among other areas:
+Most canonical masters retain an audio stream. Users are responsible for
+reviewing audio, visible screens, references, reflections, tattoos, and other
+potentially identifying or third-party content before redistribution or model
+release. A future proxy release should remove audio by default while preserving
+the archival masters.
 
-- computer vision
-- temporal and longitudinal analysis
-- self-supervised video learning
-- motion and motor-control analysis
-- drawing-process analysis
-- optical flow and trajectory extraction
-- representation learning
-- change over long time spans
-- comparison of acquisition conditions and recording eras
+## Access and Reproducibility
 
-These are possible uses rather than claims that the dataset has been optimized or validated for any particular downstream task.
+The archive is approximately 1.86 TB. Canonical files are often several
+gigabytes each, so users should select records through `metadata.csv` before
+downloading video.
+
+For reproducible work:
+
+1. Pin a repository commit or release tag.
+2. Record the selected `record_id` values.
+3. Preserve the published `hub_xet_hash` values.
+4. State whether audio was retained.
+5. Document all rotation, crop, sampling, decoding, and proxy-generation steps.
+
+Run the included validator before publishing a metadata change:
+
+```bash
+python tools/validate_metadata.py metadata.csv
+```
+
+The v0.2 index can be regenerated from the pinned 335-row source index plus the
+three audited additions:
+
+```bash
+python tools/repair_metadata.py source_metadata.csv metadata.csv \
+  --additions tools/metadata_additions_v0.2.csv
+```
+
+## Versioning
+
+The repository is under active construction. Record identity is carried by
+`record_id`; a filename may be corrected without creating a new conceptual
+record. Content replacement must retain provenance and should receive a new
+content hash.
+
+See `CHANGELOG.md` for release-level changes.
 
 ## Documentation
 
-See:
+- `metadata.csv` — canonical machine-readable index
+- `Pre_Standard_Time_Lapses/metadata.csv` — active pre-standard public index
+- `SCHEMA.md` — filename and metadata contracts
+- `Archive_Specifications.md` — acquisition specifications and known gaps
+- `VALIDATION.md` — validation scope and current results
+- `Standard_Time_Lapses/Provenance/` — historical evidence and mappings
 
-- `metadata.csv` — machine-readable dataset index
-- `SCHEMA.md` — filename and metadata conventions
-- `Archive_Specifications.md` — standard acquisition specifications
-- `Standard_Time_Lapses/Provenance/` — historical mapping and provenance records
+## Citation
 
-## License
+Until a DOI-backed release is available, cite the dataset and pinned revision:
+
+```bibtex
+@dataset{maxwellinked_time_lapse_artifacts_2026,
+  author    = {maxwellinked},
+  title     = {time-lapse-artifacts},
+  year      = {2026},
+  publisher = {Hugging Face},
+  url       = {https://huggingface.co/datasets/maxwellinked/time-lapse-artifacts},
+  note      = {Active construction release; include the repository revision used}
+}
+```
 
-This dataset is released under **CC BY-NC 4.0**.
+## License
 
-See the repository license metadata for the applicable terms.
+The dataset is released under **CC BY-NC 4.0**. The license applies only to
+material for which the licensor holds the necessary rights. Users must conduct
+their own review of incidental third-party content and downstream use.

RELEASE_MANIFEST.json

@@ -0,0 +1,81 @@
+{
+    "release_candidate":  "v0.2.0",
+    "base_revision":  "339e6149de130cbe2d001f7a0e5247872bc788ec",
+    "amended_pr":  2,
+    "created_date":  "2026-08-18",
+    "master_videos_modified":  false,
+    "excluded_media_uploaded":  false,
+    "active_index_counts":  {
+                                "canonical":  338,
+                                "pre_standard":  129,
+                                "total":  467
+                            },
+    "repository_inventory":  {
+                                 "direct_standard_mp4":  338,
+                                 "standard_pending_metadata":  0,
+                                 "direct_pre_standard_mp4":  129,
+                                 "provenance_mp4":  2
+                             },
+    "files":  [
+                  {
+                      "path":  ".gitattributes",
+                      "size_bytes":  3136,
+                      "sha256":  "a9c5b148f359c60350c3ac47566271b418beb657c21b0ee3dba990fb5807b748"
+                  },
+                  {
+                      "path":  "Archive_Specifications.md",
+                      "size_bytes":  3229,
+                      "sha256":  "ecebd17d5cb534b2b867846b471dcd4552108e563c0d8d4428b730785ca7df72"
+                  },
+                  {
+                      "path":  "CHANGELOG.md",
+                      "size_bytes":  2941,
+                      "sha256":  "fa9cc8a55b577c098adf9bc48f906a3d3d20644be1b180ca188036f0f2dac9a5"
+                  },
+                  {
+                      "path":  "metadata.csv",
+                      "size_bytes":  176413,
+                      "sha256":  "ecad3455dea0c65b54a201e226aaf527f52607abab3582f49f065758c434e0aa"
+                  },
+                  {
+                      "path":  "Pre_Standard_Time_Lapses/metadata.csv",
+                      "size_bytes":  17197,
+                      "sha256":  "714029e7acde57ffad0288499cdd4b2784cd585a17ae49710970a929d96eb3cf"
+                  },
+                  {
+                      "path":  "README.md",
+                      "size_bytes":  11297,
+                      "sha256":  "1a34bea1924882a5b4156f4602e05fc96e5c9da465aa25c645fc2a538bf88584"
+                  },
+                  {
+                      "path":  "SCHEMA.md",
+                      "size_bytes":  8179,
+                      "sha256":  "191c3fae0f67d1abb31d6639c4f720ae6610e8ffb5150ca45fed0591f3c97ce7"
+                  },
+                  {
+                      "path":  "tools/metadata_additions_v0.2.csv",
+                      "size_bytes":  1879,
+                      "sha256":  "9227bcd891afdcaa9e8547e34b0a573d45109f7035b0baf645ee03ba757b069c"
+                  },
+                  {
+                      "path":  "tools/repair_metadata.py",
+                      "size_bytes":  5115,
+                      "sha256":  "1be4ed29793f1f6a734e496341acd9700ef87559c3ba5008b8671cd46c80978e"
+                  },
+                  {
+                      "path":  "tools/validate_metadata.py",
+                      "size_bytes":  14924,
+                      "sha256":  "06a9662b28c591c2f30d869f0c1c666bfd60344fffee8efbd8988012fc2fe3a8"
+                  },
+                  {
+                      "path":  "VALIDATION.md",
+                      "size_bytes":  3240,
+                      "sha256":  "f41bab27192abf4fa273df0d6111a75874bdf65ff37466242d5f260530e3ed1d"
+                  },
+                  {
+                      "path":  "VIEWER_MIGRATION.md",
+                      "size_bytes":  2329,
+                      "sha256":  "9f33fda262f34afbf76fa4738a4269952b36ac61c7c572d85ec4546f04ffd8b8"
+                  }
+              ]
+}

SCHEMA.md

@@ -1,43 +1,199 @@
-# Filename Schema
+# Schema
 
-## Format
+This document defines the contracts for canonical filenames and
+`metadata.csv`. Filenames are human-readable descriptions; `record_id` is the
+persistent machine identity.
+
+## Canonical Filename Format
 
 `YYYY-MM-DD.HHMM.Tool.Medium.Support.Dimensions[.RecNN].ext`
 
-Example:
+Examples:
 
-`2025-07-15.0448.FountainPen.Ink.Paper.9x12.mp4`
+```text
+2025-07-15.0448.FountainPen.Ink.Paper.9x12.mp4
+2026-06-29.UnknownTime.FountainPen.Ink.Paper.30x40.Rec01.mp4
+2026-06-29.UnknownTime.FountainPen.Ink.Paper.30x40.Rec02.mp4
+```
 
-Multiple recordings requiring distinction:
+### Filename fields
 
-`2026-06-29.UnknownTime.FountainPen.Ink.Paper.30x40.Rec01.mp4`
+- `YYYY-MM-DD` — documented finish date.
+- `HHMM` — documented finish time in 24-hour local civil time.
+- `UnknownTime` — used when no reliable finish time is documented.
+- `Tool` — drawing instrument, such as `Pencil`, `FountainPen`, `BallpointPen`,
+  or `Brush`.
+- `Medium` — applied material, such as `Graphite` or `Ink`.
+- `Support` — drawing surface, such as `Paper`, `IllustrationBoard`, or
+  `WatercolorPaper`.
+- `Dimensions` — support dimensions in inches; order preserves support
+  orientation, such as `9x12` or `12x9`.
+- `RecNN` — optional recording-sequence identifier used only when otherwise
+  identical filenames require distinction. Numbering begins with `Rec01`.
+- `ext` — file extension.
 
-`2026-06-29.UnknownTime.FountainPen.Ink.Paper.30x40.Rec02.mp4`
+### Filename rules
 
-`2026-06-29.UnknownTime.FountainPen.Ink.Paper.30x40.Rec03.mp4`
+- Fields are separated by periods.
+- Style and aesthetic classifications are excluded.
+- `TL` is excluded because time-lapse is implicit in the primary collection.
+- `0000` is reserved for a documented midnight finish time.
+- `RecNN` identifies recording sequence only. It does not by itself establish a
+  separate drawing, session, or completion state.
+- `UnknownTime` records uncertainty; it is not replaced by an inferred time.
+- Paper-surface variants such as `HotPress` and `ColdPress` may remain encoded
+  in filenames until a dedicated metadata field is added.
+- Historical relationships must not be inferred from filename similarity
+  alone.
 
-## Fields
+Tool and medium tokens in a filename describe documented materials, but older
+filenames do not always encode an unambiguous one-to-one pairing. Metadata must
+therefore preserve tool and medium sets without manufacturing positional pairs.
 
-- `YYYY-MM-DD` — documented finish date.
-- `HHMM` — documented finish time in 24-hour format.
-- `UnknownTime` — used only when the finish time is unknown.
-- `Tool` — drawing instrument, such as `FountainPen` or `BallpointPen`.
-- `Medium` — material applied, such as `Graphite` or `Ink`.
-- `Support` — physical drawing surface, such as `Paper`, `IllustrationBoard`, or `WatercolorPaper`.
-- `Dimensions` — support dimensions in inches; order preserves orientation, such as `9x12` or `12x9`.
-- `RecNN` — optional recording sequence identifier used when multiple distinct recordings would otherwise produce the same filename; numbering begins with `Rec01`.
-- `ext` — original file extension.
+## Canonical Metadata Contract
 
-## Rules
+`metadata.csv` indexes the validated canonical subset of MP4 files stored
+directly in `Standard_Time_Lapses/`. Direct placement alone does not make a file
+canonical: every released canonical file must have exactly one metadata row,
+complete required technical fields, and the necessary rights and provenance
+clearance. The v0.2 index contains 338 validated records and covers every MP4
+stored directly in `Standard_Time_Lapses/` at the pinned revision.
 
-- Fields are separated by periods.
-- Style classifications are excluded.
-- `TL` is excluded because time-lapse is implicit in the dataset's primary collection.
-- `0000` is reserved for an actual midnight finish time.
-- `RecNN` is used only when required to distinguish multiple recordings with otherwise identical filename fields.
-- `RecNN` identifies recording sequence only; it does not imply a separate drawing, session, or completion state.
-- When independently documented, original filenames are preserved exactly in `metadata.csv`; otherwise `original_filename` remains blank.
-- In `metadata.csv`, multiple tools and media are stored as ordered semicolon-separated values.
-- Example: `Pencil.Graphite.FountainPen.Ink` becomes `tool = Pencil;FountainPen` and `medium = Graphite;Ink`.
-- Each tool corresponds by position to the medium in the same position.
-- Paper surface variants such as `HotPress` and `ColdPress` remain preserved in `file_name`; the current `metadata.csv` schema does not yet include a dedicated `press_type` field.
+The canonical index does not make provenance, unresolved, pre-standard, or
+legacy files canonical.
+
+### Identity and collection fields
+
+| Field | Type | Meaning |
+| --- | --- | --- |
+| `file_name` | string | Repository-relative canonical video path. |
+| `record_id` | string | Persistent identity in the form `tla-` plus 16 lowercase hexadecimal characters. |
+| `collection_tier` | enum | `standard` for this index. |
+| `is_canonical` | boolean | `true` for every row in this index. |
+| `acquisition_protocol_version` | string | Protocol identifier; currently `standard-2025-07-15`. |
+
+The initial `record_id` values were generated from a namespaced SHA-256 of the
+canonical path. Once assigned, the stored value is authoritative and must be
+retained across filename corrections. New tooling must assign an ID only when a
+record does not already have one.
+
+### Session and material fields
+
+| Field | Type | Meaning |
+| --- | --- | --- |
+| `date` | `YYYY-MM-DD` | Documented finish date. |
+| `time` | `HH:MM` or blank | Documented finish time; blank means unknown. |
+| `tool` | semicolon list | Ordered unique set of documented tools. |
+| `medium` | semicolon list | Ordered unique set of documented media. |
+| `support` | string | Physical drawing surface. |
+| `dimensions` | string | Support dimensions in inches, such as `9x12`. |
+
+Multiple values use semicolons:
+
+```text
+tool = Pencil;FountainPen
+medium = Graphite;Ink
+```
+
+`tool` and `medium` are independent observed-value lists. Equal lengths and
+positional correspondence are **not guaranteed**. Repeated media are collapsed
+to one value. A future paired representation should use a structured field and
+must be based on documented evidence rather than token position.
+
+### Encoded-video fields
+
+| Field | Type | Meaning |
+| --- | --- | --- |
+| `duration_seconds` | float | Encoded playback duration. |
+| `width_px` | integer | Stored encoded width. |
+| `height_px` | integer | Stored encoded height. |
+| `orientation` | enum | Stored pixel orientation derived from width and height. |
+| `encoded_frame_rate` | integer | Encoded playback frame rate. |
+| `video_codec` | string | Encoded video codec. |
+| `pixel_format` | string | Encoded pixel format. |
+| `container` | string | Media container. |
+| `has_audio` | boolean | Whether the stored file contains an audio stream. |
+| `hub_xet_hash` | string | Repository content identity reported for the stored file. |
+| `size_bytes` | integer | Stored file size in bytes. |
+
+These fields describe the published encoded file. They do not establish the
+original frame-capture interval, real drawing duration, speed-up factor, camera
+calibration, capture orientation, rotation, crop, or transcode lineage.
+
+`encoded_frame_rate` must not be interpreted as drawing-time sampling rate.
+
+### Historical provenance fields
+
+| Field | Type | Meaning |
+| --- | --- | --- |
+| `original_filename` | string or blank | Independently documented historical filename. |
+| `provenance_status` | string or blank | Current relationship status. |
+| `legacy_provenance_path` | string or blank | Verified historical repository path. |
+| `legacy_related_references` | string or blank | Related candidates, aliases, or rejected references. |
+| `evidence_method` | string or blank | Primary evidence used for the status. |
+| `provenance_reference_duration_delta_seconds` | float or blank | Duration delta to a referenced historical file. |
+| `notes` | string or blank | Human-readable evidence or qualification. |
+
+Blank provenance values mean that the field is not documented or not
+applicable in the current release. A blank must not be promoted to a verified
+relationship.
+
+Observed provenance statuses include:
+
+- `legacy_match_verified`
+- `legacy_aliases_verified`
+- `legacy_duplicate_removed`
+- `legacy_candidates_rejected`
+- `direct_root_upload`
+
+Observed evidence methods include:
+
+- `blob_identity`
+- `visual_continuity`
+- `visual_comparison`
+- `commit_history`
+- `manual_confirmation`
+- `blob_identity_and_physical_date`
+
+New enumerated values must be documented here before use.
+
+## Missing-Value Rules
+
+- Blank means unknown, not documented, or not applicable according to the field
+  description.
+- Blank numeric values must remain blank; they are not zero.
+- Blank booleans must remain blank; they are not `false`.
+- Uncertain dates and times must not be reconstructed from unsupported clues.
+- Pre-standard and legacy metadata may be less complete and must carry a
+  collection-tier distinction when formally indexed.
+
+## Required Validation
+
+Before publication, the validator must confirm:
+
+- unique `record_id`, `file_name`, and nonblank `hub_xet_hash`
+- canonical paths remain directly under `Standard_Time_Lapses/`
+- one consistent semicolon convention for multi-value fields
+- valid dates, times, and dimensions
+- complete core encoded-video metadata for canonical rows
+- orientation agrees with stored width and height
+- canonical metadata paths match the canonical Viewer records
+- every active standard and pre-standard metadata path resolves to a published
+  repository file
+- repository media absent from an active index is reported explicitly rather
+  than silently treated as released data
+- material lacking the necessary ownership or provenance clearance is absent
+  from both active indexes
+
+Run:
+
+```bash
+python tools/validate_metadata.py metadata.csv
+```
+
+Rebuild the v0.2 index from its pinned 335-row source and audited additions:
+
+```bash
+python tools/repair_metadata.py source_metadata.csv metadata.csv \
+  --additions tools/metadata_additions_v0.2.csv
+```

VALIDATION.md

@@ -0,0 +1,95 @@
+# Validation Report
+
+## Scope
+
+Release candidate metadata was compared with:
+
+- source commit `339e6149de130cbe2d001f7a0e5247872bc788ec`
+- all 335 rows in the pinned source `metadata.csv`
+- 3 previously unindexed direct Standard files
+- all 465 rows in the corresponding base Dataset Viewer snapshot
+- the complete PR-side repository inventory
+- the revised 129-row active pre-standard index
+
+The full archive was not downloaded. For the three added Standard records, the
+local inbox files were read without modification to extract container and
+stream metadata, and their complete SHA-256 hashes were verified against the
+current Hub LFS identities. The excluded local holding was not uploaded, copied
+into the release candidate, or represented by an active metadata row.
+
+## Result
+
+**PASS — zero structural validation errors**
+
+| Check | Result |
+| --- | ---: |
+| Canonical metadata rows | 338 |
+| Metadata columns | 29 |
+| Unique `record_id` values | 338 |
+| Unique canonical paths | 338 |
+| Unique nonblank content hashes | 338 |
+| Canonical repository paths matched | 338/338 |
+| Direct standard MP4 files | 338 |
+| Standard files pending metadata ingestion | 0 |
+| Published pre-standard MP4 files | 129 |
+| Active pre-standard metadata rows | 129 |
+| Missing active metadata targets | 0 |
+| Pipe separators remaining in canonical tool/media fields | 0 |
+| Stored-orientation conflicts | 0 |
+| Duplicate canonical content hashes | 0 |
+
+## Audited statistics
+
+- Canonical playback duration: 40.46 hours
+- Median canonical playback duration: 340.955 seconds
+- Canonical indexed size: 1.647 TB decimal
+- Stored orientation: 336 portrait, 2 landscape
+- Audio streams: 334 present, 4 absent
+- Unknown canonical finish times: 33
+
+## Outstanding warnings
+
+1. All 129 active pre-standard rows lack duration and core technical metadata.
+2. Most canonical masters retain audio; rights and privacy review remains
+   necessary.
+3. Encoded frame rate is playback rate and cannot be used as original
+   drawing-time sampling.
+4. The 2012–2016 layer has no public video payload at this snapshot.
+5. No leakage-safe train/validation/test partition has been defined.
+
+## Reproduction
+
+Generate the candidate canonical index:
+
+```bash
+python tools/repair_metadata.py source_metadata.csv metadata.csv \
+  --additions tools/metadata_additions_v0.2.csv
+```
+
+Validate the canonical index:
+
+```bash
+python tools/validate_metadata.py metadata.csv
+```
+
+Use a Viewer snapshot only when it was built from the same metadata revision:
+
+```bash
+python tools/validate_metadata.py metadata.csv --viewer-json viewer_rows.json
+```
+
+For a repository-inventory check, also provide the active pre-standard index
+and a JSON list of repository paths:
+
+```bash
+python tools/validate_metadata.py metadata.csv \
+  --pre-standard-metadata Pre_Standard_Time_Lapses/metadata.csv \
+  --repo-files-json repo_files.json
+```
+
+The available `viewer_rows.json` records the pinned base Viewer state. It is an
+historical audit input, not part of the proposed upload, and is expected to
+differ from the revised indexes until the Viewer rebuilds.
+
+`repo_files.json` is an audit input captured from the pinned PR repository
+inventory; it is not part of the proposed repository upload.

VIEWER_MIGRATION.md

@@ -0,0 +1,65 @@
+# Dataset Viewer Migration
+
+## Goal
+
+Expose canonical and pre-standard recordings as separate Hub configurations so
+users do not mistake metadata-incomplete historical material for canonical
+standard data.
+
+## Why this is not applied in v0.2.0
+
+The current repository places canonical MP4 files directly in
+`Standard_Time_Lapses/` and supporting provenance holdings beneath
+`Standard_Time_Lapses/Provenance/`. A broad `data_dir` configuration could
+therefore ingest noncanonical provenance videos.
+
+Hugging Face supports multiple configurations and folder-local video metadata,
+but the exact direct-file glob and metadata discovery behavior should be tested
+through a Hub pull request before changing the public default configuration.
+
+## Proposed pull-request test
+
+1. Add a folder-local canonical metadata file whose `file_name` values are
+   relative to `Standard_Time_Lapses/`.
+2. Add a folder-local pre-standard metadata file with explicit
+   `collection_tier = pre_standard` and `is_canonical = false`.
+3. Add two README configurations:
+   - `canonical` — default, direct MP4 files only
+   - `pre_standard` — separate historical subset
+4. Use a direct-file glob for canonical MP4 files so the `Provenance/` subtree
+   is excluded.
+5. Confirm the PR preview reports the exact active-index counts, currently 338
+   canonical and 129 pre-standard rows, with zero provenance videos and zero
+   unindexed direct files.
+6. Confirm folder-local metadata is attached to every expected row.
+7. Merge only after Viewer and Parquet generation succeed.
+
+## Candidate YAML for PR testing
+
+This block is a test proposal, not the current card configuration:
+
+```yaml
+configs:
+- config_name: canonical
+  default: true
+  data_dir: Standard_Time_Lapses
+  data_files:
+  - split: archive
+    path: "*.mp4"
+  drop_labels: true
+- config_name: pre_standard
+  data_dir: Pre_Standard_Time_Lapses
+  data_files:
+  - split: archive
+    path: "*.mp4"
+  drop_labels: true
+```
+
+If folder-local metadata is not discovered with the direct-file glob, do not
+broaden the glob. Instead, create explicit Parquet manifests for the two
+configurations so row membership is unambiguous.
+
+## Split terminology
+
+Use `archive` rather than `train` if supported by the Viewer build. Neither
+configuration should imply an official machine-learning evaluation split.

tools/metadata_additions_v0.2.csv

@@ -0,0 +1,4 @@
+"file_name","date","time","tool","medium","support","dimensions","duration_seconds","width_px","height_px","orientation","encoded_frame_rate","video_codec","pixel_format","container","has_audio","hub_xet_hash","original_filename","provenance_status","legacy_provenance_path","legacy_related_references","evidence_method","provenance_reference_duration_delta_seconds","size_bytes","notes"
+"Standard_Time_Lapses/2026-08-09.1601.BallpointPen.Ink.Paper.9x12.mp4","2026-08-09","16:01","BallpointPen","Ink","Paper","9x12","954.523","2160","3840","portrait","60","h264","yuv420p","mp4","True","197b061a90e85ff746383c0c5418e192520b66121580a0d4563ce39c0fd05b0c","2026-08-09.1601.BallpointPen.Ink.Paper.9x12.mp4","direct_root_upload","","","commit_history","","12486472605","Added directly in Hub commit 7bc09318; complete payload replaced an incomplete upload in commit 87efa084; local inbox SHA-256 matches the current Hub LFS identity."
+"Standard_Time_Lapses/2026-08-10.0921.BallpointPen.Ink.Paper.9x12.mp4","2026-08-10","09:21","BallpointPen","Ink","Paper","9x12","1131.523","2160","3840","portrait","60","h264","yuv420p","mp4","True","00092a669ea61049135e9764c6e67f6d99b0dd7f65d8d20ec932917906a92862","2026-08-10.0921.BallpointPen.Ink.Paper.9x12.mp4","direct_root_upload","","","commit_history","","14803829762","Added directly in Hub commit 7bc09318; local inbox SHA-256 matches the current Hub LFS identity."
+"Standard_Time_Lapses/2026-08-10.1201.BallpointPen.Ink.Paper.9x12.mp4","2026-08-10","12:01","BallpointPen","Ink","Paper","9x12","81.566","2160","3840","portrait","60","h264","yuv420p","mp4","True","c6bca64dfcda4d8a63e4827e9a56ca9e1c87faef31646aa042389c84655c38d5","2026-08-10.1201.BallpointPen.Ink.Paper.9x12.mp4","direct_root_upload","","","commit_history","","1087300262","Added directly in Hub commit 7bc09318; local inbox SHA-256 matches the current Hub LFS identity."

tools/repair_metadata.py

@@ -0,0 +1,144 @@
+#!/usr/bin/env python3
+"""Create the v0.2 canonical metadata index from the pinned v0.1 index.
+
+The transform is intentionally conservative. It adds fields that can be
+derived without reviewing video content, normalizes multi-value separators,
+and does not infer missing provenance or acquisition facts. Validated records
+published after the pinned source index can be supplied through an explicit
+additions CSV with the same source schema.
+"""
+
+from __future__ import annotations
+
+import argparse
+import csv
+import hashlib
+import json
+from pathlib import Path
+
+
+ADDED_FIELDS = [
+    "record_id",
+    "collection_tier",
+    "is_canonical",
+    "acquisition_protocol_version",
+]
+
+
+def stable_record_id(file_name: str) -> str:
+    digest = hashlib.sha256(
+        f"time-lapse-artifacts:{file_name}".encode("utf-8")
+    ).hexdigest()
+    return f"tla-{digest[:16]}"
+
+
+def normalize_multi_value(value: str) -> str:
+    """Normalize legacy pipe/semicolon lists as ordered unique values."""
+    if not value:
+        return ""
+    seen: set[str] = set()
+    values: list[str] = []
+    for part in value.replace("|", ";").split(";"):
+        item = part.strip()
+        if item and item not in seen:
+            seen.add(item)
+            values.append(item)
+    return ";".join(values)
+
+
+def repair(
+    input_csv: Path,
+    output_csv: Path,
+    additions_csv: Path | None = None,
+) -> dict[str, object]:
+    with input_csv.open("r", encoding="utf-8-sig", newline="") as handle:
+        reader = csv.DictReader(handle)
+        if reader.fieldnames is None:
+            raise ValueError("metadata file has no header")
+        source_fields = list(reader.fieldnames)
+        rows = list(reader)
+
+    source_row_count = len(rows)
+    additions_row_count = 0
+    if additions_csv:
+        with additions_csv.open("r", encoding="utf-8-sig", newline="") as handle:
+            additions_reader = csv.DictReader(handle)
+            if additions_reader.fieldnames is None:
+                raise ValueError("additions file has no header")
+            additions_fields = list(additions_reader.fieldnames)
+            if additions_fields != source_fields:
+                raise ValueError("additions columns must exactly match the source schema")
+            additions = list(additions_reader)
+        additions_row_count = len(additions)
+        rows.extend(additions)
+
+    required = {"file_name", "tool", "medium", "hub_xet_hash"}
+    missing = sorted(required.difference(source_fields))
+    if missing:
+        raise ValueError(f"metadata file is missing required fields: {missing}")
+
+    output_fields = [source_fields[0], *ADDED_FIELDS, *source_fields[1:]]
+    changed_tool = 0
+    changed_medium = 0
+
+    for row in rows:
+        old_tool = row.get("tool", "")
+        old_medium = row.get("medium", "")
+        row["tool"] = normalize_multi_value(old_tool)
+        row["medium"] = normalize_multi_value(old_medium)
+        changed_tool += row["tool"] != old_tool
+        changed_medium += row["medium"] != old_medium
+
+        row["record_id"] = stable_record_id(row["file_name"])
+        row["collection_tier"] = "standard"
+        row["is_canonical"] = "true"
+        row["acquisition_protocol_version"] = "standard-2025-07-15"
+
+    record_ids = [row["record_id"] for row in rows]
+    file_names = [row["file_name"] for row in rows]
+    hashes = [row["hub_xet_hash"] for row in rows if row["hub_xet_hash"]]
+    if len(record_ids) != len(set(record_ids)):
+        raise ValueError("generated record_id values are not unique")
+    if len(file_names) != len(set(file_names)):
+        raise ValueError("file_name values are not unique")
+    if len(hashes) != len(set(hashes)):
+        raise ValueError("hub_xet_hash values are not unique")
+
+    output_csv.parent.mkdir(parents=True, exist_ok=True)
+    with output_csv.open("w", encoding="utf-8", newline="") as handle:
+        writer = csv.DictWriter(handle, fieldnames=output_fields, lineterminator="\n")
+        writer.writeheader()
+        writer.writerows(rows)
+
+    return {
+        "rows": len(rows),
+        "source_rows": source_row_count,
+        "addition_rows": additions_row_count,
+        "columns_before": len(source_fields),
+        "columns_after": len(output_fields),
+        "tool_values_normalized": changed_tool,
+        "medium_values_normalized": changed_medium,
+        "unique_record_ids": len(set(record_ids)),
+        "unique_file_names": len(set(file_names)),
+        "unique_nonempty_hashes": len(set(hashes)),
+    }
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("input_csv", type=Path)
+    parser.add_argument("output_csv", type=Path)
+    parser.add_argument("--additions", type=Path)
+    parser.add_argument("--summary", type=Path)
+    args = parser.parse_args()
+
+    summary = repair(args.input_csv, args.output_csv, args.additions)
+    rendered = json.dumps(summary, indent=2, sort_keys=True)
+    if args.summary:
+        args.summary.parent.mkdir(parents=True, exist_ok=True)
+        args.summary.write_text(rendered + "\n", encoding="utf-8")
+    print(rendered)
+
+
+if __name__ == "__main__":
+    main()

tools/validate_metadata.py

@@ -0,0 +1,368 @@
+#!/usr/bin/env python3
+"""Validate the v0.2 active indexes and optional Viewer/repository snapshots."""
+
+from __future__ import annotations
+
+import argparse
+import csv
+import json
+import re
+import statistics
+import sys
+from collections import Counter
+from pathlib import Path
+from urllib.parse import unquote, urlparse
+
+
+RECORD_ID_RE = re.compile(r"^tla-[0-9a-f]{16}$")
+DATE_RE = re.compile(r"^\d{4}-\d{2}-\d{2}$")
+TIME_RE = re.compile(r"^(?:[01]\d|2[0-3]):[0-5]\d$")
+DIMENSIONS_RE = re.compile(r"^\d+(?:\.\d+)?x\d+(?:\.\d+)?$")
+DIRECT_STANDARD_RE = re.compile(r"^Standard_Time_Lapses/[^/]+\.mp4$")
+DIRECT_PRE_STANDARD_RE = re.compile(r"^Pre_Standard_Time_Lapses/[^/]+\.mp4$")
+
+REQUIRED_FIELDS = [
+    "file_name",
+    "record_id",
+    "collection_tier",
+    "is_canonical",
+    "acquisition_protocol_version",
+    "date",
+    "time",
+    "tool",
+    "medium",
+    "support",
+    "dimensions",
+    "duration_seconds",
+    "width_px",
+    "height_px",
+    "orientation",
+    "encoded_frame_rate",
+    "video_codec",
+    "pixel_format",
+    "container",
+    "has_audio",
+    "hub_xet_hash",
+    "size_bytes",
+]
+
+PRE_STANDARD_REQUIRED_FIELDS = [
+    "file_name",
+    "date",
+    "time",
+    "tool",
+    "medium",
+    "support",
+    "dimensions",
+    "original_filename",
+]
+
+
+def viewer_path(row: dict[str, object]) -> str:
+    video = row.get("video")
+    if not isinstance(video, dict) or not isinstance(video.get("src"), str):
+        return ""
+    path = unquote(urlparse(video["src"]).path)
+    marker = "/resolve/"
+    if marker not in path:
+        return ""
+    after_revision = path.split(marker, 1)[1]
+    parts = after_revision.split("/", 1)
+    return parts[1] if len(parts) == 2 else ""
+
+
+def load_csv(path: Path) -> tuple[list[str], list[dict[str, str]]]:
+    with path.open("r", encoding="utf-8-sig", newline="") as handle:
+        reader = csv.DictReader(handle)
+        if reader.fieldnames is None:
+            raise ValueError("metadata file has no header")
+        return list(reader.fieldnames), list(reader)
+
+
+def validate(
+    metadata_csv: Path,
+    viewer_json: Path | None,
+    pre_standard_metadata: Path | None,
+    repo_files_json: Path | None,
+) -> dict[str, object]:
+    fields, rows = load_csv(metadata_csv)
+    errors: list[str] = []
+    warnings: list[str] = []
+    pre_standard_rows: list[dict[str, str]] = []
+    repo_counts: dict[str, int] = {}
+    repo_unindexed_standard: list[str] = []
+    repo_unindexed_pre_standard: list[str] = []
+
+    missing_fields = [field for field in REQUIRED_FIELDS if field not in fields]
+    if missing_fields:
+        errors.append(f"missing required columns: {missing_fields}")
+
+    def duplicate_values(field: str) -> list[str]:
+        values = [row.get(field, "") for row in rows if row.get(field, "")]
+        return sorted(value for value, count in Counter(values).items() if count > 1)
+
+    for field in ("file_name", "record_id", "hub_xet_hash"):
+        duplicates = duplicate_values(field)
+        if duplicates:
+            errors.append(f"duplicate {field} values: {duplicates[:5]}")
+
+    for index, row in enumerate(rows, start=2):
+        label = row.get("file_name") or f"CSV row {index}"
+        if not row.get("file_name", "").startswith("Standard_Time_Lapses/"):
+            errors.append(f"{label}: canonical path is outside Standard_Time_Lapses")
+        if not RECORD_ID_RE.fullmatch(row.get("record_id", "")):
+            errors.append(f"{label}: invalid record_id")
+        if row.get("collection_tier") != "standard":
+            errors.append(f"{label}: collection_tier must be standard")
+        if row.get("is_canonical", "").lower() != "true":
+            errors.append(f"{label}: is_canonical must be true")
+        if row.get("acquisition_protocol_version") != "standard-2025-07-15":
+            errors.append(f"{label}: unexpected acquisition protocol version")
+        if not DATE_RE.fullmatch(row.get("date", "")):
+            errors.append(f"{label}: invalid or missing finish date")
+        if row.get("time") and not TIME_RE.fullmatch(row["time"]):
+            errors.append(f"{label}: invalid finish time")
+        if not DIMENSIONS_RE.fullmatch(row.get("dimensions", "")):
+            errors.append(f"{label}: invalid dimensions")
+        for field in ("tool", "medium"):
+            value = row.get(field, "")
+            if "|" in value:
+                errors.append(f"{label}: {field} still uses a pipe separator")
+            if value and any(not item.strip() for item in value.split(";")):
+                errors.append(f"{label}: {field} contains an empty list item")
+        for field in (
+            "duration_seconds",
+            "width_px",
+            "height_px",
+            "orientation",
+            "encoded_frame_rate",
+            "video_codec",
+            "pixel_format",
+            "container",
+            "has_audio",
+            "hub_xet_hash",
+            "size_bytes",
+        ):
+            if not row.get(field, ""):
+                errors.append(f"{label}: canonical technical field {field} is blank")
+
+        try:
+            width = int(row.get("width_px", ""))
+            height = int(row.get("height_px", ""))
+            expected_orientation = (
+                "portrait" if height > width else "landscape" if width > height else "square"
+            )
+            if row.get("orientation") != expected_orientation:
+                errors.append(
+                    f"{label}: orientation {row.get('orientation')!r} conflicts with {width}x{height}"
+                )
+        except ValueError:
+            pass
+
+    if pre_standard_metadata:
+        pre_fields, pre_standard_rows = load_csv(pre_standard_metadata)
+        missing_pre_fields = [
+            field for field in PRE_STANDARD_REQUIRED_FIELDS if field not in pre_fields
+        ]
+        if missing_pre_fields:
+            errors.append(
+                f"pre-standard metadata is missing required columns: {missing_pre_fields}"
+            )
+        pre_names = [row.get("file_name", "") for row in pre_standard_rows]
+        duplicate_pre_names = sorted(
+            value for value, count in Counter(pre_names).items() if value and count > 1
+        )
+        if duplicate_pre_names:
+            errors.append(
+                f"duplicate pre-standard file_name values: {duplicate_pre_names[:5]}"
+            )
+        for index, row in enumerate(pre_standard_rows, start=2):
+            name = row.get("file_name", "")
+            if not name or "/" in name or "\\" in name or not name.endswith(".mp4"):
+                errors.append(
+                    f"pre-standard CSV row {index}: file_name must be a direct MP4 basename"
+                )
+
+    if repo_files_json:
+        loaded_paths = json.loads(repo_files_json.read_text(encoding="utf-8"))
+        if not isinstance(loaded_paths, list) or not all(
+            isinstance(path, str) for path in loaded_paths
+        ):
+            errors.append("repository snapshot must be a JSON list of paths")
+        else:
+            repo_paths = set(loaded_paths)
+            standard_files = {path for path in repo_paths if DIRECT_STANDARD_RE.fullmatch(path)}
+            pre_standard_files = {
+                path for path in repo_paths if DIRECT_PRE_STANDARD_RE.fullmatch(path)
+            }
+            provenance_files = {
+                path
+                for path in repo_paths
+                if path.startswith("Standard_Time_Lapses/Provenance/")
+                and path.endswith(".mp4")
+            }
+            canonical_paths = {row.get("file_name", "") for row in rows}
+            active_pre_paths = {
+                f"Pre_Standard_Time_Lapses/{row.get('file_name', '')}"
+                for row in pre_standard_rows
+                if row.get("file_name")
+            }
+            missing_canonical_targets = sorted(canonical_paths.difference(repo_paths))
+            missing_pre_targets = sorted(active_pre_paths.difference(repo_paths))
+            repo_unindexed_standard = sorted(standard_files.difference(canonical_paths))
+            repo_unindexed_pre_standard = sorted(
+                pre_standard_files.difference(active_pre_paths)
+            )
+            if missing_canonical_targets:
+                errors.append(
+                    f"{len(missing_canonical_targets)} canonical metadata paths have no repository file"
+                )
+            if missing_pre_targets:
+                errors.append(
+                    f"{len(missing_pre_targets)} active pre-standard metadata paths have no repository file"
+                )
+            if repo_unindexed_pre_standard:
+                errors.append(
+                    f"{len(repo_unindexed_pre_standard)} published pre-standard files are absent from the active index"
+                )
+            if repo_unindexed_standard:
+                warnings.append(
+                    f"{len(repo_unindexed_standard)} direct standard files are pending metadata ingestion"
+                )
+            repo_counts = {
+                "repository_files": len(repo_paths),
+                "direct_standard_video_files": len(standard_files),
+                "direct_pre_standard_video_files": len(pre_standard_files),
+                "provenance_video_files": len(provenance_files),
+                "missing_canonical_metadata_targets": len(missing_canonical_targets),
+                "missing_pre_standard_metadata_targets": len(missing_pre_targets),
+            }
+
+    viewer_rows: list[dict[str, object]] = []
+    viewer_counts: dict[str, int] = {}
+    if viewer_json:
+        loaded = json.loads(viewer_json.read_text(encoding="utf-8"))
+        if not isinstance(loaded, list):
+            errors.append("Viewer snapshot must be a JSON list")
+        else:
+            viewer_rows = [row for row in loaded if isinstance(row, dict)]
+            paths = [viewer_path(row) for row in viewer_rows]
+            tiers = Counter(path.split("/", 1)[0] for path in paths if "/" in path)
+            viewer_counts = dict(sorted(tiers.items()))
+
+            canonical_viewer = {
+                viewer_path(row): row
+                for row in viewer_rows
+                if viewer_path(row).startswith("Standard_Time_Lapses/")
+            }
+            metadata_paths = {row["file_name"] for row in rows}
+            missing_from_viewer = sorted(metadata_paths.difference(canonical_viewer))
+            missing_from_metadata = sorted(set(canonical_viewer).difference(metadata_paths))
+            if missing_from_viewer:
+                errors.append(
+                    f"{len(missing_from_viewer)} metadata paths are absent from the Viewer snapshot"
+                )
+            if missing_from_metadata:
+                errors.append(
+                    f"{len(missing_from_metadata)} canonical Viewer paths are absent from metadata.csv"
+                )
+
+            by_path = {row["file_name"]: row for row in rows}
+            for path, viewer_row in canonical_viewer.items():
+                metadata_row = by_path.get(path)
+                if not metadata_row:
+                    continue
+                for field in (
+                    "date",
+                    "time",
+                    "support",
+                    "dimensions",
+                    "orientation",
+                    "video_codec",
+                    "pixel_format",
+                    "container",
+                    "hub_xet_hash",
+                ):
+                    left = metadata_row.get(field, "")
+                    right_value = viewer_row.get(field)
+                    right = "" if right_value is None else str(right_value)
+                    if left != right:
+                        errors.append(f"{path}: {field} differs from Viewer snapshot")
+
+            pre_standard_viewer_rows = [
+                row
+                for row in viewer_rows
+                if viewer_path(row).startswith("Pre_Standard_Time_Lapses/")
+            ]
+            pre_standard_without_technical = sum(
+                row.get("duration_seconds") is None for row in pre_standard_viewer_rows
+            )
+            if pre_standard_without_technical:
+                warnings.append(
+                    f"{pre_standard_without_technical} pre-standard Viewer rows lack duration metadata"
+                )
+
+    durations = [float(row["duration_seconds"]) for row in rows if row.get("duration_seconds")]
+    sizes = [int(row["size_bytes"]) for row in rows if row.get("size_bytes")]
+    orientations = Counter(row.get("orientation", "") or "<missing>" for row in rows)
+    audio = Counter(row.get("has_audio", "") or "<missing>" for row in rows)
+
+    if audio.get("True", 0) + audio.get("true", 0) > 0:
+        warnings.append("canonical masters commonly retain audio; rights/privacy review is still required")
+    warnings.append(
+        "encoded_frame_rate is playback rate, not original drawing-time sampling"
+    )
+
+    return {
+        "ok": not errors,
+        "errors": errors,
+        "warnings": warnings,
+        "statistics": {
+            "metadata_rows": len(rows),
+            "metadata_columns": len(fields),
+            "viewer_rows": len(viewer_rows),
+            "viewer_collection_counts": viewer_counts,
+            "pre_standard_metadata_rows": len(pre_standard_rows),
+            "repository_inventory": repo_counts,
+            "unindexed_direct_standard_files": repo_unindexed_standard,
+            "unindexed_direct_pre_standard_files": repo_unindexed_pre_standard,
+            "unique_record_ids": len({row.get("record_id", "") for row in rows}),
+            "unique_nonempty_hashes": len(
+                {row.get("hub_xet_hash", "") for row in rows if row.get("hub_xet_hash")}
+            ),
+            "unknown_finish_times": sum(not row.get("time") for row in rows),
+            "orientation_counts": dict(sorted(orientations.items())),
+            "audio_counts": dict(sorted(audio.items())),
+            "playback_hours": round(sum(durations) / 3600, 2),
+            "median_playback_seconds": round(statistics.median(durations), 3),
+            "indexed_size_tb_decimal": round(sum(sizes) / 1_000_000_000_000, 3),
+        },
+    }
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("metadata_csv", type=Path)
+    parser.add_argument("--viewer-json", type=Path)
+    parser.add_argument("--pre-standard-metadata", type=Path)
+    parser.add_argument("--repo-files-json", type=Path)
+    parser.add_argument("--output", type=Path)
+    args = parser.parse_args()
+
+    report = validate(
+        args.metadata_csv,
+        args.viewer_json,
+        args.pre_standard_metadata,
+        args.repo_files_json,
+    )
+    rendered = json.dumps(report, indent=2, sort_keys=True)
+    if args.output:
+        args.output.parent.mkdir(parents=True, exist_ok=True)
+        args.output.write_text(rendered + "\n", encoding="utf-8")
+    print(rendered)
+    if not report["ok"]:
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()