Archive GTA capture image inventory with historical diagnostic limits

NOTICE.md

@@ -0,0 +1,5 @@
+# Capture archive and rights
+
+The Diogenes project owner authorized preservation of these original capture experiments and diagnostic outputs. Grand Theft Auto and its game imagery remain the property of Rockstar Games, Take-Two Interactive and other respective rights holders. This archive grants no new rights in the underlying game, extracted assets, commercial exploitation or model training. Only recorded/rendered images and derived diagnostic pictures are included here; game executables, extracted asset libraries and proprietary SDKs are excluded from this image scope.
+
+Images retain original bytes. For 72 filenames incompatible with the archive transport format, a hash-based public filename is used; MANIFEST.jsonl preserves the original logical path. This renames the container destination only. Historical rejected, non-trainable, diagnostic, incomplete and failed-QC states remain in force; publication is not certification. Related metadata and code are archived separately. No inferred licence replaces third-party rights.

README.md

@@ -0,0 +1,30 @@
+---
+pretty_name: Diogenes GTA Capture Images Archive v01
+license: other
+license_name: original-capture-archive-terms
+license_link: https://huggingface.co/datasets/DiogenesLab/Diogenes_GTA_Capture_Images_Archive_v01/blob/main/NOTICE.md
+tags:
+- archival
+- game-capture
+- non-trainable
+size_categories:
+- 10K<n<100K
+---
+
+# Diogenes GTA Capture Images Archive v01
+
+The project has closed. This owner-authorized archive preserves **35,418 original stored images / 113,487,805,836 bytes**: 35,156 PNGs and 262 JPEGs in 285 historical capture and diagnostic directories. **Transfer is in progress**; the intended-member manifest is not proof that every shard has arrived.
+
+The scope includes GTA camera captures, exported frame sequences, review contact sheets, pointer-rejection examples, masks, synthetic gradients and camera diagnostics. Stored-image counts are not counts of unique gameplay frames. Source paths retain historical scene/attempt identifiers; some runs, previews and copied handoffs duplicate content. Failed and rejected experiments are deliberately preserved and remain unsuitable for uses their original quality records disallow. Availability does not establish trainability or production certification.
+
+## Review and limits
+
+Review read all associated structured metadata in the wider capture archive and visually inspected 553 image samples: every image in small directories (fewer than 20 files), and first/middle/last samples in larger directories. Sample headers were checked for text identity metadata; the sampled pictures depict fictional game scenes and capture diagnostics. This is a sampled visual review, not exhaustive inspection of all 35,418 images. Sample dimensions vary from 64×48 test patterns to contact sheets; most sampled gameplay images are 1920×1080. The archive does not imply uniform dimensions, camera calibration, successful pointer isolation or geometry correctness.
+
+## Layout and integrity
+
+There are 590 planned tar shards containing up to 100 intact images from one directory. No image is split or re-encoded. Original relative paths are retained where transport-safe; 72 long or non-ASCII destinations use `diagnostics/images/<key-hash>.<extension>` with `original_logical_path` recorded in `MANIFEST.jsonl`. Each tar includes a per-member SHA-256/size manifest. Source hashes, pinned Hugging Face LFS hashes/sizes and anonymous download checks are verified during transfer before source cleanup.
+
+This image-only collection excludes video, audio, replay streams, ZIP handoffs, broader metadata, binaries and game assets. Adjacent timing/pose/QC data must not be assumed present in this repository. Historical JSON metadata and explicit privacy-sanitized copies are being archived separately; original hash/signature declarations in those documents describe their original bytes and do not certify sanitized copies.
+
+See [NOTICE.md](NOTICE.md) for rights and unchanged usage limits. Capture source code: [Diogenes-Archive](https://github.com/CuddlerAI/Diogenes-Archive).