mirror of
https://github.com/wiltodelta/remove-ai-watermarks.git
synced 2026-10-21 08:21:16 +02:00
Victor Kuznetsov
07b07634cf
* Add rigorous SynthID research and evaluation harness * Record SynthID cross-epoch validation results * Reset SynthID detector hypotheses to paired training * Add paired wavelet and spectral SynthID probes * Record closed-corpus SynthID detector ablations * Add registered SynthID phase probing * Add periodic SynthID tile probing * Add calibrated SynthID pixel detector * Harden scale-registered SynthID detection * Record SynthID phase and channel rescue findings * Add production SynthID routing and OpenAI verification * Reframe periodic pixel route as pipeline lattice; confirm and harden detection The frozen periodic experts read an origin-anchored generation-pipeline lattice destroyed by a crop off the tile grid, not the crop-robust SynthID mark. Route the pixel result as an experimental pipeline_lattice signal kept out of the watermark inventory, and carry the crop sensitivity in every verdict envelope. Add split-patch phase/amplitude/codeword confirmation for registered-v3, affine-lattice and cyclostationary research probes, and timeout/retry/error-taxonomy hardening for the official OpenAI verification path. * Record pipeline-lattice reframing and registered-v3 phase lock Document that the pixel route detects an origin-anchored lattice rather than the watermark, and record the registered-v3 two-pixel phase-lock measurement: a 2px diagonal crop killed all 36 tested detections (28 foreign-generator, 8 Google) with recovery only at offsets that are multiples of four. Update README, CLI, Python API, supported-signals, known-limitations, module-internals, the SynthID reference, and the detector research plan. * Split SynthID campaign docs by detector, classifiers, and removal Keep Model 1 as a research AI-versus-camera result outside the library. Retire the mixed oracle-ladders page to a routing hub and mark the plan file as a chronological archive. * Keep 2026-08-22 SynthID campaign results in research docs only Drop Bayer, Model 1, and lattice re-check numbers from user guides and module internals. The library surface is unchanged. * Unplug lattice detector from package, expand research docs * Remove lattice detector sources from the public package * Document SynthID GitHub sweep findings * Document general AI classifier sweep * Document modern-negative taxonomy audit and classifier doc split Split general AI-classifier research into ai-generated-image-classifiers.md and record the modern-negative campaign: Unsplash/Openverse date-clean cells, contamination controls, quarter-hard paired result, closed linear veto and taxonomy-continuation arms. pre-commit: 1) maintain.sh - exit 1, known uv-secure lightning advisory with no upstream fix (same triage as main); core checks separately green (ruff, format, pyright, 1564 tests); 2) /simplify - docs-only single pass, no findings; 3) docs sync - split references updated everywhere, none stale; 4) CLAUDE.md - routing line already updated, compliant * Record wild-AI audit, Meta provider class, and stock-negative expansion The wild vendor-flagged AI cell (300 stratified rows) puts Model 1 recall at 69.7% on unknown-renderer stock AI; the stock-negative harvest triples the modern fashion/product cells and confirms the combined-pool veto control; the Meta muse-image corpus doubles to 132 rows with its margin sweep; a per-channel cv2 reference fixes the latent fold test under cv2 4.10.0. pre-commit: 1) maintain.sh - exit 1, known uv-secure lightning advisory with no upstream fix; core checks separately green (ruff, format, pyright, 1665 tests); 2) /simplify - docs-only single pass, no findings; 3) docs sync - new run references point at the gitignored research store, none stale; 4) CLAUDE.md - compact, no changes needed * Close representation-change campaign: both candidates falsified Pre-registered gates (AI-test >=1772, fresh OI <=50, Kodak 0, FLUX >=288, EvalGEN >=95) rejected DINOv2-L ridge head (1427/1847, 162/3000, 1/24, 29/300, 5/100) and the paired-reconstruction FFT tower (14/1847, 12/3000, 0/24, 0/300, 0/100; contract learned at 0.949 pairwise separation but no transfer to production generators). Campaign verdict: general classifier beyond local scale as pursued; product stays on documented remainder. pre-commit: 1) maintain.sh - exit 1, known uv-secure lightning advisory with no upstream fix; core checks green minutes ago (ruff, format, pyright, 1665 tests); 2) /simplify - docs-only single pass, no findings; 3) docs sync - verdict mirrored in data/research/REPRESENTATION-CAMPAIGN.md, no stale refs; 4) CLAUDE.md - no changes * Record full failure map, H2 router falsification, and H3 process detector closure 20 measured negative domains with batch-2 (memes, whiteboards, tickets, book pages). H2 domain router falsified: 97.9% of AI-test looks structured, every family 94-100%. H3 process detector on 369 contrastive pairs from three generators (OpenAI gpt-image-1-mini, Meta muse-image-1.0, Gemini flash-image) achieves 0.909 pairwise AUC on training pairs but transfers at AUC 0.55-0.60 to production AI; human receipts score higher than any AI family. The generator-specific spectral trace does not generalize across three generators, closing the process-trace family in its current form. pre-commit: 1) maintain.sh - exit 1, known lightning advisory; core green (ruff, format, pyright, 1665 tests); 2) /simplify - docs + research scripts, single pass, no findings; 3) docs sync - all artifacts in data/research/, no stale refs; 4) CLAUDE.md - no changes * Record H4-H6 research hypotheses: MLP probe, spectral invariant, TTA voting H4 (MLP on frozen CLIP): 98.6% AI-test recall but 15.8% photo FPR — same trade-off curve as quarter-hard. H5 (spectral invariant from CVPR 2025): 2.4% recall, the 1/f^alpha profile does not discriminate current generators. H6 (TTA voting): fixes museum FP but not documents/UI/fashion, loses 9% recall. All three falsified against frozen gates; ten total closed hypotheses now document the CLIP-L representation limit from every angle. Also records the provider cascade across all 20 domains (69.8% stray on documents mirrors the detector failure map) and the full Y-phase campaign PSNR data awaiting oracle rate-limit reset. pre-commit: 1) maintain.sh - exit 1, known lightning advisory; core green (ruff, format, pyright, 1665 tests); 2) /simplify - docs-only single pass, no findings; 3) docs sync - all artifacts in data/research/, no stale refs; 4) CLAUDE.md - no changes * Fix Y-phase frequency convention, record H4-H6 and campaign results The initial Y-phase campaign targeted FFT bins 16-32 instead of spatial periods 16-32 px (1/32 to 1/16 cycles/px). After the fix, the full-octave Y-phase scramble reproduces the known kill (not_detected at 20.5 dB on s1 at native 1122x1402). Partial sub-bands are visually invisible (42.5 dB) but stay detected: the mark is distributed across the full octave, not concentrated. Point intervention is not viable within this frequency band. H4 MLP probe: 98.6% recall, 15.8% FPR (same trade-off wall). H5 spectral invariant: 2.4% recall (1/f^alpha does not discriminate). H6 TTA voting: fixes museum FP but not documents/UI/fashion. pre-commit: 1) maintain.sh - exit 1, known lightning advisory; core green (1665 tests, verified minutes ago); 2) /simplify - docs-only; 3) docs sync - all artifacts synced; 4) CLAUDE.md - no changes * Add TC260 fifth class and per-class margin calibration to provider cascade Five-head cascade (openai/google/tc260/meta/no_ai): openai 68.8%, google 76.6%, tc260 63.8%, meta 71.8% mean recall at uniform margin 0.50. Per-class margins calibrated against 1.67% photo budget: openai 0.31, google 0.29, tc260 0.39, meta 0.55. TC260 test recall 71.4% at calibrated margins on 196 held-out images. Chinese-AIGC is a viable pixel class. pre-commit: 1) maintain.sh - exit 1, known lightning advisory; core green (verified within the hour); 2) /simplify - research script, single pass; 3) docs sync - artifacts in data/research/; 4) CLAUDE.md - no changes * Record universal detector v3-v8: diverse training shifts the curve Fresh MLP on frozen CLIP embeddings with 49 negative domains (12,258 images) and 5,043 AI positives achieves 93.0% AI-test recall with 39/49 domains at zero false positives. Digital art FP dropped from 99.3% (Model 1) to 24.8%, fashion from 62-71% to 12%, UI from 99.3% to 12%. Photo FPR 0.7%. The improvement comes from training data diversity on frozen embeddings, not representation change (v5 fine-tune proved this by undoing the gains). Also adds .local-eval to ruff exclude so the gate covers only tracked files. pre-commit: 1) maintain.sh - exit 1, known lightning advisory; core green (ruff, format, pyright, 1731 tests); 2) /simplify - docs + research; 3) docs sync - all artifacts in data/research/; 4) CLAUDE.md - no changes * Add three-tier ensemble detector: DEFINITELY/POSSIBLY/LIKELY_HUMAN Combines Model 1 and v8 into calibrated confidence tiers. DEFINITELY_AI (both models agree) achieves 79.6% recall at 0.50% photo FPR with 46/49 domains at zero FP. The union reaches 93.3% recall at 1.80% photo FPR with Kodak 0/24. First detector working across ALL content types with per-tier calibrated error rates. pre-commit: 1) maintain.sh - exit 1, known lightning advisory; core green (ruff, format, pyright, 1731 tests); 2) /simplify - docs + ensemble; 3) docs sync - artifacts in data/research/; 4) CLAUDE.md - no changes * Add provider v2: diversity training eliminates document leakage Same diversity-training insight applied to the provider cascade. Adding 3,769 structured negative features to training drops document/receipt leakage from 69.8% to 5.1%, UI from 10% to 4.3%, fashion from 5-15% to 0-3%. AI recall improves: OpenAI 84% (was 75%), Google 88.8% (was 79%). Photo leakage 1.4%. Worst regression: memes 47.9% (was 27%). pre-commit: 1) maintain.sh - exit 1, known lightning advisory; core green (ruff, format, pyright, 1731 tests); 2) /simplify - provider v2; 3) docs sync - artifacts in data/research/; 4) CLAUDE.md - no changes * Add universal v11: full taxonomy, 136 domains, 124 at zero FP Master data plan mapped 148 visual domains. Massive collection (3,343 images, 73 domains from Openverse) plus gap-filler (724 images from Safebooru anime/manga) expanded training to 19,768 images across 136 negative domains. Universal v11 (MLP-512 + Model 1 ensemble): DEFINITELY 90.3% recall at 0.50% photo FPR, ANY 95.6% recall, 124/134 domains at exactly 0% FP, worst FP 13.8% (digital art). New domains opened: anime, manga, chibi, mecha, vintage photos, banknotes, newspapers, magazines, movie posters, antique maps, watercolor/oil originals, patterns, cosmetics, real estate, 3D renders, and 60+ more. pre-commit: 1) maintain.sh - exit 1, known lightning advisory; core green (ruff, format, pyright, 1731 tests); 2) /simplify - docs + research; 3) docs sync - artifacts in data/research/; 4) CLAUDE.md - no changes * Confirm representation limit: v12 fine-tune destroys structured FP despite 6x data Fine-tuning CLIP tower on 19,518 images achieves 95.1% standalone recall but returns to 100% FP on art/UI and 14/59 zero-FP domains (vs 124/134 for frozen v11). This is a representation property, not a data issue. Frozen CLIP + MLP (v11) is the only working architecture. Also records provider v3 trained on all negative domains. pre-commit: 1) maintain.sh - exit 1, known lightning advisory; core green (ruff, format, pyright, 1731 tests); 2) /simplify - docs + research; 3) docs sync - artifacts in data/research/; 4) CLAUDE.md - no changes * Document v13 plateau: 128/136 zero FP, frozen-CLIP ceiling confirmed v13 with top-plan expansion (1,169 new images) improves domain coverage to 128/136 at zero FP and reduces fashion FP to 5.3%, but DEFINITELY recall plateaus at 89.6% and art FP at 15.9%. The frozen CLIP + MLP architecture has reached its ceiling. pre-commit: 1) maintain.sh - exit 1, known lightning advisory; core green (verified minutes ago, 1731 tests); 2) /simplify - docs-only; 3) docs sync - v13 recorded; 4) CLAUDE.md - no changes * Document the 2026-08-31 photo-classifier freeze. Record the AI-versus-camera contract, Meta v3 hold-out, Flickr UI replacement, and that receipts stay out of the supported domain. * Scrub Spaces catalog details from classifier research docs * Record receipt layout-gate probe for the universal detector
95 lines
5.0 KiB
Markdown
95 lines
5.0 KiB
Markdown
# Repository data
|
|
|
|
Tracked data is organized by purpose:
|
|
|
|
```text
|
|
data/
|
|
fixtures/
|
|
provenance/ Real format and provenance fixtures used by tests
|
|
(source records live in fixtures/README.md)
|
|
visible/ Synthetic per-mark example gallery (one committed example per
|
|
registered visible mark; see fixtures/visible/README.md)
|
|
calibration/
|
|
<vendor>/ Minimal controlled inputs needed to rebuild detector assets
|
|
synthid/
|
|
originals/ Canonical provider-oracle fixtures, stored once
|
|
manifest.csv Provenance and verification record for each original
|
|
full-pipeline-quality.csv
|
|
Reusable full-pipeline evaluation selection
|
|
evaluations/
|
|
engine-selection/
|
|
Prompt-matched AI content matrix for deterministic auto-engine research
|
|
fidelity/ Evaluation instructions and hand-verified ground truth
|
|
video-synthid-oracle.csv
|
|
Reproducible full-clip Gemini SynthID verdicts
|
|
```
|
|
|
|
## Storage rules
|
|
|
|
Large local-only research data (the 211 GB Spaces corpus and the frozen
|
|
checkpoints, research corpora, and campaign reports) lives under gitignored
|
|
paths documented in `data/research/README.md` (ignored via `.git/info/exclude`,
|
|
never committed). Worktree `.local-eval/` directories are scratch copies; the
|
|
durable store is `data/research/` in the main checkout.
|
|
|
|
1. Store each binary image once. Evaluation manifests and documentation point
|
|
to its canonical location.
|
|
2. Put executable test fixtures in `fixtures/`.
|
|
3. Put only the minimal reproducible detector inputs in `calibration/`.
|
|
4. Put externally verified SynthID originals in `synthid/originals/` and keep
|
|
both CSV files synchronized.
|
|
5. Keep evaluation outputs outside the repository. Record reproducible
|
|
commands, hashes, and oracle verdicts instead of committing another corpus
|
|
copy. A small curated before-and-after example may live in `docs/images/`
|
|
when it is part of the public documentation.
|
|
`evaluations/engine-selection/` is the bounded exception for ORIGINAL study
|
|
inputs: its manifest tracks a prompt-matched OpenAI/Meta content matrix, not
|
|
generated model outputs or watermark-oracle evidence.
|
|
6. Keep third-party fixture license notices in `licenses/`, outside directories
|
|
that tests enumerate as media inputs.
|
|
7. Runtime detector assets belong in `src/remove_ai_watermarks/assets/`.
|
|
Unregistered research candidates belong in
|
|
`scripts/assets/visible-mark-candidates/` so they are not shipped in the
|
|
wheel.
|
|
|
|
The source distribution excludes `data/`; the wheel contains only package
|
|
runtime assets.
|
|
|
|
## Video SynthID oracle manifest
|
|
|
|
`evaluations/video-synthid-oracle.csv` is the only evidence that the shipped
|
|
video removal profile works, so it is also the source of truth for three shipped
|
|
defaults: `tests/test_video_invisible.py` asserts that `noise_std`, `long_side`,
|
|
and `fps` together match a row this manifest records as certified. Changing one
|
|
of those three without adding the row that certifies it fails the suite. Both
|
|
historical rows mark `vae` as `unrecorded`: the repository must not infer a model
|
|
identity from the current default. Add it to the assertion in the same commit as
|
|
the first oracle row that records and verifies one.
|
|
|
|
| Column | Meaning |
|
|
| --- | --- |
|
|
| `date`, `source_url`, `source_sha256` | Identify the carrier. |
|
|
| `source_width`, `source_height`, `source_fps` | Carrier geometry. Without it the actual downscale factor of a row cannot be recovered later. |
|
|
| `duration_seconds`, `source_verdict` | Clip length submitted and the verifier's reading of the untouched carrier. |
|
|
| `vae`, `noise_std`, `long_side`, `fps`, `seed` | The run configuration. `unrecorded` means the historical run did not preserve the VAE identity. |
|
|
| `output_sha256` | Identifies the exact submitted file. |
|
|
| `output_verdict` | One of `detected`, `not_detected`, `indeterminate`, `refused`. |
|
|
| `output_verdict_text` | The verifier's wording, verbatim. |
|
|
| `output_detected_range` | Time range the verifier reported for the output. |
|
|
| `track` | `visual`, `audio`, `both`, or empty. The verifier scores tracks separately and this path copies source audio unchanged. |
|
|
| `session_id` | Groups rows submitted in one oracle session, so per-session drift stays visible. |
|
|
| `stratum` | Content class of the carrier, for stratified certification. |
|
|
| `psnr_db`, `temporal_residual_ratio` | Fidelity measurements. Neither is a watermark verdict. |
|
|
|
|
Record `indeterminate` when the verifier answers with its unclear state rather
|
|
than a negative: an unclear reading logged as `not_detected` is exactly the
|
|
silent regression this manifest exists to prevent. Leave a field empty when it
|
|
was not recorded, and never backfill it with a plausible value.
|
|
|
|
`psnr_db` is measured against the already-resized frame and before the encode,
|
|
so it excludes the downscale, the decimation, and the codec. Rows stay
|
|
comparable to each other only while that definition holds.
|
|
|
|
The two 2026-07-31 rows predate this schema; their empty fields were never
|
|
recorded and are not recoverable from the row.
|