mirror of
https://github.com/wiltodelta/remove-ai-watermarks.git
synced 2026-10-14 13:30:55 +02:00
* 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
213 lines
8.0 KiB
Python
213 lines
8.0 KiB
Python
from __future__ import annotations
|
|
|
|
import json
|
|
import sys
|
|
from pathlib import Path
|
|
|
|
import numpy as np
|
|
import pytest
|
|
from PIL import Image
|
|
|
|
sys.path.insert(0, str(Path(__file__).resolve().parent.parent / "scripts"))
|
|
|
|
import synthid_oracle_batch as batch
|
|
from synthid_periodic_tile import fold_residual_template, unit_tile
|
|
from synthid_periodic_tile_probe import PeriodicTileModel
|
|
from synthid_periodic_tile_probe import save_model as save_tile_model
|
|
from synthid_phase_carrier import PhaseCarrierModel
|
|
from synthid_phase_carrier import save_model as save_phase_model
|
|
|
|
|
|
def _fixture(tmp_path: Path) -> tuple[Path, Path, Path]:
|
|
rng = np.random.default_rng(23)
|
|
raw_tile = rng.normal(size=(8, 8, 3))
|
|
raw_tile -= np.mean(raw_tile, axis=(0, 1), keepdims=True)
|
|
raw_tile, _ = unit_tile(raw_tile)
|
|
source = np.clip(np.rint(128.0 + 24.0 * np.tile(raw_tile, (8, 8, 1))), 0, 255).astype(np.uint8)
|
|
source_path = tmp_path / "source.png"
|
|
Image.fromarray(source, mode="RGB").save(source_path)
|
|
|
|
folded = fold_residual_template(
|
|
source,
|
|
tile_height=8,
|
|
tile_width=8,
|
|
denoise_sigma=1.0,
|
|
)
|
|
template, expected_norm = unit_tile(folded)
|
|
tile_model = PeriodicTileModel(64, 64, 8, 8, 1.0, template, expected_norm)
|
|
tile_model_path = tmp_path / "tile-model.npz"
|
|
save_tile_model(tile_model_path, tile_model)
|
|
|
|
spectra = np.stack([np.fft.rfft2(source[:, :, channel]) for channel in range(3)], axis=2)
|
|
magnitude = np.abs(spectra)
|
|
magnitude[0, 0, :] = 0.0
|
|
row, column, channel = np.unravel_index(int(np.argmax(magnitude)), magnitude.shape)
|
|
phase_model = PhaseCarrierModel(
|
|
64,
|
|
64,
|
|
np.asarray([row], dtype=np.int32),
|
|
np.asarray([column], dtype=np.int32),
|
|
np.asarray([channel], dtype=np.int8),
|
|
np.asarray([np.angle(spectra[row, column, channel])]),
|
|
np.asarray([1.0]),
|
|
np.asarray([magnitude[row, column, channel]]),
|
|
)
|
|
phase_model_path = tmp_path / "phase-model.npz"
|
|
save_phase_model(phase_model_path, phase_model)
|
|
return source_path, tile_model_path, phase_model_path
|
|
|
|
|
|
def _build(tmp_path: Path) -> tuple[Path, Path]:
|
|
source, tile_model, phase_model = _fixture(tmp_path)
|
|
output_dir = tmp_path / "oracle-batch"
|
|
manifest_path = batch.build_batch(
|
|
[source],
|
|
output_dir=output_dir,
|
|
tile_model_path=tile_model,
|
|
phase_model_path=phase_model,
|
|
tile_threshold=0.5,
|
|
phase_threshold=0.5,
|
|
active_threshold=0.0,
|
|
strength=2.0,
|
|
seed=20260810,
|
|
provider="google",
|
|
repository_root=Path(__file__).resolve().parent.parent,
|
|
)
|
|
return output_dir, manifest_path
|
|
|
|
|
|
def test_build_preregisters_fixed_request_order_without_copying_source(tmp_path: Path) -> None:
|
|
output_dir, manifest_path = _build(tmp_path)
|
|
|
|
manifest = json.loads(manifest_path.read_text(encoding="utf-8"))
|
|
|
|
assert manifest["status"] == "preregistered_unsubmitted"
|
|
assert manifest["request_count"] == 5
|
|
assert [row["role"] for row in manifest["rows"]] == list(batch.ROLE_ORDER)
|
|
assert manifest["rows"][0]["in_batch"] is False
|
|
assert not (output_dir / "source.png").exists()
|
|
assert manifest["provider"] == "google"
|
|
assert "shifted, and orthogonal_random are detected" in manifest["decision_rule"]
|
|
assert all(row["synthid_result"] is None for row in manifest["rows"])
|
|
assert all(row["c2pa_result"] is None for row in manifest["rows"])
|
|
template = json.loads((output_dir / "results-template.json").read_text(encoding="utf-8"))
|
|
assert template["manifest_sha256"] == batch.artifact_sha256(manifest_path)
|
|
assert [row["artifact_sha256"] for row in template["rows"]] == [row["artifact_sha256"] for row in manifest["rows"]]
|
|
assert (
|
|
batch.verify_batch(
|
|
manifest_path,
|
|
repository_root=Path(__file__).resolve().parent.parent,
|
|
)["source_count"]
|
|
== 1
|
|
)
|
|
|
|
|
|
def test_verify_rejects_derivative_mutation(tmp_path: Path) -> None:
|
|
output_dir, manifest_path = _build(tmp_path)
|
|
manifest = json.loads(manifest_path.read_text(encoding="utf-8"))
|
|
aligned_row = next(row for row in manifest["rows"] if row["role"] == "aligned")
|
|
aligned_path = output_dir / aligned_row["path"]
|
|
with Image.open(aligned_path) as image:
|
|
pixels = np.asarray(image.convert("RGB"), dtype=np.uint8).copy()
|
|
pixels[0, 0, 0] ^= 1
|
|
Image.fromarray(pixels, mode="RGB").save(aligned_path)
|
|
assert batch.artifact_sha256(aligned_path) != aligned_row["artifact_sha256"]
|
|
|
|
with pytest.raises(ValueError, match="artifact hash mismatch"):
|
|
batch.verify_batch(
|
|
manifest_path,
|
|
repository_root=Path(__file__).resolve().parent.parent,
|
|
)
|
|
|
|
|
|
def test_build_rejects_output_inside_repository(tmp_path: Path) -> None:
|
|
source, tile_model, phase_model = _fixture(tmp_path)
|
|
repository_root = Path(__file__).resolve().parent.parent
|
|
|
|
with pytest.raises(ValueError, match="outside the repository"):
|
|
batch.build_batch(
|
|
[source],
|
|
output_dir=repository_root / ".local-eval/oracle-batch-test",
|
|
tile_model_path=tile_model,
|
|
phase_model_path=phase_model,
|
|
tile_threshold=0.5,
|
|
phase_threshold=0.5,
|
|
active_threshold=0.0,
|
|
strength=2.0,
|
|
seed=20260810,
|
|
provider="google",
|
|
repository_root=repository_root,
|
|
)
|
|
|
|
|
|
def test_build_rejects_nonempty_output_directory(tmp_path: Path) -> None:
|
|
source, tile_model, phase_model = _fixture(tmp_path)
|
|
output_dir = tmp_path / "existing"
|
|
output_dir.mkdir()
|
|
(output_dir / "marker.txt").write_text("occupied", encoding="utf-8")
|
|
|
|
with pytest.raises(ValueError, match="must not already contain"):
|
|
batch.build_batch(
|
|
[source],
|
|
output_dir=output_dir,
|
|
tile_model_path=tile_model,
|
|
phase_model_path=phase_model,
|
|
tile_threshold=0.5,
|
|
phase_threshold=0.5,
|
|
active_threshold=0.0,
|
|
strength=2.0,
|
|
seed=20260810,
|
|
provider="google",
|
|
repository_root=Path(__file__).resolve().parent.parent,
|
|
)
|
|
|
|
|
|
def test_evaluate_requires_controls_and_aligned_outcome(tmp_path: Path) -> None:
|
|
output_dir, manifest_path = _build(tmp_path)
|
|
results = json.loads((output_dir / "results-template.json").read_text(encoding="utf-8"))
|
|
for row in results["rows"]:
|
|
row["synthid_result"] = "not_detected" if row["role"] == "aligned" else "detected"
|
|
row["c2pa_result"] = "unavailable"
|
|
row["raw_response"] = f"verbatim {row['role']} result"
|
|
row["submitted_at"] = "2026-08-10T20:00:00Z"
|
|
results_path = output_dir / "results.json"
|
|
results_path.write_text(json.dumps(results, indent=2) + "\n", encoding="utf-8")
|
|
|
|
report = batch.evaluate_results(
|
|
manifest_path,
|
|
results_path,
|
|
repository_root=Path(__file__).resolve().parent.parent,
|
|
)
|
|
|
|
assert report["counts"] == {
|
|
"causal_success": 1,
|
|
"aligned_still_detected": 0,
|
|
"control_failed": 0,
|
|
"indeterminate": 0,
|
|
}
|
|
|
|
shifted = next(row for row in results["rows"] if row["role"] == "shifted")
|
|
shifted["synthid_result"] = "not_detected"
|
|
results_path.write_text(json.dumps(results, indent=2) + "\n", encoding="utf-8")
|
|
control_report = batch.evaluate_results(
|
|
manifest_path,
|
|
results_path,
|
|
repository_root=Path(__file__).resolve().parent.parent,
|
|
)
|
|
assert control_report["counts"]["control_failed"] == 1
|
|
|
|
|
|
def test_evaluate_rejects_incomplete_result(tmp_path: Path) -> None:
|
|
output_dir, manifest_path = _build(tmp_path)
|
|
results = json.loads((output_dir / "results-template.json").read_text(encoding="utf-8"))
|
|
results["rows"].pop()
|
|
results_path = output_dir / "incomplete-results.json"
|
|
results_path.write_text(json.dumps(results, indent=2) + "\n", encoding="utf-8")
|
|
|
|
with pytest.raises(ValueError, match="cover every"):
|
|
batch.evaluate_results(
|
|
manifest_path,
|
|
results_path,
|
|
repository_root=Path(__file__).resolve().parent.parent,
|
|
)
|