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.

README.md

@@ -30,7 +30,7 @@ removal.
 | Goal | Command | GPU |
 | --- | --- | --- |
 | Find provenance signals and watermarks | `identify` | No |
-| Detect the SynthID pixel carrier in the calibrated image-size range | `detect-synthid` | No |
+| Detect a generation-pipeline pixel lattice, experimental, not the watermark | `detect-synthid` | No |
 | Verify supported OpenAI SynthID from pixels with the official remote API | `verify-openai-synthid` | No |
 | Remove known visible AI marks | `visible` | No |
 | Erase a region you select | `erase` | No |
@@ -78,31 +78,54 @@ Inspect an image:
 remove-ai-watermarks identify image.png
 ```
 
-Detect the supported SynthID pixel carrier after installing the pixel runtime:
+Signed provenance is the supported route for SynthID and `identify` reads it.
+The pixel route below is experimental and answers a different question. Install
+the pixel runtime first:
 
 ```bash
 uv tool install --force "remove-ai-watermarks[pixels]"
 remove-ai-watermarks detect-synthid image.png
-remove-ai-watermarks detect-synthid resized.png --register-scale
+remove-ai-watermarks detect-synthid native-period.png --fixed-period
 ```
 
-This detector is positive-only and limited to one measured carrier family in
+This is experimental and does not detect the SynthID watermark. The statistic it reports is
+destroyed by cropping off the tile grid — a seven-pixel crop on the large branch
+and a two-pixel crop on the ordinary-size registered-v3 route — while SynthID's
+published evaluation keeps 99.97% of its detection rate under aggressive crop
+and resize, so a positive identifies the generation pipeline rather than the
+mark. Measured on 2026-08-16: it accepted 8 of 11 fresh large Google generations
+and 29 of 223 images from other generators, and 0 of 6 when the same positives
+were cropped off the tile grid; a two-pixel crop also removed all 36 tested
+registered-v3 detections across foreign-generator and Google images. It is
+positive-only
+and limited to one measured lattice family in
 the [calibrated image-size range](docs/synthid.md#32-how-our-tool-detects-the-supported-carrier).
-The native default uses the fixed fold through 10 megapixels and a separately
-challenged opponent-color large-image branch above 10 through 18 megapixels;
-the large branch requires both sides to be at least 2,048 pixels. Both expect
-the recovered carrier at its measured 16-pixel sampling scale.
-`--register-scale` opts into a much slower bounded scale search
-for resized images from 250,000 through 10,000,000 decoded pixels, with both
-sides at least 64 pixels. Its measured positive scale range is approximately
-0.65 through 1.5; 0.5x resizes remain outside reliable detection. `identify`
-keeps the fast default. `not_detected` or `unsupported` is not a clean-image
-guarantee.
+The production default uses registered-v3 from 250,000 through 10,000,000
+decoded pixels with both sides at least 256 pixels. When that route abstains,
+a narrower opponent-registered-v1 fallback covers 1 through 10 megapixels,
+both sides at least 768 pixels, and carrier periods 7.9 through 12.0. Period-8
+candidates additionally require a decoded-pixel block-edge veto that separates
+the measured carrier from ordinary JPEG lattices. The
+separately challenged opponent-color large-v1 branch runs
+above 10 through 18 megapixels; large-v1 requires both sides to be at least
+2,048 pixels. Registered-v3 is a bounded scale search with a measured positive
+scale range of approximately 0.65 through 1.5. The fallback recovered 49/49
+losslessly resized views at scales 0.5 through 0.75 from seven official
+pixel-positive parents. The period-8 veto rejected all 1,790 previously crossing
+codec-lattice controls, while 350 matched 0.5x Picsum controls had no base
+crossing; the earlier period-band rule also accepted 0/1,000 frozen controls.
+`--fixed-period` explicitly selects the faster legacy 16-pixel diagnostic; it
+is not a production positive route. `--register-scale` forces the registered
+expert, including on geometry where the default would choose large-v1.
+`identify` uses the production router. `indeterminate` means no qualified local
+expert crossed its threshold; `unsupported` means no expert covers the input.
+Neither is a clean-image guarantee.
 
-The large native branch is not recompression-robust: all seven official large
-positives fell below its frozen threshold after same-size JPEG-95 and JPEG-90
-re-encoding. Use it for original or losslessly copied pixels, and treat a miss
-after lossy transcoding as inconclusive.
+The large native and opponent-registered branches are not recompression-robust.
+The large branch lost all seven official positives after same-size JPEG-95 and
+JPEG-90; the fallback retained 0/63 JPEG-95, JPEG-85, and WebP-95 views. Use
+them for original or losslessly copied pixels, and treat a miss after lossy
+transcoding as inconclusive.
 
 For supported OpenAI images, the optional official verifier provides a broader
 pixel-watermark verdict than the incomplete local OpenAI research signal:

docs/cli.md

@@ -57,36 +57,58 @@ remove-ai-watermarks identify image.png --no-visible
 ```
 
 Despite the historical option name, `--no-visible` skips all pixel detectors,
-including the supported SynthID carrier, visible marks, open DWT-DCT, and
+including the pipeline lattice described below, visible marks, open DWT-DCT, and
 TrustMark. Metadata inspection still runs.
 
-## Detect the supported SynthID pixel carrier
+## Detect the generation-pipeline pixel lattice (experimental)
 
 ```bash
 remove-ai-watermarks detect-synthid image.png
 remove-ai-watermarks detect-synthid image.png --json
-remove-ai-watermarks detect-synthid resized.png --register-scale
+remove-ai-watermarks detect-synthid native-period.png --fixed-period
 ```
 
-The command returns one of `detected`, `not_detected`, or `unsupported`. The
-runtime detector covers one frozen periodic carrier family in the
+This route is experimental. Signed provenance, read by `identify` and confirmed
+against the provider by `verify-openai-synthid`, remains the supported way to
+establish SynthID. The command returns one of `detected`, `indeterminate`, or
+`unsupported`, and it does not detect the SynthID watermark: its statistic disappears when the image
+is cropped off the tile grid, and it changes when the generator's pipeline
+changes, so read a positive as evidence about the pipeline and never as a
+watermark claim. The JSON carries `identifies_watermark` and
+`tile_aligned_crop_required` for exactly this reason. The
+runtime detector covers one frozen periodic lattice family in the
 [calibrated image-size range](synthid.md#32-how-our-tool-detects-the-supported-carrier)
-and needs the `pixels` extra. The native default uses the fixed fold from
-1,000,000 through 10,000,000 decoded pixels and the separately challenged
-opponent-color large branch above 10,000,000 through 18,000,000 pixels when
-both sides are at least 2,048 pixels. It never resizes the input and does not
-register a carrier whose sampling period changed through spatial resampling.
-`--register-scale` enables a substantially slower bounded search over measured
-carrier periods for images from 250,000 through 10,000,000 decoded pixels, with
-both sides at least 64 pixels. It is opt-in and is not used by `identify`.
-The measured positive scale range is approximately 0.65 through 1.5; 0.5x
-resizes are not reliably detected.
-The native large branch is also codec-sensitive: same-size JPEG-95 and JPEG-90
-re-encoding reduced its seven official large positives from 7/7 to 0/7. A miss
-on a lossy re-encode is therefore inconclusive.
-It is positive-only: `not_detected` means the score stayed below this detector's
-threshold, while `unsupported` means the image geometry is outside its scope.
-Neither result proves that another SynthID epoch or payload is absent.
+and needs the `pixels` extra. The production default uses registered-v3 from
+250,000 through 10,000,000 decoded pixels with both sides at least 256 pixels.
+An opponent-registered-v1 fallback covers 1 through 10 megapixels, both sides
+at least 768 pixels, and selected carrier periods 7.9 through 12.0. Period-8
+candidates must also pass an opponent-color block-edge veto for the native JPEG
+lattice. The
+separately challenged opponent-color large-v1 branch covers above 10,000,000
+through 18,000,000 pixels when both sides are at least 2,048 pixels.
+Registered-v3 performs a bounded carrier-period search and independent split-
+patch confirmation. Its measured positive
+scale range is approximately 0.65 through 1.5. The fallback recovered 49/49
+lossless 0.5x-0.75x views from seven official positives. Its period-8 veto
+rejected 1,790 codec-lattice crossings, and 350 matched 0.5x controls produced no
+base crossing. The earlier period-band rule accepted 0/1,000 post-freeze Picsum
+controls. `identify` uses this production router.
+
+`--fixed-period` explicitly selects the faster legacy fixed-v2 diagnostic below
+10 megapixels. It does not resize or register the carrier and is not a
+production positive route. `--register-scale` forces the registered-v3 cascade,
+including its opponent fallback, on geometry where the default would select
+large-v1.
+The native large and opponent-registered branches are codec-sensitive. The
+large branch fell from 7/7 to 0/7 after same-size JPEG-95 or JPEG-90; the
+fallback retained 0/63 JPEG-95, JPEG-85, and WebP-95 views. A miss on a lossy
+re-encode is therefore inconclusive.
+It is positive-only: `indeterminate` means the score stayed below this
+detector's threshold, while `unsupported` means the image geometry is outside
+its scope. Neither result proves that another SynthID epoch or payload is
+absent.
+JSON output includes the exact reason plus provider scope, backend, pixel
+preservation, and metadata-use audit fields.
 
 ## Verify OpenAI SynthID from pixels
 
@@ -110,6 +132,16 @@ Retention, `--acknowledge-upload` is mandatory. This command is never called by
 `identify`. `not_detected` means only that OpenAI's verifier did not recognize a
 supported watermark in this file; it is not proof of human authorship.
 
+The built-in client bounds the request at 120 seconds and disables automatic
+SDK retries, so one acknowledgement cannot silently upload the media multiple
+times. A timeout, disconnect, malformed response, access failure, or rate limit
+is an error, never a negative watermark verdict. API failures expose status,
+error code, request id, `Retry-After`, and whether an explicit caller-controlled
+retry is appropriate through `OpenAIProvenanceError`; the verifier itself never
+retries an upload.
+The JSON result uses the same provider-scope, backend, pixel-preservation, and
+metadata-use audit fields as the local detector.
+
 The Python API enforces the same boundary with the required explicit intent
 flag `verify_openai_synthid(path, acknowledge_upload=True)`.
 

docs/known-limitations.md

@@ -64,22 +64,72 @@ identity or exact texture, and each then runs the same face stage.
 `qwen-zimage` is the higher fidelity of the two. Both are large, slow, and may
 still alter small text or difficult faces.
 
-### Local SynthID detection covers one image-carrier family
+### The experimental pixel route reads a pipeline lattice, not the watermark
 
-Google does not publish the proprietary SynthID payload decoder. The project
-ships a positive-only detector for one measured periodic image carrier in a
-calibrated image-size range, plus provenance-based recognition
-from Google AI C2PA or current OpenAI C2PA with an explicit watermark action.
-It does not cover images outside the selected mode's size range, crop, strong
-JPEG compression, video, or future carrier epochs. Arbitrary dimensions inside
-the default range are supported only while the recovered carrier retains its
-measured 16-pixel sampling lattice. A spatial resize changes that period. The
-opt-in `detect-synthid --register-scale` mode searches a bounded
-fractional-period range for images from 250,000 through 10,000,000 pixels, but
-requires both sides to be at least 64 pixels. Its measured positive range is
-approximately scale 0.65 through 1.5; a 0.5 resize aliases the carrier with
-ordinary resize lattices and is not reliably detected. It is also slower. A
-`not_detected` or `unsupported` result is not a negative universal verdict, and
+Signed provenance is the primary and supported route for SynthID in this
+project; the pixel route described here is experimental. Google does not
+publish the proprietary SynthID payload decoder, and nothing
+in this project reproduces one. What the local pixel route detects is a
+periodic lattice anchored at the image origin: it is destroyed by a crop of
+seven pixels on the large branch and by a crop of two pixels on the
+ordinary-size registered-v3 route, whereas the published SynthID evaluation
+retains 99.97% of its detection rate under aggressive crop and resize. A
+positive therefore
+identifies the generation pipeline that leaves the lattice, and says nothing
+about whether the image carries a watermark. It also changes when a provider
+changes pipeline: output from `gemini-2.5-flash-image` scores at the level of
+unrelated photographs. The project additionally offers provenance-based
+recognition from Google AI C2PA or current OpenAI C2PA with an explicit
+watermark action, and that provenance route is the only one that speaks about
+a watermark at all.
+
+The ordinary-size route shares the same phase lock. On 2026-08-16 a two-pixel
+diagonal crop removed all 28 in-geometry foreign-generator detections and all 8
+detected Google provenance positives from registered-v3, with the statistic
+recovering only at crop offsets that are multiples of four. The route therefore
+fires on other vendors' images because those pipelines leave a matching
+origin-anchored lattice, not because a watermark is present.
+
+The published control rates for this route were all measured on photographs.
+Against other generators' output, which is the population a provenance tool is
+actually pointed at, the shipped runtime accepted 29 of 223 signed non-Google
+images on 2026-08-16: 13.0% overall, 24.1% on Adobe Firefly, with a top score
+of 3.01 against a threshold of 1.0. Treat a positive as evidence only when the
+alternative generators have been ruled out by other means.
+
+Sensitivity has only ever been reported on images the route was calibrated
+against. The first out-of-distribution measurement was taken on 2026-08-16: 11
+images generated at 5632x3072 by `gemini-3.1-flash-image`, a model that did not
+exist when the large branch was calibrated, written straight from the API with
+no lossy-codec history and watermarked by Google's documented policy that all
+generated images carry SynthID. Eight were detected and three returned
+indeterminate, a rate of 72.7% with a one-sided 95% lower bound of 43.6%. The
+sample is small and one provider wide, so read it as an order of magnitude
+rather than an operating point.
+
+Cropping the same images to 4096x2560 on the tile grid retained 4 of 6. Taking
+the identical crop seven pixels off the grid retained 0 of 6, with every score
+negative. A crop that is not tile-aligned removes the signal entirely, which is
+the sharpest way to see that this route does not read the watermark: the
+published SynthID evaluation keeps 99.97% of its detection rate under
+aggressive crop and resize.
+
+The route's declared scope is narrower still. It does not cover images outside
+the selected mode's size range, crop, strong
+JPEG compression, video, or future carrier epochs. The production default
+searches a bounded fractional-period range from 250,000 through 10,000,000
+pixels and requires both sides to be at least 256 pixels. Its measured positive
+range is approximately scale 0.65 through 1.5. A precision-first opponent-color
+fallback covers only 1 through 10 megapixels, sides of at least 768 pixels, and
+periods 7.9 through 12.0; it recovered the measured lossless 0.5x-0.75x cohort.
+At period 8, an additional block-edge veto rejects the ordinary JPEG lattice;
+this does not make arbitrary crops or codecs supported. The fallback retained
+0/63 lossy JPEG/WebP views, so a lossy miss remains inconclusive. Above 10
+through 18
+megapixels, the separate large-v1 route requires native, lossless pixels and
+both sides at least 2,048 pixels. The fixed-v2 branch is retained only behind
+the explicit `detect-synthid --fixed-period` diagnostic. A
+`indeterminate` or `unsupported` result is not a negative universal verdict, and
 removal still requires the matching provider oracle for confirmation.
 
 For important outputs:

docs/module-internals.md

@@ -451,7 +451,7 @@ metadata extraction from verdict logic:
   both extractors reach the same answer. It did not, and the record path silently
   reported no SynthID for images the file path flagged.
 - `identify` preserves the path-based API and adds the optional registered
-  visible-mark, open invisible-watermark, and supported SynthID carrier
+  visible-mark, open invisible-watermark, and generation-pipeline lattice
   detectors after extraction.
 
 ### SynthID periodic carrier detector
@@ -468,17 +468,29 @@ ceiling does not require multiple three-channel float workspaces. The model hash
 is pinned by a test, and the unchanged operating threshold is
 `0.17357069773071196` through 10 megapixels.
 
-The direct API returns `detected`, `not_detected`, or `unsupported`; the last is
-distinct because no resize is performed. The fixed branch is selected from
-1,000,000 through 10,000,000 decoded pixels. The frozen threshold accepted
+That fixed threshold is not production-qualified. A later source-fresh
+Open Images test-split challenge produced 5 crossings among the 211 images in
+its supported geometry. A precision-first `0.28` replacement is frozen as a
+research candidate after retaining all 12 available source-diverse native
+positives and rejecting those five crossings. It then failed the second
+untouched holdout at 1/213, with score `0.322542963`. The replacement is
+rejected and fixed-v2 remains an explicit diagnostic only.
+
+The direct API returns `detected`, `indeterminate`, or `unsupported`. Passing
+`register_scale=False` selects fixed-v2 from 1,000,000 through 10,000,000
+decoded pixels as an explicit diagnostic. Its frozen threshold accepted
 none of 5,000 public COCO views balanced across every observed target geometry,
 and none of a separate 5,000-view challenge over 256 generated geometries
 covering every pair of modulo-16 edge remainders. The original 2048x2048
 verdicts and exact scores remain unchanged. Runtime matches do not attribute a
 provider. `identify` adds only positive matches as high-confidence
 evidence and never turns a local negative into a clean verdict.
+The result envelope also names the signal family, provider scope, backend,
+whether metadata contributed to the verdict, whether pixels were preserved,
+and an explicit reason for unsupported or indeterminate results. These fields
+are shared with the official OpenAI verifier's JSON boundary.
 
-The native default selects `synthid-periodic-tile-large-v1` above 10 through 18
+The production router selects `synthid-periodic-tile-large-v1` above 10 through 18
 megapixels when both dimensions are at least 2,048 pixels. It evaluates all
 phase-aligned 2,048-square windows and combines the minimum fixed-template,
 Red-minus-Green, and Blue-minus-Yellow spatial correlations with the most
@@ -532,7 +544,67 @@ final controls. Multiplying the canonical score by spectral-period confidence
 also accepted two. That baseline was rejected rather than shipped at its
 discovery threshold.
 
-The opt-in `register_scale=True` path is the later calibrated successor. It
+The research-only
+[`synthid_affine_lattice_probe.py`](../scripts/synthid_affine_lattice_probe.py)
+adds split-confirm synchronization. It estimates complex harmonic coherence on
+one checkerboard of patches, confirms the selected period on the other, and
+reports amplitude-aware confirmation, a locally content-whitened multichannel
+code match, and phase-preserving and cyclically registered template scores. A
+0.1-pixel grid recovered the expected
+12.8, 14.4, 17.6, and 19.2 periods in 20/20 transformed views from five official
+positive parents. A provisional conjunction accepted none of 800 corresponding
+views from 200 oracle-negative parents, none of 469 OpenAI-labeled rows, and
+three of 276 broad non-Google rows whose TC260 or Samsung provenance prevents
+treating them as clean oracle negatives. This probe is not runtime routing:
+positive diversity is still inadequate, period 8 remains rejected, and the
+amplitude stage currently handles isotropic scale at zero rotation only. Full
+protocol and caveats are in the detector research plan.
+
+The whitened match uses neighboring noncarrier bins to estimate complex
+Green/opponent-color covariance around every selected harmonic. It corrected a
+period-24 alias on one native official positive, giving the broad native search
+the correct period 16 on all five ordinary-size positives. It was retained only
+as a candidate reranker: its positive-to-negative margin was smaller than the
+existing spatial-template margin in both the native pilot and a locked 0.8
+resize challenge. Two of the three TC260/Samsung-provenance challenge crossings
+also matched the full whitened code more strongly than the weakest official
+large positive, so the pixel result identifies a compatible signal family, not
+a provider.
+
+The same probe also exposes a payload-agnostic H5 confirmation. It estimates a
+complex harmonic vector independently on the two checkerboards and measures
+their fixed relative-phase inner product; the codeword need not match the known
+template. Free cyclic-shift selection was rejected because natural phase aliases
+overlapped the controls. The fixed all-harmonic statistic separated five native
+official positives from 130 controls with margin `0.1619`, and five 0.8-resized
+views from 200 fresh-parent controls with margin `0.1541`; all seven large
+official positives also passed the observed gap. It remains research-only
+because these tests supplied the period, reuse one negative source family, and
+contain only 12 independent positive parents.
+
+The separate research-only
+[`synthid_cyclostationary_probe.py`](../scripts/synthid_cyclostationary_probe.py)
+measures full complex cross-channel spectral correlation at carrier shifts
+against neighboring-shift same-image nulls. It recognizes synthetic amplitude
+modulation and rejects both independent noise and a purely additive carrier.
+The real native pilot overlapped completely: five official-positive joint
+contrasts ranged from `-0.0027` to `0.0337`, while 130 controls reached `0.0593`.
+The cyclostationary branch is rejected rather than tuned on new frequency bands.
+
+The affine-lattice research CLI also supports symmetric in-memory JPEG and
+top-left crop challenges. At fixed native period 16, codec-specific
+phase-preserving template thresholds retained all five repeated positive views
+and accepted none of 80 locked controls at JPEG qualities 95, 90, and 80. The
+locked margins declined from `0.1054` to `0.0257`; whitened and unknown-codeword
+scores were rejected for lossy JPEG. A separate registered crop conjunction
+retained all five views and accepted none of 80 controls after both 5% and 10%
+top-left crops, with registered-template margins above `0.22`. These are
+research pilots, not runtime experts: the period was supplied, positive parents
+were reused, and only one codec implementation and one negative family were
+tested.
+
+The `register_scale=True` path is the calibrated ordinary-size production
+expert. It
 keeps the harmonic search, adds three-level scale-space consistency and quadrant
 agreement. Its calibrated decision requires the canonical winner to be the
 strongest spectral-period candidate, its three-way amplitude to cross the
@@ -556,11 +628,51 @@ accepted seven, so fixed and registered results cannot safely be unioned. In
 overlapping geometry the registered decision remains the validated path;
 fixed-only evidence is a diagnostic rather than a universal-cascade positive.
 
-The research-only router in `scripts/synthid_routed_expert_bank.py` encodes that
-precedence and always abstains on fixed-only evidence. Its three-observation
-schema keeps the fixed, registered, and large identities explicit. Registered
-and large crossings are positive routes only in their disjoint calibrated
-ranges; the bank never returns a clean-image verdict.
+A later source-fresh 3,000-image Open Images test-split challenge superseded the
+registered-v2 precision claim: it accepted 4 of 2,997 supported controls. The
+registered-v3 confirmation uses disjoint patch groups for phase, cyclic
+amplitude, and held-out-codeword evidence. It rejected all four crossings,
+retained 147 of 148 base-accepted source-diverse positives, and retained all 359
+base-accepted views in a dense 0.65-1.50 transform matrix over 12 independent
+parents. Frozen unchanged, it then accepted 0/2,996 controls from a second
+nonoverlapping Open Images cohort and 0/2,366 supported controls from a
+3,000-image COCO second-family challenge. Registered-v3 is now the default
+ordinary-size positive route. The exact gates, acquisition hashes, and rejected
+weak-signal rescue are recorded in the detector research plan.
+
+Every one of those control rates is photographic. Against 223 corpus images
+whose C2PA names a non-Google generator, the unchanged entry point accepted 29
+(`0.130`, Adobe Firefly `0.241`, highest foreign score `3.01`), all from
+registered-v3. The branch reads a lattice shared across generation pipelines,
+which is why `identify` reports it as the experimental `pipeline_lattice`
+signal and never as a watermark.
+
+The branch is also phase-locked to the image origin, exactly like the large
+expert. A two-pixel diagonal crop killed all 28 in-geometry foreign detections
+and all 8 detected Google provenance positives (maximum remaining scores
+`0.779` and `0.311` against the `1.0` threshold); the signal recovers only at
+offsets that are multiples of four. Registered-v3 therefore detects the same
+crop-destroyed generation-pipeline lattice as `large-v1`, on ordinary sizes.
+
+When registered-v3 abstains, opponent-registered-v1 searches the same frozen
+template in Red-minus-Green and Blue-minus-Yellow space. It reranks three
+separated scale candidates with fixed RGB and two spatial opponent-color gates,
+then accepts only periods 7.9-12.0 on 1-10 megapixel rasters whose sides are at
+least 768 pixels. Period-8 candidates additionally require Red-Green and
+Blue-Yellow 8-pixel edge ratios no greater than `1.05`; this vetoes the
+deterministic JPEG block lattice without using container metadata. The final
+rule recovered 49/49 lossless 0.5x-0.75x views from seven official-positive
+parents. It rejected all 1,790 measured period-8 codec crossings, while 350
+identically resized controls had no base crossing and the earlier period-band
+rule accepted 0/1,000 post-freeze Picsum controls. Period 12.8 remains excluded,
+and lossy JPEG/WebP views remain inconclusive.
+
+The runtime precedence is registered-v3, the bounded opponent fallback, then
+large-v1 above 10 megapixels. Passing
+`register_scale=False`, or the CLI's `--fixed-period`, selects the legacy fixed
+diagnostic explicitly. The research bank in
+`scripts/synthid_routed_expert_bank.py` keeps all three observations for audits.
+Neither runtime nor research routing returns a clean-image verdict.
 
 An unchanged registered challenge from 10 to 18 megapixels retained only 1 of
 37 Google candidates and zero of 89 non-Google controls. Twenty-eight positives
@@ -635,6 +747,17 @@ period-8 model retained 5 of 158 and zero controls, then 1 of 40 fresh OpenAI
 rows and 1 of 50 fresh controls. They neither explain the known oracle-positive
 miss nor justify runtime routing.
 
+An image-level patch-bag model over nine normalized high-pass patches was the
+strongest later local OpenAI classifier. It accepted 324/1,364 development-test
+OpenAI rows with 4/3,373 heterogeneous controls, then transferred to 120/486
+OpenAI rows, 8/504 general controls, and 5/33 same-provider non-assertion rows
+in the later temporal challenge. Assertion enrichment over the same-provider
+stratum was not independently significant (`p = 0.151`, one-sided exact test).
+JPEG-95 reduced the unchanged conjunction to
+1/486 positives and zero controls. The native temporal hits were complementary
+to the signed period-8 hits, but the same-provider rate and codec collapse
+identify another export noiseprint rather than a runtime SynthID expert.
+
 The separate OpenAI period-8 DTCWT component is persistent rather than tied to
 one short rollout: exact-generator asserted hits were 3/16, 46/365, and 26/200
 from May through July, with 75/581 overall versus 1/52 same-generator rows
@@ -655,12 +778,12 @@ development positives and finished at 0/50 positives, 0/199 controls, and AUC
 0.374. Neither branch is part of runtime routing; full split and oracle details
 are in the detector research plan.
 
-The separately measured geometry range remains 250,000 through 10,000,000
-decoded pixels with both sides at least 64 pixels. The default path and
-`identify` remain native-only and select either the fixed or large branch by
-geometry; scale registration stays opt-in. A 20-image real-corpus drift check was
-byte-identical after integration. The calibration history and caveats are in the
-linked detector research plan.
+The separately measured registered geometry range remains 250,000 through
+10,000,000 decoded pixels with both sides at least 256 pixels. The default path
+and `identify` use registered-v3, then the narrower opponent-registered-v1
+fallback in its 1-10 megapixel domain, and large-v1 above 10 megapixels. A
+20-image real-corpus drift check was byte-identical after the earlier v2 integration.
+The calibration history and caveats are in the linked detector research plan.
 
 ### Official OpenAI SynthID verifier
 
@@ -681,6 +804,19 @@ multipart file to `content_provenance_checks.create` and parses exactly one
 used as fallback evidence. Missing, duplicate, or unknown SynthID outcomes are
 errors rather than negative detections.
 
+The default SDK client has a 120-second request timeout and zero automatic
+retries. One upload acknowledgement therefore authorizes at most one media
+transmission rather than inheriting the SDK's retry default. Request logs keep
+the endpoint, temporary basename, media type, byte count, timeout, retry policy,
+duration, HTTP status, error code, and request id when available, but omit the
+source path, image bytes, credentials, and decoded-pixel fingerprint.
+
+`OpenAIProvenanceError` preserves the status, API error code, request id,
+`Retry-After` value, and a transient-only `retryable` flag. The library does not
+automatically act on that flag: an explicit caller invocation is required for
+every additional upload. Transport and schema failures remain errors rather
+than becoming `not_detected` or a local detector result.
+
 The result remains provider-scoped and positive-evidence-only. `not_detected`
 does not mean human-created, and the official endpoint's published prohibition
 on repeated reverse-engineering or evasion queries prevents using this backend

docs/python-api.md

@@ -12,7 +12,7 @@ path that still runs on CPU and combines `video` and `diffusion`. Add `heif`
 independently when path-based pixel APIs must decode HEIC, HEIF, or AVIF. See
 the complete [feature-extra matrix](installation.md#feature-extras).
 
-## Detect the supported SynthID carrier
+## Detect the generation-pipeline lattice (experimental)
 
 Install `remove-ai-watermarks[pixels]`, then call the lazy top-level API:
 
@@ -20,26 +20,48 @@ Install `remove-ai-watermarks[pixels]`, then call the lazy top-level API:
 import remove_ai_watermarks as raiw
 
 result = raiw.detect_synthid("input.png")
-print(result.status)     # "detected" | "not_detected" | "unsupported"
+print(result.status)     # "detected" | "indeterminate" | "unsupported"
 print(result.score)      # float for a supported image size, otherwise None
 print(result.threshold)  # frozen operating point
+print(result.reason)     # support or non-detection reason, otherwise None
 
-# Opt in when the image may have been spatially resized.
-registered = raiw.detect_synthid("resized.png", register_scale=True)
+# Force the legacy native-period diagnostic only when auditing that expert.
+fixed_diagnostic = raiw.detect_synthid("native-period.png", register_scale=False)
 ```
 
-The detector is positive-only and covers one measured periodic carrier family
+This is experimental and is not a watermark detector. Signed provenance through
+`identify` is the supported SynthID route. Its statistic is destroyed by a seven-pixel
+crop while the published SynthID evaluation survives aggressive crop and
+resize, so a positive identifies the generation pipeline. The result exposes
+`identifies_watermark` and `tile_aligned_crop_required` so a caller cannot
+reach the wrong conclusion from `status` alone. It is positive-only and covers
+one measured periodic lattice family
 in the [calibrated image-size range](synthid.md#32-how-our-tool-detects-the-supported-carrier).
-Arbitrary dimensions are accepted inside the default range, but spatial
-resampling can change the carrier period. `register_scale=True` enables the
-slower scale-registered operating point over 250,000 through 10,000,000 decoded
-pixels, with both sides at least 64 pixels; the default remains the native-period
-detector used by `identify`. Its score is a normalized multi-gate statistic with
-a threshold of `1.0`, not the native detector's raw template correlation. Scale
-0.5 is outside its reliable positive range.
-`not_detected` means only that the selected model did not find its carrier;
-`unsupported` is kept separate from a negative result. Neither is proof that
-the image contains no SynthID watermark.
+The default and `identify` use registered-v3 over 250,000 through 10,000,000
+decoded pixels, with both sides at least 256 pixels. An opponent-registered-v1
+fallback covers 1 through 10 megapixels, sides of at least 768 pixels, and
+periods 7.9 through 12.0; period-8 candidates also require the opponent-color
+block-edge codec veto. Large-v1 covers 10 through 18 megapixels. Each score
+uses a threshold of `1.0`; registered-v3 also requires independent split-patch
+confirmation. The fallback is qualified for measured lossless 0.5x-0.75x
+views, not lossy retranscodes.
+`register_scale=True` forces the registered-v3 cascade, including its bounded
+opponent fallback, while `False` explicitly selects the legacy native-period
+fixed-v2 diagnostic below the large-image boundary.
+`indeterminate` means only that the selected local model did not find its
+carrier; `unsupported` means no local expert covers the input geometry. Neither
+is proof that the image contains no SynthID watermark.
+Both local and official OpenAI JSON results expose `signal_family`,
+`provider_scope`, `backend`, `metadata_used_for_verdict`, and
+`pixels_preserved`; the local result adds `identifies_watermark` and
+`tile_aligned_crop_required`, so callers do not need to infer the evidence boundary from
+the detector name.
+
+Remote transport and response failures raise `OpenAIProvenanceError`. Its
+`status_code`, `error_code`, `request_id`, `retry_after`, and `retryable`
+attributes let a caller implement bounded backoff or a circuit breaker without
+turning an API outage into a false `not_detected` result. One function call still
+performs at most one upload.
 
 ## Remove visible marks
 
@@ -330,8 +352,9 @@ Timings and spatial artifacts are opt-in. Artifacts include image-identifying da
 such as a thumbnail and perceptual hash; aggregate feature families do not.
 
 `identify_from_evidence` does not reopen the source file by default: it evaluates
-metadata only, and registered visible marks and pixel-backed invisible watermarks
-remain in the path-based `identify` call.
+metadata only, and the pixel-backed checks remain in the path-based `identify`
+call: registered visible marks, open invisible-watermark decoders, and the
+experimental generation-pipeline lattice signal.
 
 Pass `image_path` together with `check_visible` or `check_invisible` to add those
 pixel detectors on top of the SAME evidence. That is how a caller asking one file

docs/supported-signals.md

@@ -79,7 +79,9 @@ The inspection and stripping code handles signals in these groups:
 - xAI and Grok EXIF signature fields;
 - Samsung AI editing markers;
 - Hugging Face job metadata;
-- one positive-only SynthID periodic pixel carrier in a calibrated image-size range;
+- one positive-only generation-pipeline pixel lattice in a calibrated image-size
+  range, experimental, which identifies the pipeline and not the SynthID
+  watermark; signed provenance remains the supported SynthID route;
 - open Stable Diffusion style DWT-DCT watermarks with the `detect` extra;
 - Adobe TrustMark with the `trustmark` extra.
 
@@ -127,32 +129,51 @@ The `controlnet`, `sdxl`, `qwen` and `default` values were removed. A retired na
 is rejected at parse time rather than remapped onto a surviving profile.
 
 Google does not publish the SynthID payload decoder. This project ships a
-positive-only detector for one measured periodic image-carrier family in a
+positive-only detector for one measured periodic image-lattice family in a
 calibrated image-size range, available through `detect-synthid`
 and the default pixel pass in `identify` when the `pixels` extra is installed.
-The unchanged fixed threshold accepted none of the public COCO views in both
-an observed-geometry challenge and a generated-geometry challenge covering all
-modulo-16 edge cases. Above 10 through 18 megapixels, the native default uses a
+That lattice is not the watermark. It is anchored at the image origin: a
+seven-pixel crop removes it from the large branch and a two-pixel crop removes
+it from registered-v3 (all 36 tested detections across foreign-generator and
+Google images), while the published SynthID evaluation survives aggressive crop
+and resize, so every control rate below describes a generation-pipeline
+signature and not watermark detection. Every rate quoted
+below was also measured on photographs; on 223 signed non-Google generator
+images the same runtime accepted 29, a rate of 13.0%. Sensitivity outside the
+calibration distribution was measured once, on 11 fresh 5632x3072 images from
+`gemini-3.1-flash-image`: 8 detected, 72.7%, one-sided 95% lower bound 43.6%.
+The same images cropped seven pixels off the tile grid returned 0 of 6.
+The default ordinary-size route uses registered-v3, including independent
+split-patch phase and codeword confirmation. It accepted none of 5,993
+supported controls across two nonoverlapping Open Images test cohorts and none
+of 2,366 supported controls in a second-family COCO challenge. A precision-first
+opponent-registered-v1 fallback covers 1 through 10 megapixels, sides of at
+least 768 pixels, and carrier periods 7.9 through 12.0. Period-8 candidates must
+also pass an opponent-color block-edge codec veto. It recovered 49/49 lossless
+0.5x-0.75x views from seven separate official positives. The veto rejected all
+1,790 measured period-8 codec crossings, 350 matched 0.5x controls had no base
+crossing, and the earlier period-band rule accepted 0/1,000 post-freeze
+controls. Above 10 through
+18 megapixels, the production router uses a
 separately challenged large branch over phase-aligned windows and opponent-color
 phase agreement; both sides must be at least 2,048 pixels. It retained all seven
 officially verified large Google pixel positives and accepted none of 2,637
 feature-unseen, decoded-pixel-unique natural controls. A smaller post-freeze
-Open Images acquisition also produced 0/41 detections. Arbitrary dimensions in
-the default calibrated ranges are accepted, but the input must retain the
-measured 16-pixel carrier scale. The large branch retained 0/7 official
-positives after either JPEG-95 or JPEG-90 re-encoding, so its native-size scope
-does not include lossy retranscodes. The
-opt-in `detect-synthid --register-scale` mode performs a slower bounded scale
-search over its separately measured 250,000-through-10,000,000-pixel range and
-requires both sides to be at least 64 pixels. Its measured positive scale range
-is approximately 0.65 through 1.5; 0.5x resizes are not reliably detected.
-`identify` does not enable that search. Neither detector attributes a provider
-locally.
+Open Images acquisition also produced 0/41 detections. Registered-v3 has a
+measured scale range of approximately 0.65 through 1.5; the narrower fallback
+adds the measured lossless 0.5x-0.75x range. The large branch retained 0/7
+official positives after either
+JPEG-95 or JPEG-90 re-encoding, and the opponent-registered fallback retained
+0/63 JPEG-95, JPEG-85, and WebP-95 views. Their scope does not include lossy
+retranscodes. `detect-synthid --fixed-period` exposes
+fixed-v2 only as a legacy diagnostic; its fresh-source false-positive rate
+disqualified it as a production positive route. No local production expert
+attributes a provider.
 
 The tool also recognizes presence from supported provenance: Google AI C2PA
 under Google's all-media watermark policy, and current OpenAI C2PA carrying an
 explicit `c2pa.watermarked.*` action. Legacy OpenAI C2PA without that action
-does not assert SynthID. A pixel result of `not_detected` or `unsupported`
+does not assert SynthID. A local pixel result of `indeterminate` or `unsupported`
 remains inconclusive for other sizes, epochs, codecs, and payloads.
 
 The optional `verify-openai-synthid` command is a separate official remote

docs/synthid.md

@@ -585,21 +585,32 @@ codecs, video, or future epochs. The evidence and gates are documented in
 
 ### 3.2 How our tool detects the supported carrier
 
-`remove-ai-watermarks detect-synthid image.png` folds the image residual modulo
-16x16 and compares it with a frozen float64 template. It evaluates only native
-input, without resize. Exact-multiple dimensions retain the original folding
-path; non-divisible dimensions use count-correct modulo folding. The model and
+The heading is kept because README, `cli.md` and `python-api.md` link to this
+anchor, but the name is inherited and inaccurate. What the local experts read
+is a periodic lattice anchored at the image origin, measured on 2026-08-16 to
+vanish under a seven-pixel crop that the published SynthID evaluation survives
+at 99.97% TPR. Everything in this section describes that pipeline signature,
+not watermark recovery; the measurement is in the empirical log of
+[`synthid-detector-removal-plan.md`](synthid-detector-removal-plan.md).
+
+`remove-ai-watermarks detect-synthid image.png --fixed-period` exposes the
+legacy fixed-v2 diagnostic. It folds the image residual modulo 16x16 and
+compares it with a frozen float64 template. It evaluates only native input,
+without resize. Exact-multiple dimensions retain the original folding path;
+non-divisible dimensions use count-correct modulo folding. The model and
 threshold remain frozen from the 2048x2048 experiment. Through 10 megapixels,
 the fixed threshold
 accepted none of 5,000 public COCO views balanced across every observed target
 geometry. A separate 5,000-view challenge used 256 generated geometries from one
 through 18 megapixels and covered every pair of width/height remainders modulo
-16; it also produced no accepted view. Fixed-branch support is therefore the
-challenged interval of 1,000,000 through 10,000,000 decoded pixels rather than
-an enumerated width-by-height registry. The original 2048x2048 scores remain
-exactly unchanged.
+16; it also produced no accepted view. Later fresh-source challenges invalidated
+that precision claim: fixed-v2 accepted 5/211 supported controls in one Open
+Images cohort, and a proposed `0.28` threshold still accepted 1/213 in the
+second. The branch is therefore diagnostic only and is never unioned into the
+production positive route. The original 2048x2048 scores remain exactly
+unchanged.
 
-Above 10,000,000 through 18,000,000 decoded pixels, the native default selects
+Above 10,000,000 through 18,000,000 decoded pixels, the production default selects
 the separate `synthid-periodic-tile-large-v1` branch when both sides are at
 least 2,048 pixels. It scores every phase-aligned 2,048-square window without
 resizing. Every window must retain the combined template, Red-minus-Green, and
@@ -646,15 +657,19 @@ less conservative per-period thresholds accepted five final controls. The
 runtime therefore detects arbitrary decoded dimensions only when the carrier
 retains its measured 16-pixel scale.
 
-`remove-ai-watermarks detect-synthid image.png --register-scale` opts into the
-separately calibrated scale-registered detector. It samples the 30 strongest
+Through 10,000,000 decoded pixels, `remove-ai-watermarks detect-synthid
+image.png` selects the registered-v3 detector by default. `--register-scale`
+forces this branch even where the router would otherwise select large-v1. The
+registered expert samples the 30 strongest
 template harmonics over fractional periods, reconstructs the three strongest
 canonical candidates, and combines full-frame correlation with quadrant and
 three-level scale-space consistency. A positive result additionally requires
 the canonical winner to be the strongest spectral candidate, a calibrated
 threshold for its recovered period, and agreement in two independent
-high-frequency template bands. The reported score is the minimum normalized
-margin across these requirements; its threshold is `1.0`.
+high-frequency template bands. Registered-v3 then confirms base crossings on
+disjoint patch groups using origin-corrected harmonic phase, cyclic amplitude,
+and a held-out unknown-codeword statistic. The reported score preserves the
+base normalized margin only after confirmation; its threshold is `1.0`.
 
 The final rule retained 229 of 355 source-disjoint transformed positives and
 accepted none of 499 previously untouched Open Images controls. It retained
@@ -662,8 +677,9 @@ accepted none of 499 previously untouched Open Images controls. It retained
 A separate period-8 rescue was rejected because symmetric 0.5x control resizing
 reproduced the same lattice and fully overlapped the positive feature range.
 The measured input range is 250,000 through 10,000,000 decoded pixels.
-Both dimensions must also be at least 64 pixels so every canonical quadrant can
-retain one 16-by-16 carrier tile.
+Both dimensions must also be at least 256 pixels because the frozen positive
+confirmation requires complete disjoint 256-pixel patch groups. Narrower
+geometries need a separately calibrated adaptive-patch expert.
 
 A later frozen 1,000-image Open Images reserve preserved zero registered
 detections. Only 81 of those images were large enough for the fixed expert, and
@@ -673,12 +689,151 @@ is the validated decision and a fixed-only crossing remains ambiguous. The
 fixed expert is retained as a separately scoped carrier diagnostic, not as an
 independent recall addition to registered mode.
 
-The research router in `scripts/synthid_routed_expert_bank.py` makes this policy
-explicit and always abstains on fixed-only evidence. It records fixed,
-registered, and large observations separately: the registered threshold is the
-positive route in its measured scale-search range, while the large threshold is
-the positive route in its disjoint native large-image range. It never claims
-absence because encoder-version coverage remains incomplete.
+Registered-v2 later accepted 4/2,997 supported controls in a source-fresh Open
+Images test cohort. The frozen v3 confirmation rejected all four while retaining
+147/148 base-accepted source-diverse positives and all 359 base-accepted views
+in a dense 0.65-1.50 scale matrix. Without changing the rule, v3 then accepted
+0/2,996 controls from a second nonoverlapping Open Images cohort and 0/2,366
+supported controls from a 3,000-image COCO second-family challenge. The two Open
+Images cohorts give a one-sided 95% zero-error false-positive upper bound of
+about 0.050%. These results qualify a precision-first positive route, not an
+absence verdict; the v2 sensitivity benchmark and scale-0.5 exclusion remain.
+
+Those control rates are photographic. On 223 corpus images whose C2PA names a
+non-Google generator, the unchanged production entry point accepted 29, a rate
+of `0.130`, with `0.241` on Adobe Firefly and a highest foreign score of `3.01`
+against the `1.0` threshold. Every acceptance came from registered-v3. The
+lattice this branch reads is therefore shared across generation pipelines, and
+the photographic upper bound above does not describe the rate a user meets on
+AI-generated input. The branch is also phase-locked to the image origin: a
+two-pixel diagonal crop killed all 28 in-geometry foreign detections and all 8
+detected Google provenance positives, with the signal recovering only at
+offsets that are multiples of four. A mark that survived cropping would not do
+that; the route detects pipeline lattice on ordinary sizes as well as large
+ones.
+
+An exact runtime-parity check subsequently applied the recorded transform to
+each of the 226 source-diverse assignments before invoking registered-v3. Two
+half-scale outputs became too small for the research confirmation grid; 224
+were scored. Registered-v3 detected 147/224 overall. All 40 supported 0.5x
+views missed, while the measured 0.65x-1.5x range detected 147/184 (79.9%). The
+unchanged registered-v2 base accepted 148 of those 184; v3 confirmation removed
+only one true base crossing. The remaining sensitivity loss is therefore
+period selection, amplitude, or high-band failure after resampling, not an
+overly strict confirmation gate.
+
+A preregistered same-image-null follow-up compares the selected carrier on the
+two patch checkerboards with 16 neighboring periods from plus or minus 0.35 to
+2.0 pixels. On two combined Open Images development cohorts, a near-threshold
+base floor of 0.8, the existing frozen confirmation, and minimum joint
+coherence excess `0.05695968150082637` rescued 7 of the 36 sub-threshold
+0.65x-1.5x views. The frozen rule accepted none of the first 3,000 COCO
+controls. It remains research-only until the untouched final control partition
+and new positive parents pass unchanged. The final 2,000 COCO files contained
+1,598 registered-supported images and seven near-threshold candidates. Four
+candidates could not form the required two 256-pixel patch groups; the three
+scored candidates had same-image excess at most zero. The frozen recovery rule
+therefore accepted 0/1,598, as did registered-v3 itself. The negative-side gate
+passed, but the lack of unseen positive parents still keeps the rescue outside
+runtime.
+
+Same-image contrast also failed to rescue spatial/spectral period mismatches.
+Across the two 3,000-file Open Images cohorts, 469 controls passed the
+preregistered mismatch prefilters and 212 passed frozen split confirmation.
+Their maximum joint excess, `0.1762448251`, exceeded the `0.1464860382` maximum
+of the five eligible transformed positive mismatches. The requirement that the
+canonical and spectral period winners agree remains part of registered-v3.
+
+A later content-adaptive research branch uses the registered tile only as a
+synchronization basis. It standardizes every patch against all cyclic tile
+shifts, chooses one phase on a checkerboard of selection patches, and confirms
+the same phase on the disjoint checkerboard. A frozen joint standardized score
+of `2.5`, combined with matching periods, the existing confirmation, both base
+gate margins at least `0.45`, and one margin at least `1.0`, recovered 5 of 18
+scored amplitude/high-band failures. It accepted none of 88 first-cohort gate
+failures and none of 334 previously scored control candidates, but those are
+development results. Its separately started Picsum feature-holdout then
+accepted 0/350 decoded-pixel-unique controls unchanged. Only two controls
+reached the mechanism pre-gates and their maximum joint patch-shift score was
+`0.7029283` against `2.5`. The branch remains research-only pending unseen
+positive parents. That check then failed: 21 resized views at 0.65, 0.70, and
+0.75 from seven separate official pixel-positive large Gemini parents produced
+0/21 registered-v3 detections and 0/21 patch-phase rescues. Their surviving
+amplitude but inverted or weak RGB high-band response points to a different
+carrier epoch or representation. The patch branch is not promoted; the next
+test registers the separately measured large-v1 opponent-color signal.
+
+The first opponent-color registration experiment then recovered 21/21 of
+those views and selected the exact transformed period every time. It searches
+7.5-14.5 pixels in the Red-minus-Green and Blue-minus-Yellow spectrum, then
+requires fixed RGB, Red-minus-Green spatial, and Blue-minus-Yellow spatial
+scores of at least `0.16`, `0.60`, and `0.55`. The same frozen candidate
+accepted 0/21 symmetrically resized views from seven large Open Images
+controls. The frozen candidate then accepted 0/350 decoded-pixel-unique Picsum
+controls with a maximum decision margin of `0.7778322`.
+
+The unrestricted branch was not production-safe. It accepted 605/2,997 and
+587/2,996 controls from two wider Open Images cohorts; almost every crossing
+selected the period-8 or period-12.8 codec lattice. It also accepted one of
+3,964 COCO controls at period 10, below one megapixel, and 28/3,772
+non-Google/non-OpenAI Spaces controls, all at period 8 or 12.8. Lossy
+re-encoding destroyed the measured positive route: raw lossless resize retained
+21/21 views at scales 0.5, 0.625, and 0.75, while JPEG-95, JPEG-85, and WebP-95
+retained 0/63.
+
+Those failures defined a narrower precision-first expert before a new control
+holdout: 1-10 megapixels, both sides at least 768 pixels, and a selected period
+from 8.1 through 12.0. It keeps the original three score thresholds and does
+not require the strongest spectral alias to equal the spatially reranked
+period. The latter restriction was removed after a public-API test showed that
+it reduced the 0.55x cohort from 7/7 to 1/7 while the geometry gate already
+excluded its only reserve control crossing. The final rule then accepted
+0/1,000 untouched Picsum controls, with a maximum unrestricted decision margin
+of `0.7953734`, and the production API detected 42/42 lossless views at scales
+0.55 through 0.75 from the seven official-positive parents. It now ships as
+`synthid-periodic-tile-opponent-registered-v1` after registered-v3 abstains.
+
+A later period-8 codec-confound experiment extended that frozen expert without
+mistaking the JPEG grid for a watermark. Half-scale lossless views retained 7/7
+official positives at period 8, but ordinary JPEG controls produced the same
+large template correlation. Their decoded opponent-color edge geometry was
+different: every positive had both the Red-Green and Blue-Yellow native
+8-pixel edge ratio at or below `1.05`. The threshold was selected against 600
+development codec crossings, then rejected all 582 crossings in the held-out
+Open Images cohort, all 601 in the feature-unseen frozen reserve, and all seven
+in the heterogeneous Spaces challenge. Separately, 350 controls passed through
+the same lossless 0.5x resize had no base carrier crossing. The runtime now
+accepts periods 7.9-12.0 and applies this pixel-only veto through period 8.1.
+The production API consequently detects 49/49 measured lossless views at scales
+0.5-0.75 from the seven official-positive parents.
+
+The same work did not recover lossy retranscodes. JPEG 4:2:0 creates a strong
+inverted period-8 template in positives and natural controls alike. A
+matched-transform 512-feature codec-residual discriminant accepted up to
+86/525 holdout controls and did not generalize across positive parents. Eighty
+block-DCT coherence variants also failed the precision-first gate: the best
+zero-holdout-error statistic retained only 2/21 JPEG views. These are rejected
+hypotheses, not hidden fallback branches.
+
+The wider provider challenge produced a useful but non-attributive observation:
+the unrestricted score accepted 95/7,254 OpenAI files with an explicit SynthID
+provenance assertion, all at periods 10.70-10.80. It also accepted 4/287 OpenAI
+lineage files without that assertion, including three in the production period
+band. Missing assertion is not an oracle-negative watermark label, and the
+feature does not identify a provider. All 95 decoded rasters were unique and
+none overlapped the 94 hits from the earlier period-8 OpenAI candidate, so the
+union covered 189/7,259 asserted files. The new component was nevertheless
+codec-fragile: JPEG-95 and WebP-95 each reduced 95/95 native hits to 0/95.
+It may be a complementary carrier epoch or an associated export/resampling
+lattice; current labels cannot distinguish those causes. The local detector
+therefore remains provider-neutral; OpenAI-wide coverage still comes from the
+official pixel verifier.
+
+Runtime routing now applies these policies directly and always abstains on
+fixed-only evidence. The research bank in
+`scripts/synthid_routed_expert_bank.py` still records fixed, registered, and
+large observations separately for audits. No route claims absence because
+encoder-version coverage remains incomplete.
 
 Running the unchanged registered components directly above the declared range
 did not justify a larger ceiling. From 10 to 18 megapixels it accepted 1 of 37
@@ -931,10 +1086,12 @@ the aligned result. Together with the failed third-carrier fit, this exhausts
 the current linear native 16-by-16 Google hypothesis as two jointly controllable
 states. It still does not replace a matching-provider oracle result.
 
-This mode is not the default because the bounded search is materially slower;
-scale registration is explicit and `identify` keeps the native detector. The
-registered mode does not reliably detect 0.5x carriers and does not make the
-detector universal across crop, codecs, carrier states, or providers.
+The bounded search is materially slower, but registered-v3 is now the default
+ordinary-size route because fixed-v2 failed its fresh-source precision gate.
+Registered-v3 itself does not reliably detect 0.5x carriers. The later bounded
+opponent fallback covers the measured lossless 0.5x case with its codec-grid
+veto, but does not make the detector universal across crop, codecs, carrier
+states, or providers.
 
 A crop-specific follow-up tested cube-root LMS and OKLab projections, all six
 DTCWT orientations, and explicit FFT phase-lock metrics after period-and-phase
@@ -1185,8 +1342,9 @@ runtime until a new future holdout confirms it.
 A positive result identifies the carrier but does not attribute a provider.
 Provider identity still comes from provenance.
 
-The command reports `not_detected` separately from `unsupported`. Both are
-inconclusive outside the measured carrier family and calibrated image-size range.
+The command reports `indeterminate` separately from `unsupported`. Both are
+inconclusive: the first means no qualified local expert crossed its threshold,
+and the second means no local expert covers the geometry.
 
 The same modulo-folding method has been tested separately on a large,
 temporally split OpenAI-labeled corpus. Its strongest native-size template was
@@ -1543,6 +1701,23 @@ the signed OpenAI period-8 candidate accepted 4 of 417 unique positives and 1
 of 454 general-control groups, the OpenAI full-image noiseprint accepted 68 of
 417, and the Google runtime detector accepted 153 of 211 unique positives.
 
+A previously prepared image-level patch-bag hypothesis was then executed. It
+aggregates the mean, standard deviation, minimum, and maximum of nine learned
+embeddings from normalized high-pass luminance and opponent-color patches. The
+validation-selected model accepted 285 of 1,280 OpenAI rows above every
+validation control. Without changing its weights or threshold, it accepted 324
+of 1,364 development-test rows, 3 of 1,373 difficult controls, and 1 of 2,000
+COCO controls. The later August temporal challenge accepted 120 of 486 OpenAI
+rows, 8 of 504 general controls, and 5 of 33 same-provider rows without a
+watermark assertion. Assertion enrichment over that same-provider stratum was
+not independently significant (`p = 0.151`, one-sided exact test). Content
+grouping reduced those counts to 107 of 417, 4 of 453, and 4 of 30,
+respectively. A JPEG-95 conjunction retained only 1 of 486
+OpenAI rows and no controls. Its native hits did not overlap the frozen signed
+period-8 temporal hits. The model therefore reveals a complementary OpenAI
+export noiseprint, not a codec-robust or watermark-specific decoder, and does
+not enter runtime.
+
 ### 3.3 Official OpenAI pixel verification
 
 `remove-ai-watermarks verify-openai-synthid image.png
@@ -1559,6 +1734,11 @@ entry. The source is not modified. Tests deliberately cover C2PA-only positive
 responses, pixel mutation, surviving metadata, malformed response shapes, and
 documented access and rate-limit failures.
 
+The default SDK client fixes a 120-second request timeout and disables
+automatic retries. One explicit acknowledgement therefore cannot silently
+transmit the sanitized raster more than once. Timeout and connection failures
+are errors, and request logs omit source paths and decoded-pixel hashes.
+
 A live 2026-08-14 web-verifier smoke used the same sanitization invariant. Two
 metadata-stripped, pixel-identical OpenAI images at 1536 by 1024 and 1024 by
 1536 both returned `SynthID detected` with `Content Credentials not detected`.