Color Calibration

Derive reproducible channel scaling from catalog spectra, the imaging system response, and an explicit white reference instead of balancing color by eye.

schema 1 · revision 2

Recognize it

The linear image retains the camera/filter response rather than an intentional color reference.
Star colors or background channel ratios show a broad systematic cast after gradient removal.

Intended output

A linear, astrometrically valid image with recorded calibration inputs, catalog matches, fit evidence, and channel coefficients that downstream linked operations preserve.

Required input state

Linear, unstretched color data after accepted background extraction.
A correct plate solution whose field center, scale, and orientation match the image.
Known OSC/mono state, sensor response, filter or passband information, and chosen white reference.

What NOVA does

As of a 2026-09-02 policy change, NOVA attempts SSSC first for both broadband and LP/dual-band data, not PixInsight SPCC. A 17-target same-methodology comparison found SSSC tied PixInsight in every case where SSSC reached at least Stage 1 calibration (13 of 17 targets, zero exceptions); every real divergence happened at the weaker Stage 2, and even there SSSC's worst case was about 5% behind while its best cases were 3-8x better. PixInsight is therefore consulted as a second opinion only when SSSC's own solve lands at Stage 2 -- SPCC for broadband, PixInsight ColorCalibration for LP/dual-band (SPCC's broadband white reference is deliberately never applied there). The trigger is SSSC's solve stage, not a raw fit-residual threshold, because stage reflects how many usable calibration stars were available, which is what the evidence showed actually predicts reliability -- a legitimately-agreeing target can carry a higher residual than some Stage-2 disagreements. A sanity gate then checks the signal region's G/R ratio before and after calibration and discards a solve that moved sharply away from neutral, regardless of filter type; that check exists because a plausible-looking fit can still pair detected stars with the wrong catalog entries in dense fields.

Use when

The linear image has trustworthy astrometry and documented sensor/filter response.
The chosen calibration mode matches broadband or narrowband acquisition.

Skip when

The image is stretched, monochrome without an intended color composition, or lacks valid astrometry.
Required sensor/filter/passband information is unknown.
An aesthetic false-color palette is the goal and physical line intensity is not the intended balance.

Limits

A process-complete message does not prove that the WCS, sensor, filter, white reference, or fit was valid.
Broadband SPCC and narrowband/dual-band calibration are not interchangeable prescriptions.
Physically calibrated narrowband intensities do not automatically produce an aesthetic SHO/Hubble palette.
Background neutralization is not a substitute for a valid spectrophotometric solve.
NOVA's own choice not to apply broadband SPCC to LP/dual-band data is a project engineering decision about its white reference, not a claim that filter-aware SPCC implementations in general cannot support narrowband or dual-narrowband response models -- modern SPCC versions can.
That the sanity gate did not reject a solve is evidence the result was not caught by that specific check, not evidence the calibration is physically correct; not rejected is not validated.
The Experiment Mode `cc_sssc` candidate calls SSSC with the LP flag off and without the native LP curve, so it does not exercise the LP/dual-band path the production policy actually uses for LP targets; a recorded `cc_sssc` result is evidence about the broadband-only path only.
SSSC's LP throughput curve is a real, hash-verified SeeStar S50 transmission curve, not a commercial-filter proxy -- but it remains a model of the filter's response, and any measured system-response solution built on it inherits that model's uncertainty.

Scientific and aesthetic notes

Siril SPCC compares measured image flux ratios with Gaia DR3 spectra folded through the selected sensor and filters.
White reference is an explicit rendering choice inside a physically grounded calibration; Average Spiral Galaxy is Siril's broad default, not a universal truth.
A linked downstream stretch preserves established channel relationships; unlinked stretching can overwrite them.

Evidence and measurements

after color calibration before and after

beforeafter

color calibration before and after

Cross-tool matrix

Tool	Equivalence	Expected result	Evidence
nova	Exact replay	The selected route completes with valid fit evidence and produces channel balance appropriate to the acquisition mode.	nova-color-source, nova-m66-run-1.24.7
pixinsight	Same engine, adapted host	The fit is defensible and star colors show a plausible range without a global cast.	m66-manual-verification
siril	Algorithmically equivalent	Robust catalog-to-image fits support the applied coefficients and downstream linked operations preserve them.	siril-spcc-1.4.4, m66-manual-verification
saspro	Functional alternative	A documented system-response solution calibrates the linear image without being mislabeled as SPCC.	saspro-source-1.18.0, nova-color-source

NOVA Python pipeline 1.24.7

Verify or repair WCS and record acquisition/filter state before calibration runs.
Let SSSC attempt first for both broadband and LP/dual-band data; do not treat PixInsight SPCC as the default route.
Record SSSC's solve stage, whether a PixInsight second opinion was consulted, the sanity-gate's before/after G/R values, and any fallback.

Controls and starting ranges

default route: SSSC first, both broadband and LP/dual-band
PixInsight second opinion: consulted only when SSSC's solve lands at Stage 2 -- SPCC for broadband, PixInsight ColorCalibration for LP/dual-band
sanity gate: reject when the signal-region G/R ratio moves sharply away from neutral after calibration, applied regardless of filter type

Mask behavior: Calibration is global; background/white reference selection and catalog-star rejection are engine-specific safeguards.

Failure modes

Stale or malformed WCS allows a misleading run or forces fallback.
Broadband SPCC's white reference is applied to LP/dual-band data, which the current policy deliberately avoids.
A fallback or second-opinion result is described as if SSSC's own solve had succeeded outright.

Recovery

Restore the linear input, repair astrometry/configuration, and rerun while preserving the failed-route record.

PixInsight SPCC M66 validation environment; exact version not recorded

Confirm the image is linear and its astrometric solution is correct.
Select the SeeStar S50 sensor, actual filter/LP mode, and intended white reference.
Apply SPCC, then record catalog, matched-star/fit evidence, coefficients, and whether any fallback occurred.

Controls and starting ranges

white reference: Average Spiral Galaxy for the verified M66 broadband run
filter mode: match acquisition; do not infer broadband from image appearance
M66 LP mode: disabled

Mask behavior: Global calibration; use the process's source-detection and fit-rejection controls rather than a decorative image mask.

Failure modes

Wrong WCS or response profile produces invalid matching or misleading balance.
Fallback is not disclosed.

Recovery

Undo, correct WCS/sensor/filter/reference inputs, and repeat from the linear post-background image.

Siril SPCC 1.4.4

Run only on linear data and plate-solve with correct center, scale, and orientation first.
Select OSC mode, the closest valid SeeStar sensor/filter response, broadband or narrowband mode, and an explicit white reference.
Inspect and record the R/G and B/G fit plots, matched-star count, and resulting coefficients.

Controls and starting ranges

catalog: Gaia DR3; local catalog supported in Siril 1.4+
M66 white reference: Average Spiral Galaxy
M66 result: 33 stars; R 0.837, G 0.759, B 1.000

Mask behavior: SPCC uses detected/catalog-matched stars and robust fitting; it is not ordinarily applied through an image mask.

Failure modes

The remote Gaia service is unavailable and no local catalog is configured.
Existing but inaccurate WCS is trusted without checking image scale and field.
SHO data are expected to become an aesthetic Hubble palette through physical SPCC alone.

Recovery

Use the local Gaia catalog or repair astrometry/configuration; for aesthetic SHO balance, use a documented manual palette workflow.

Seti Astro Suite Pro SSSC 1.18.0 source-inspected

Use SSSC on a linear, plate-solved color image when the field supplies enough spectrum-bearing calibration stars.
Provide valid astrometry and record Gaia-XP matches, solved response, coefficients, and fallback behavior.
Label SSSC as a related physical calibration approach, not the identically implemented SPCC process -- even though it is now NOVA's first-attempted route for both broadband and LP/dual-band data.

Controls and starting ranges

catalog: Gaia-XP spectra
NOVA integration minimum: 20 enriched spectrum-bearing stars; otherwise explicit fallback
SASpro solution stages: scalar below 50 stars; color model at 50+; full response at 200+ with adequate color span

Mask behavior: Global calibration based on matched stellar spectra; no ordinary image-mask claim is established.

Failure modes

Too few matched spectra produce fallback or an unstable solution.
SSSC and SPCC are described as exact equivalents.

Recovery

Repair astrometry, use a star-richer field where appropriate, or fall back explicitly to a validated SPCC route.

Validation status

Not yet Jeff-validated. Use this as sourced guidance, not a certification.

Sources

M66 manual recipe verification feedback M66_Manual_Recipe_Verification_Feedback.md — Jeff validated
NOVA SPCC/SSSC selection, fallback logic, and ontology nas_server/seti_astro.py:4150; nas_server/processing_ontology.json:492 — NOVA source-confirmed
M66 workflow 1.24.7 execution record critiques/20260802_045251_seestar_galaxy.json — NOVA execution record
Siril 1.4.4 spectrophotometric color calibration documentation — Vendor-documented
Installed Seti Astro Suite Pro 1.18.0 processing source saspro/pedestal.py; saspro/numba_utils.py; saspro/stacking_suite.py; saspro/abe.py; saspro/sssc.py — Artifact-confirmed

Related evidence

The M66 recipe and Jeff-validation ledger will link here when their static exports are available. Until then, this article remains explicitly unvalidated.