Input Catalog Reference¶

This page is the definitive reference for the three input files consumed by tract7dt:

Input source catalog (inputs.input_catalog) — the CSV table of sources to be fit.
Image list (inputs.image_list_file) — the list of FITS image paths.
GaiaXP synphot catalog (inputs.gaiaxp_synphot_csv) — optional Gaia DR3 synthetic photometry for ePSF star selection.

Understanding the format requirements, column semantics, and physical units of these files is essential before running the pipeline.

1. Input Source Catalog¶

Config key: inputs.input_catalog

Format: CSV (comma-separated values), readable by pandas.read_csv().

This is the primary science input: a table of astronomical sources to be modeled and fit by the Tractor. Each row represents one source. The pipeline projects source sky positions onto the image pixel grid, assigns each source to a spatial patch, initializes a Tractor source model (point source or galaxy), and fits that model simultaneously across all bands.

1.1 Required Columns¶

Column	Type	Unit / Convention	Description
`RA`	float	degrees (ICRS)	Right Ascension of the source.
`DEC`	float	degrees (ICRS)	Declination of the source.

Column name matching

Column lookup is case-insensitive (e.g. ra, Ra, RA are all accepted), but the pipeline emits a warning if the exact column names are not RA and DEC. For clarity and to avoid warnings, always use uppercase RA and DEC.

Coordinate system

RA/DEC are interpreted through the WCS of the input FITS images via astropy.wcs.WCS.all_world2pix(). The coordinate frame must match the WCS frame of your images — typically ICRS (J2000 FK5). If your catalog uses a different epoch or frame, you must preprocess it to match before running the pipeline.

Validation rules:

If RA or DEC columns cannot be found (case-insensitive), the pipeline raises ValueError.
Rows with NaN or non-finite RA/DEC are silently excluded from spatial assignment to patches, but are preserved in the final merged catalog (with empty fit columns).

1.2 Recommended Columns¶

These columns are not strictly required but significantly improve model initialization and diagnostic quality.

Column	Type	Unit / Convention	Description
`ID`	string	—	Unique source identifier. Strongly recommended. Used as the merge key when joining patch fit results back to the input catalog. If absent, the pipeline falls back to using `RA`+`DEC` as a composite merge key, which is fragile for sources at identical coordinates. IDs are treated as format-free strings — the pipeline reads and preserves them exactly as written in the CSV. Leading zeros, non-numeric characters, and arbitrary text are all preserved (e.g. `00019`, `1858283`, `my_star_alpha`). The ID is also used in cutout filenames (`src_{ID}.png`).
`TYPE`	string	One of: `STAR`, `GAL`, `EXP`, `DEV`, `SERSIC`	Morphological classification that determines the Tractor source model. See Source Model Mapping below.

1.3 Optional Columns¶

Column	Type	Unit / Convention	Description
`FLUX_{band}`	float	Scaled flux (in the reference zeropoint system set by `image_scaling.zp_ref`, default 25.0 mag)	Per-band initial flux estimate. `{band}` must exactly match the `FILTER` keyword in the corresponding FITS header (e.g. `FLUX_m400`, `FLUX_m625`). When provided, these values seed the Tractor optimizer. When absent, the pipeline performs SEP aperture photometry on the images to initialize fluxes.
`ELL`	float	Dimensionless, range [0, 1)	Ellipticity, defined as `1 - b/a` where `b/a` is the axis ratio. Used to initialize galaxy model shape. If absent, defaults to `0.2` (axis ratio `b/a = 0.8`). Only used for galaxy-type sources.
`THETA`	float	degrees	Position angle in SExtractor convention: measured counter-clockwise from the image x-axis. Internally converted to the Tractor's `EllipseESoft` parameterization. If absent, defaults to `0`. Only used for galaxy-type sources.
`Re`	float	pixels	Effective (half-light) radius. Used to initialize the galaxy model size parameter. If absent, falls back to `patch_run.re_fallback_pix` (default: 3.0 pixels). Values are clipped to [0.3, 100.0] pixels internally. Only used for galaxy-type sources.
`SERSIC_n`	float	Dimensionless	Per-source initial Sersic index. Only used for `SERSIC`-type sources. If absent or `NaN`, falls back to `patch_run.sersic_n_init` (default: 3.0).
`POS_ERR`	float	pixels	Positional uncertainty. When provided and finite, a Gaussian prior is applied to the fitted position using this value as sigma for both x and y. For newly injected Gaia sources, this is set automatically from `zp.gaia_pos_err_pix` (default: 0.1) during Gaia augmentation and written into the saved augmented catalog. Matched original catalog rows are not auto-filled.
`FLUX_{band}_ERR`	float	Scaled flux (same unit as `FLUX_{band}`)	Per-band flux uncertainty. When provided and the corresponding `FLUX_{band}` value was used (not fallback aperture photometry), a Gaussian prior is applied to that band's flux.
`Re_ERR`	float	pixels	Uncertainty in effective radius. When provided and `Re` was used from catalog (not fallback), a Gaussian prior is applied to `log(Re)` with `sigma_logre = Re_ERR / Re`. Only used for galaxy-type sources.
`ELL_ERR`	float	Dimensionless	Uncertainty in ellipticity. Must be provided together with `THETA_ERR` for ellipticity/PA priors to take effect. Only used for galaxy-type sources.
`THETA_ERR`	float	degrees	Uncertainty in position angle. Must be provided together with `ELL_ERR` for ellipticity/PA priors to take effect. Errors are propagated to the internal `(ee1, ee2)` parameterization via the analytical Jacobian. Only used for galaxy-type sources.

Gaussian priors from error columns

When error columns are present and the corresponding parameter value came from the catalog (not a fallback default), the pipeline applies a Gaussian prior: -0.5 * ((fitted - initial) / error)^2 is added to the log-probability. This softly pulls the optimizer toward the catalog value. Sources without error values (missing column or NaN) are fitted without any prior (current default behavior). Priors work per-source: some sources can have priors while others do not.

ELL_ERR and THETA_ERR must be provided together

Because the internal shape parameterization (ee1, ee2) depends on both ellipticity and position angle, providing only one of ELL_ERR or THETA_ERR results in infinite propagated error. Both must be present for ellipticity/PA priors to take effect. Re_ERR is independent and works on its own.

1.4 Source Model Mapping¶

The TYPE column determines which Tractor source model is used for each source. Matching is case-insensitive (the value is uppercased internally).

`TYPE` value	Tractor model	Description
`STAR`	`PointSource`	Unresolved point source. Only position and per-band fluxes are fit.
`EXP`	`ExpGalaxy`	Exponential-profile galaxy (Sersic n=1). Fits position, fluxes, effective radius, ellipticity, and position angle.
`DEV`	`DevGalaxy`	de Vaucouleurs-profile galaxy (Sersic n=4). Same free parameters as `EXP`.
`SERSIC`	`SersicGalaxy`	General Sersic profile. Fits all `EXP`/`DEV` parameters plus the Sersic index `n`. Initialized from `SERSIC_n` column if present and finite; otherwise from `patch_run.sersic_n_init` (default 3.0).
`GAL`	(same as fallback)	Alias. Treated as unknown type; the fallback model is used.
(missing, empty, or any other value)	Fallback model	Determined by `patch_run.gal_model` config key (default: `exp`). Allowed values: `exp`, `dev`, `sersic`, `star`.

TYPE column entirely absent

If the TYPE column does not exist at all, every source is assigned the fallback model set by patch_run.gal_model. In overlay diagnostic plots, all such sources appear as yellow squares labeled UNKNOWN (fallback=<model>).

Choosing a fallback model

For fields dominated by faint, unresolved sources, consider patch_run.gal_model: "star". For deep extragalactic fields with mostly galaxies, "exp" is a common choice.

1.5 Band Consistency Check¶

If the input catalog contains any column whose name starts with FLUX_ (e.g. FLUX_m400, FLUX_m625), the pipeline performs a band consistency check:

It extracts band names from catalog column names: {FLUX_m400 → m400, FLUX_m625 → m625, ...}.
It extracts band names from image FILTER headers: {m400, m625, ...}.
If these two sets do not match exactly, the pipeline raises RuntimeError with a message listing which bands are catalog-only and which are image-only.

This prevents silent mismatches between your catalog flux columns and your actual images.

No FLUX columns? No check.

If your catalog has no FLUX_* columns at all, this check is skipped and fluxes are initialized purely from aperture photometry on the images.

1.6 Flux Initialization Logic¶

For each source and each band, the Tractor optimizer is seeded with an initial flux value. The selection logic is:

If FLUX_{band} exists in the catalog and the value is finite and positive → use it.
Otherwise → perform circular aperture photometry on the scaled image using sep.sum_circle() with radius patch_run.r_ap (default: 5.0 pixels).
The result is clamped to a minimum floor of patch_run.eps_flux (default: 1e-4) to avoid zero-flux initialization failures.

If FLUX_{band}_ERR exists and the flux value was taken from the catalog (step 1), a Gaussian prior is applied to that band's flux using the catalog value as the mean and the error as the sigma. If the flux was from aperture photometry fallback (step 2), the error column is ignored for that source and band.

1.7 Coordinate Projection¶

Source RA/DEC are projected to pixel coordinates using the WCS of the first loaded FITS image (astropy.wcs.WCS.all_world2pix()). The projection is done once during load_inputs() and reused for:

Crop filtering (is the source inside the crop box?)
Saturation filtering (is the source near saturated pixels?)
Patch assignment (which spatial patch does the source belong to?)
Overlay diagnostic plots

Sources whose RA/DEC are NaN, non-finite, or project outside the image bounds are tracked but do not participate in fitting. They still appear in the final merged catalog.

1.8 Minimal Example¶

A minimal input catalog CSV:

ID,RA,DEC,TYPE
1,34.40625,-5.22306,STAR
2,34.40750,-5.22100,EXP
3,34.40500,-5.22500,DEV

A more complete example with flux priors and shape parameters:

ID,RA,DEC,TYPE,FLUX_m400,FLUX_m475,FLUX_m550,FLUX_m625,ELL,THETA,Re,SERSIC_n
1,34.40625,-5.22306,STAR,150.3,200.1,250.5,180.2,,,,
2,34.40750,-5.22100,EXP,50.0,60.0,70.0,55.0,0.3,45.0,5.2,
3,34.40500,-5.22500,SERSIC,80.0,90.0,100.0,85.0,0.15,120.0,8.0,2.5

An example with error columns for Gaussian priors:

ID,RA,DEC,TYPE,POS_ERR,FLUX_m400,FLUX_m400_ERR,ELL,ELL_ERR,THETA,THETA_ERR,Re,Re_ERR
1,34.40625,-5.22306,STAR,0.5,150.3,5.0,,,,,,
2,34.40750,-5.22100,EXP,1.0,50.0,3.0,0.3,0.05,45.0,5.0,5.2,0.5

Stars and shape columns

For STAR-type sources, ELL, THETA, Re, and SERSIC_n are ignored (point sources have no shape parameters). It is safe to leave them blank or NaN for stars.

1.9 Common Mistakes and Validation Errors¶

Symptom	Cause	Fix
`ValueError: input_catalog must have RA/DEC columns`	No column named `RA` or `DEC` (case-insensitive).	Rename your coordinate columns to `RA` and `DEC`.
`RuntimeError: Band mismatch between input catalog and image list`	`FLUX_*` column band names don't match image `FILTER` headers.	Ensure `FLUX_` suffixes exactly match FITS `FILTER` values, or remove all `FLUX_*` columns to use aperture-photometry initialization.
Warning: `input_catalog uses ra/dec columns; interpreting as Right Ascension / Declination`	Columns are lowercase `ra`/`dec` instead of `RA`/`DEC`.	Rename to uppercase. The pipeline works but warns.
Many sources with empty fit columns in output	Sources have `NaN` coordinates, project outside image bounds, or were flagged by crop/saturation filters.	Check `excluded_*` columns in the output. Verify coordinate frame matches WCS.
All sources labeled `UNKNOWN` in overlay plots	`TYPE` column is missing or contains unrecognized values.	Add a `TYPE` column, or accept the fallback model set by `patch_run.gal_model`.
`ValueError: Duplicate keys in patch results`	Multiple sources share the same `ID` (or same `RA`+`DEC` when `ID` is absent).	Ensure `ID` values are unique, or ensure no two sources share identical coordinates.

2. Image List File¶

Config key: inputs.image_list_file

Format: Plain text file. One FITS image path per line.

2.1 Syntax¶

One path per line. Each non-empty, non-comment line is interpreted as a FITS file path.
Comments: Lines starting with # are ignored.
Blank lines are ignored.
Relative paths are resolved against the directory containing the YAML config file.
Absolute paths are used as-is.

2.2 Example¶

# 7DT UDS Field - All Bands
/data/7dt/7DT_UDS_m400.fits
/data/7dt/7DT_UDS_m475.fits
/data/7dt/7DT_UDS_m550.fits
/data/7dt/7DT_UDS_m625.fits

2.3 FITS Image Requirements¶

Each FITS file listed must satisfy specific header requirements. The pipeline reads the primary HDU (hdul[0]) and checks for the following:

Required FITS Header Keywords¶

Keyword	Type	Unit / Convention	Description
`FILTER`	string	—	Band/filter name (e.g. `m400`, `m625`). Must not be empty. This value is used as the band identifier throughout the pipeline and must match `FLUX_{band}` column suffixes if present in the input catalog.
`ZP_AUTO`	float	mag (AB)	Photometric zero-point. Used to scale all images to a common reference zero-point (`image_scaling.zp_ref`, default 25.0). The scaling factor applied is `10^(-0.4 * (ZP_AUTO - zp_ref))`.
`SKYSIG`	float	ADU	Sky background noise (1-sigma) per pixel, in ADU. Must be positive and finite. Used to build the sky noise map and inverse-variance weights.
`EGAIN`	float	e-/ADU	Effective gain. Must be positive and finite. Used to compute source Poisson noise: `var_source = max(image, 0) / EGAIN`.

Optional FITS Header Keywords¶

Keyword	Type	Unit / Convention	Description
`SATURATE`	float	ADU	Saturation level of the detector. If present and finite, pixels at or above `SATURATE / saturation_divisor` are flagged as saturated in the saturation mask (see `source_saturation_cut.saturation_divisor`; default `1.3`). If absent, no saturation mask is built (pipeline warns and uses `inf`).
`PEEING`	float	pixels	PSF FWHM in pixels. Used as fallback when the ePSF is not available for a given band/cell. Takes priority over `SEEING`.
`SEEING`	float	arcseconds	Atmospheric seeing FWHM. Used as fallback PSF when `PEEING` is absent. Converted to pixels using the WCS pixel scale.

WCS Requirements¶

Each image must contain a valid WCS (World Coordinate System) in its primary header, parsable by astropy.wcs.WCS(). All images must share a consistent, aligned WCS:

CRVAL (reference position) must agree within checks.wcs_tolerance.crval (default: 1e-6 degrees).
CRPIX (reference pixel) must agree within checks.wcs_tolerance.crpix (default: 1e-6 pixels).
CD/PC matrix must agree within checks.wcs_tolerance.cd (default: 1e-9).
CDELT must agree within checks.wcs_tolerance.cdelt (default: 1e-9 deg/pix).
CTYPE must be identical across all images.

If checks.require_wcs_alignment: true (default) and any image fails this check, the pipeline raises RuntimeError.

Shape Requirement¶

If checks.require_same_shape: true (default), all images must have identical (NAXIS2, NAXIS1) dimensions. This is expected because 7DT images for a given tile are pre-registered and stacked to a common grid.

2.4 Image Scaling¶

All images are scaled to a common photometric reference system before any analysis. For each image:

scale_factor = 10^(-0.4 * (ZP_AUTO - zp_ref))
scaled_image = raw_image * scale_factor

Where zp_ref is set by image_scaling.zp_ref (default: 25.0). This means that in the scaled system:

A source with flux F (in scaled counts) has approximate magnitude: mag ≈ -2.5 * log10(F) + zp_ref. This is exact only if ZP_AUTO in the FITS header perfectly represents the true zero-point. In practice, ZP_AUTO has errors, so the pipeline's ZP calibration stage (when zp.enabled) derives the actual ZP from Gaia-matched stars and applies calibrated MAG_{band}_fit columns.
All FLUX_* values in the input catalog (if provided) should be in this same scaled system.
All fitted flux values in the output catalog are in this system.

2.5 Noise Model¶

For each pixel in each band, the total noise variance is:

var_total = SKYSIG^2 + max(raw_image, 0) / EGAIN
sigma_total = sqrt(var_total) * scale_factor

The pipeline uses the sky-only sigma (not total sigma) as the default noise estimate for Tractor fitting (sigma_sky_scaled), which means:

sigma_sky_scaled = SKYSIG * scale_factor
invvar = 1 / sigma_sky_scaled^2   (for good pixels)
invvar = 0                         (for bad/saturated pixels)

2.6 Common FITS Errors¶

Error	Cause	Fix
`RuntimeError: Missing FILTER in <file>`	The FITS header has no `FILTER` keyword.	Add `FILTER` keyword to the header.
`RuntimeError: Empty FILTER in <file>`	`FILTER` header value is blank or whitespace.	Set `FILTER` to the band name string.
`RuntimeError: Missing ZP_AUTO in <file>`	No `ZP_AUTO` in header.	Run photometric calibration and write `ZP_AUTO`.
`RuntimeError: Bad/Missing SKYSIG in <file>`	`SKYSIG` is absent, non-finite, or ≤ 0.	Compute and write a valid `SKYSIG` value.
`RuntimeError: Bad/Missing EGAIN in <file>`	`EGAIN` is absent, non-finite, or ≤ 0.	Write a valid `EGAIN` value (typically from detector specs).
`RuntimeError: WCS mismatch with reference: <file>`	WCS does not agree with the first image within tolerances.	Re-register images to a common WCS, or relax `checks.wcs_tolerance`.
`RuntimeError: Image shape mismatch: <file>`	Array dimensions differ from the first image.	Ensure all images are resampled to the same pixel grid.
Warning: `SATURATE missing in <file>; using inf.`	No `SATURATE` keyword.	Add `SATURATE` if you want saturation masking; otherwise this warning is harmless.

3. GaiaXP Synphot Catalog¶

Config key: inputs.gaiaxp_synphot_csv

Format: CSV, readable by pandas.read_csv().

This catalog provides Gaia DR3 XP synthetic photometry for stars in the field. It is used during ePSF (effective Point Spread Function) construction to identify high-quality PSF star candidates. The pipeline cross-matches these positions with SEP source detections in the image to select clean, unsaturated, isolated stars for ePSF building.

3.1 Required Columns¶

Column	Type	Unit / Convention	Description
`ra`	float	degrees (ICRS)	Gaia DR3 right ascension. Note: lowercase.
`dec`	float	degrees (ICRS)	Gaia DR3 declination. Note: lowercase.
`mag_{band}`	float	mag (AB, synthetic)	Synthetic magnitude in the band matching the FITS `FILTER` name. One column per band. For example, if your images have `FILTER=m400` and `FILTER=m625`, this catalog needs columns `mag_m400` and `mag_m625`.

Column names are lowercase

Unlike the input source catalog (which uses uppercase RA/DEC), the GaiaXP catalog uses lowercase ra and dec. This reflects Gaia catalog conventions.

3.2 Example¶

ra,dec,mag_m400,mag_m475,mag_m550,mag_m625
34.405,−5.223,18.2,17.8,17.5,17.3
34.410,−5.225,19.1,18.7,18.4,18.2

3.3 Magnitude Filtering¶

Only GaiaXP sources within the configured magnitude range are used as PSF star seeds:

epsf.gaia_mag_min (default: 10.0 mag) — reject sources brighter than this (likely saturated).
epsf.gaia_mag_max (default: 30.0 mag) — reject sources fainter than this (too faint for PSF).

3.4 How It Is Used¶

The GaiaXP catalog is used only during the ePSF building stage (run-epsf or the ePSF step within run). The flow is:

For each band, look for column mag_{band} in the GaiaXP CSV.
Filter to sources with magnitudes within [gaia_mag_min, gaia_mag_max].
Project GaiaXP RA/DEC to pixel coordinates using image WCS.
For each ePSF grid cell, gather GaiaXP seeds that fall within the cell.
Depending on epsf.psfstar_mode:
"gaia" — use only GaiaXP positions.
"sep+gaia" (default) — use GaiaXP positions as priority seeds, supplemented by SEP-detected sources.
"sep" — ignore GaiaXP entirely, use only SEP detections.
If epsf.gaia_snap_to_sep: true (default), GaiaXP seed positions are refined by snapping to the nearest SEP detection within epsf.gaia_match_r_pix pixels.
Each candidate undergoes SNR and saturation checks before being accepted.

3.5 When It Can Be Omitted¶

You can omit the GaiaXP catalog by either:

Setting epsf.use_gaiaxp: false in the config.
Setting epsf.psfstar_mode: "sep" (SEP-only star selection).
Pointing the path to a non-existent file (the pipeline warns but does not crash).

In these cases, PSF star selection relies entirely on automated SEP source detection and selection (roundness filter, brightness ranking, saturation avoidance).

3.6 Generating a GaiaXP Synphot Catalog¶

The GaiaXP synphot catalog is typically generated by:

Querying Gaia DR3 sources within the tile footprint (e.g. via TAP, astroquery, or a local copy).
Selecting sources with XP spectra available.
Computing synthetic photometry through the 7DT filter response curves.
Writing the result as a CSV with the required columns.

This catalog is field-specific and must be regenerated for each new tile/pointing.

4. Summary: Input File Checklist¶

Before running the pipeline, verify:

[ ] Input catalog CSV exists at the path specified in inputs.input_catalog.
[ ] Catalog has RA and DEC columns (uppercase recommended).
[ ] Catalog has a unique ID column (strongly recommended).
[ ] If FLUX_* columns exist, their band suffixes exactly match FITS FILTER values.
[ ] If FLUX_{band}_ERR columns exist for Gaussian priors, their band suffixes also match.
[ ] If using ELL_ERR/THETA_ERR for shape priors, both columns are present (one without the other has no effect).
[ ] If using POS_ERR, values are in pixels (not degrees).
[ ] RA/DEC coordinates are in the same frame as the image WCS (typically ICRS).
[ ] Image list file exists at inputs.image_list_file.
[ ] All listed FITS files exist and are accessible.
[ ] All FITS files have FILTER, ZP_AUTO, SKYSIG, EGAIN headers.
[ ] All FITS files have valid, consistent WCS.
[ ] All FITS files have the same pixel dimensions.
[ ] GaiaXP CSV exists at inputs.gaiaxp_synphot_csv (if using GaiaXP-based ePSF).
[ ] GaiaXP CSV has lowercase ra, dec, and mag_{band} columns for each band.

5. Physical Units Summary¶

This table collects all physical units in one place for quick reference.

Quantity	Unit	Where used
RA / DEC (source catalog)	degrees (ICRS)	`input_catalog` columns `RA`, `DEC`
RA / DEC (GaiaXP catalog)	degrees (ICRS)	`gaiaxp_synphot_csv` columns `ra`, `dec`
Flux (`FLUX_{band}`)	scaled counts (ZP = `zp_ref`)	`input_catalog` optional columns
Ellipticity (`ELL`)	dimensionless, 1 − b/a	`input_catalog` optional column
Position angle (`THETA`)	degrees (SExtractor convention: CCW from x-axis)	`input_catalog` optional column
Effective radius (`Re`)	pixels	`input_catalog` optional column
Sersic index (`SERSIC_n`)	dimensionless	`input_catalog` optional column
Position uncertainty (`POS_ERR`)	pixels	`input_catalog` optional column
Flux uncertainty (`FLUX_{band}_ERR`)	scaled counts (ZP = `zp_ref`)	`input_catalog` optional column
Re uncertainty (`Re_ERR`)	pixels	`input_catalog` optional column
Ellipticity uncertainty (`ELL_ERR`)	dimensionless	`input_catalog` optional column
PA uncertainty (`THETA_ERR`)	degrees (SExtractor convention)	`input_catalog` optional column
`ZP_AUTO`	magnitudes (AB)	FITS header
`SKYSIG`	ADU (per pixel)	FITS header
`EGAIN`	electrons per ADU (e⁻/ADU)	FITS header
`SATURATE`	ADU	FITS header
`PEEING`	pixels (FWHM)	FITS header
`SEEING`	arcseconds (FWHM)	FITS header
GaiaXP magnitudes (`mag_{band}`)	magnitudes (AB, synthetic)	`gaiaxp_synphot_csv`
WCS tolerance: `crval`	degrees	config `checks.wcs_tolerance.crval`
WCS tolerance: `crpix`	pixels	config `checks.wcs_tolerance.crpix`
WCS tolerance: `cd`, `cdelt`	degrees/pixel	config `checks.wcs_tolerance.cd`, `.cdelt`
Crop margin	pixels	config `crop.margin`
Saturation search radius	pixels	config `source_saturation_cut.radius_pix`
Aperture radius (flux init)	pixels	config `patch_run.r_ap`
Effective radius fallback	pixels	config `patch_run.re_fallback_pix`
Fallback PSF FWHM	pixels	config `patch_run.fallback_default_fwhm_pix`