This paper presents a large, public catalog of 2,452 star‑forming galaxy candidates that the authors identify at photometric redshifts between 6.5 and 16.5. The sample was found across about 650–740 square arcminutes of deep imaging from six widely used blank‑field JWST Near‑Infrared Camera (NIRCam) surveys. The catalog and the software used to make it are released together so other teams can use the same, reproducible measurements.

The team built the sample with a modular Python pipeline called galfind. They reprocessed the NIRCam images with a modified JWST reduction, applied custom corrections for known imaging artefacts, aligned all images to the Gaia reference frame, and measured object colors across many filters. The release includes optical and near‑infrared catalogs, empirical point‑spread functions (PSFs, which describe how a point source appears in the images), and spectral energy distribution (SED) fits using several public tools (eazy‑py, bagpipes, and bdfinder). SED fitting is the method of comparing measured colors to models to estimate properties such as redshift and dust.

Where possible the authors checked photometric redshifts against spectra. The sample contains 293 objects with spectroscopic confirmation. A comparison between photometric and spectroscopic redshifts gives a normalized median absolute deviation of 0.017, an outlier rate of 4.3%, and a small mean offset of −0.022. The team also searched for and removed likely contaminants: 34 “little red dots” with unusual “V‑shaped” color curves and a set of brown dwarf candidates (69 L‑type, 36 T‑type and 2 compact Y‑type). From the cleaned sample they report a flattening in the relation between UV slope (βUV, a measure of how blue or red the ultraviolet light is) and UV luminosity. At z ≈ 12.5 the median UV slope at MUV = −19 is βUV = −2.32 (+0.08/−0.07), which the authors say implies the presence of some dust in these early galaxies.