PHOTcube: IFU Spectral Extraction Module

• PHOTcube is a module that extends PHOTfun to perform crowded-field spectral extraction from IFU datacubes using slice-by-slice PSF photometry.
• It processes datacubes by slicing them into monochromatic images, applying established DAOPHOT routines, and reconstructing per-source stellar spectra.
• PHOTcube enhances data quality in dense fields by enabling local sky subtraction and interactive PSF-star selection, outperforming methods like PampelMuse.

Searching arXiv for PHOTcube and related sources. PHOTcube is an extension integrated into the PHOTfun software suite for crowded-field spectral extraction from integral-field spectroscopic datacubes. In the formulation used in the Milky Way bulge kinematics study, PHOTfun is a custom-developed Python-based tool for PSF photometry, and PHOTcube is its dedicated extension optimized for IFU datacubes. Its operational role is to extract stellar spectra in dense stellar fields by slicing a datacube into monochromatic images, performing PSF-based photometry on each slice, and reconstructing per-source spectra from the sequence of measured fluxes. In the paper that introduces it, PHOTcube is applied to MUSE datacubes of highly extincted and crowded bulge fields near the Galactic plane, where blending and sky-subtraction residuals complicate standard extraction workflows (Quezada et al., 8 Sep 2025).

1. Definition, scope, and problem setting

PHOTcube is described explicitly as a module rather than as a standalone package. The source paper states that PHOTfun “included a dedicated extension, PHOTcube, optimized for IFU datacubes,” and further specifies that PHOTcube is “an extension integrated into the PHOTfun GUI” (Quezada et al., 8 Sep 2025). Its function is to extend PHOTfun’s DAOPHOT-based PSF-photometry workflow from ordinary image sets to IFU datacubes.

The immediate motivation is crowded-field spectroscopy in the inner Milky Way bulge. The relevant MUSE fields lie close to the Galactic plane, where the study emphasizes severe extinction, high extinction and stellar crowding, and source blending. In that regime, simple aperture extraction is inadequate. The paper states the core requirement directly: “In crowded stellar fields, spectral extraction requires PSF-based photometry to accurately deblend sources.” PHOTcube is therefore a practical extraction layer for IFU data, intended for dense stellar environments rather than a general-purpose 3D spectral-spatial forward-modeling framework (Quezada et al., 8 Sep 2025).

The instrumental context in the paper is specific. PHOTcube is used on VLT/MUSE observations, in wide-field adaptive-optics mode, and the extracted spectra are used primarily in the Ca II triplet region around $8500$–$8700\,\AA$, from which radial velocities are later measured. In the paper’s own division of labor, PHOTcube is directly relevant to the new MUSE data reduction and spectrum extraction; subsequent radial-velocity measurement, foreground-contamination modeling, and bulge kinematic fitting occur downstream.

2. Software architecture and relation to PHOTfun and DAOPHOT-II

PHOTcube inherits its photometric core from PHOTfun. PHOTfun is presented as a GUI wrapper around established DAOPHOT-II and ALLSTAR routines rather than as a replacement for them. The paper is explicit that PHOTfun “does not add new capabilities to DAOPHOT-II”; its contribution is a user-friendly interface that facilitates source detection, PSF modeling, photometry, and interactive inspection (Quezada et al., 8 Sep 2025).

Within that architecture, PHOTfun includes the DAOPHOT-II subroutines FIND, PICK, PHOT, PSF, SUBTRACT, and DAOMATCH, and it uses ALLSTAR for PSF-fitting photometry. PHOTcube extends this environment to datacubes. The extension “enables the extraction of stellar spectra by concatenating the flux measurements … of each detected source across the datacube,” which defines its fundamental computational model: repeated 2D crowded-field photometry across wavelength slices, followed by spectral assembly.

The implementation details given in the paper are limited but concrete. PHOTfun is a Python package, its GUI is based on Shiny, the software is publicly available via pip, source code is on GitHub, and DAOPHOT compatibility is provided through a Docker container. The paper also highlights an integrated GUI, interactive PSF-star rejection, real-time display of intermediate products, and SAMP interoperability with TOPCAT and DS9. This suggests that PHOTcube’s distinctive contribution is not a new PSF formalism, but integration of established crowded-field photometry with an IFU-specific workflow and interactive control surfaces (Quezada et al., 8 Sep 2025).

3. Extraction workflow for IFU datacubes

The PHOTcube workflow is described as a slice-by-slice extraction sequence. It begins with master source detection on the white-light FoV image, which integrates total flux over wavelength. Source detection is therefore performed once on a high-S/N collapsed image, rather than independently in each wavelength plane (Quezada et al., 8 Sep 2025).

The next stage is PSF-star selection. Candidate PSF stars are pre-selected with PICK, after which the user is encouraged to refine them manually through the GUI by inspecting light profiles. The paper emphasizes this interactive rejection as a practical advantage in crowded fields.

PHOTcube then loads the datacube and slices it along the wavelength direction in a set of sequential monochromatic images. Using the fixed master and PSF target lists, each monochromatic slice is processed with DAOPHOT/ALLSTAR. The output of this stage is a monochromatic magnitude or flux measurement for each detected source at each wavelength slice.

Finally, PHOTcube performs spectrum reconstruction. For each source, it compiles the monochromatic magnitudes, converts them into fluxes, and generates the corresponding spectrum. The paper also summarizes this stage as reconstruction by “concatenating the measured fluxes across the full wavelength range” (Quezada et al., 8 Sep 2025).

This workflow implies a specific methodological choice. PHOTcube does not fit a full 3D model to the datacube. Instead, it applies 2D PSF photometry repeatedly to monochromatic images and assembles spectra afterward. A plausible implication is that the approach prioritizes compatibility with robust crowded-field tools already established in stellar photometry, while avoiding a more elaborate IFU-specific forward model.

4. Functional boundaries, assumptions, and quality control

The paper delineates PHOTcube’s capabilities with unusual clarity. PHOTcube does process IFU datacubes, slice them by wavelength into monochromatic images, use PSF photometry on each slice, extract deblended source fluxes, convert monochromatic magnitudes to fluxes, reconstruct stellar spectra, and operate within the PHOTfun GUI (Quezada et al., 8 Sep 2025).

It is equally clear about what PHOTcube is not. Detection is described as occurring on the white-light FoV image prior to slicing, so independent per-slice source detection is not part of the stated method. Astrometric alignment is not described as a PHOTcube function. Cube reconstruction is not performed, because the datacube is an input product from the MUSE pipeline. Radial-velocity determination is downstream from PHOTcube, as is foreground-contamination modeling. The safest interpretation is therefore that PHOTcube is an IFU spectral-extraction extension based on slice-wise PSF photometry, not a full end-to-end IFU reduction package.

Several assumptions are explicit. The white-light image is assumed to define the source list adequately across the wavelength range. The same master and PSF target lists are reused for all slices. Spectra are reconstructed by concatenating monochromatic fluxes, and no more sophisticated forward modeling of wavelength dependence is described. The paper also states, in the later radial-velocity stage, that template interpolation is not performed because “for RV measurements a perfect match between the observed and template spectra was not necessary.” That decision belongs downstream of PHOTcube, but it defines the processing environment into which PHOTcube outputs are inserted.

The uncertainty model described for PHOTcube itself is similarly pragmatic. Flux uncertainties come from magnitude errors provided by DAOPHOT and are propagated into spectral flux errors. Spectrum SNR is defined as the mean SNR per wavelength pixel, where per-pixel SNR is flux divided by associated uncertainty. Quality control includes manual inspection of PSF stars, visual inspection of blended profiles, and later visual inspection of the selected synthetic template and cross-correlation function. The paper notes that spurious radial velocities are rejected, typically removing 10–30% of the faintest stars, and that blends can sometimes be identified through broader or double CaT lines (Quezada et al., 8 Sep 2025).

5. Validation, comparison with PampelMuse, and limitations

The paper’s validation strategy is empirical rather than benchmark-driven. Its main direct comparator is PampelMuse, another crowded-field IFU extraction code. Three advantages are claimed for PHOTfun/PHOTcube: PampelMuse requires a predefined list of targets, does not allow interactive selection of PSF-model stars, and is described as less effective at subtracting sky artifacts compared to DAOPHOT-II-based methods (Quezada et al., 8 Sep 2025).

The sky-subtraction distinction is central to the comparison. PampelMuse relies on global sky subtraction provided by the MUSE pipeline, whereas PHOTfun/DAOPHOT computes a local sky estimate within an annulus around each detected source. The paper argues that this local sky subtraction improves removal of residual OH sky features. In the comparison discussed by the authors, PHOTfun-PHOTcube spectra are described as smoother around strong OH lines than PampelMuse spectra, and an “empty sky” extraction is used to display residuals left after global MUSE-pipeline subtraction.

The paper also states the regime in which PHOTcube performs especially well: crowded fields, fields with strong blending, fields with significant local sky variability, and dense stellar regions such as the inner bulge. These are all observational conditions that directly motivated the method.

Its limitations are also explicitly stated or inferable from the workflow. Extraction depends on good PSF-star selection, and manual intervention is recommended. Faint stars remain problematic, as shown by later RV quality-control rejection. Blends can survive extraction and may only be recognized subsequently from broadened or double CaT lines. No public quantitative benchmark is given for completeness, extraction bias, or runtime. The paper also does not provide a PHOTcube-specific mathematical PSF model or a full uncertainty-propagation formalism. This suggests that PHOTcube is best understood as a usable, DAOPHOT-based crowded-field extraction extension whose validation rests on practical spectral quality and downstream scientific utility, rather than on a standalone formal performance study (Quezada et al., 8 Sep 2025).

6. Mathematical content, scientific role, and reproducibility

The source paper is explicit that it provides no PHOTcube-specific extraction equation, no PSF functional form for the extension itself, and no explicit likelihood for slice-by-slice photometry. PHOTcube is therefore described procedurally rather than through a dedicated analytical formalism (Quezada et al., 8 Sep 2025).

The nearest formal equations in the paper belong to downstream analysis. The radial-velocity uncertainty relation adopted from Valenti et al. (2018) is

$$\ln(\epsilon_{RV}) = 4.624 - 1.023 \ln(S/N) - 0.159 [Fe/H] + 0.120 [Fe/H]^2.$$

The heliocentric-to-galactocentric conversion is

$$\begin{aligned} V_{GC} &= V_{HC} + 220 \sin(l) \cos(b) \ &\quad + 16.5 \left[ \sin(b) \sin(25) + \cos(b) \cos(25) \cos(l - 53) \right]. \end{aligned}$$

The analytical fits for the bulge kinematic maps are

$$\mathrm{RV}_{GC} = \left(A + B b^2\right) + \left(C b^2\right) l + \left(D + E b^2\right) \tanh(F l),$$

$$\sigma \mathrm{RV}_{GC} = \left(A + B e^{-b^2/C}\right) + D l^2 + \left(E e^{-b^4/r}\right) e^{-l^2/s}.$$

These equations are not PHOTcube algorithms. Their significance is that they define the downstream scientific products to which PHOTcube-derived spectra contribute.

That contribution is substantial within the paper’s scientific program. The new MUSE data were processed using PHOTfun, the PHOTcube extension extracted stellar spectra from the MUSE datacubes, and those spectra then yielded radial velocities for the new inner and outer MUSE fields. The study reports nine new MUSE fields, a final kinematic map based on 57 bulge fields, and about 23000 individual RV measurements. For several of the new fields, the adopted stellar samples are large, including 1873, 1707, and 1990 stars in three example fields. The paper states that these new MUSE fields improved sampling close to the Galactic plane and that the updated velocity-dispersion morphology became smoother and boxier, with previously inferred wing-like features near the plane disappearing (Quezada et al., 8 Sep 2025).

Reproducibility is addressed directly. PHOTfun/PHOTcube is public, with source code on GitHub, distribution via PyPI, and Docker-based DAOPHOT enablement. The same paper also notes that there is no separate dedicated PHOTcube methods paper; the method is presented within this work itself and in the appendix. This suggests that, in the present literature, PHOTcube is defined primarily by its integration into PHOTfun and by its role in the MUSE crowded-field extraction workflow, rather than by an independent software publication.