What’s New in Photutils 1.13?

Here we highlight some of the new functionality of the 1.13 release. In addition to these changes, Photutils 1.13 includes a large number of smaller improvements and bug fixes, which are described in the full Changelog.

Testing the installed version of Photutils

To test your installed version of Photutils, you can run the test suite using the pytest command. Running the test suite requires installing the pytest-astropy (0.11 or later) package.

To run the test suite, use the following command:

pytest --pyargs photutils

This method replaces the old method of running the test suite using the photutils.test() Python function, which has been removed.

Datasets subpackage reoganization

The photutils.datasets subpackage has been reorganized and the make module has been deprecated. Instead of importing functions from photutils.datasets.make, import functions from photutils.datasets.

Changed noise pixel values in example datasets

The randomly-generated optional noise in the simulated example images returned by the make_4gaussians_image and make_100gaussians_image is now slightly different. The noise sigma is the same, but the pixel values differ. This is due to a change from the legacy NumPy random number generator to the redesigned and preferred random number generator introduced in NumPy 1.17.

Making simulated images with model sources

The new make_model_image() function creates a simulated image with model sources. This function is useful for testing source detection and photometry algorithms. This function has more options and is significantly faster than the now-deprecated mask_model_sources_image function.

A new make_model_params() function was also added to make a table of randomly generated model positions, fluxes, or other parameters for simulated sources.

These two new functions along with the existing make_random_models_table() function provide a complete set of tools for creating simulated images with model sources. Please see the examples in the documentation of these functions.

The make_model_sources_image, make_gaussian_sources_image, make_gaussian_prf_sources_image, make_test_psf_data, and make_random_gaussians_table functions are now deprecated and will be removed in a future release.

Making simulated images with a PSF model

A specialized function, make_psf_model_image() function was added to generate simulated images from a PSF model. This function returns both an image and a table of the model parameters.

PSF photometry initial parameter guesses

The init_params table input when calling the PSFPhotometry or IterativePSFPhotometry class now allows the user to input columns for additional model parameters other than x, y, and flux if those parameters are free to vary in the fitting routine (i.e., not fixed parameters). The column names must match the parameter names in the PSF model. They can also be suffixed with either the “_init” or “_fit” suffix.

Removed deprecated PSF photometry tools

The deprecated BasicPSFPhotometry, IterativelySubtractedPSFPhotometry, DAOPhotPSFPhotometry, DAOGroup, DBSCANGroup, and GroupStarsBase, and NonNormalizable classes and the prepare_psf_model, get_grouped_psf_model, and subtract_psf functions were removed.

Updates to Star finders

The DAOStarFinder, IRAFStarFinder, and StarFinder classes and the find_peaks functions now support input arrays with units. This requires inputing a threshold value that also has compatible units to the input data array.

Sources that have non-finite properties (e.g., centroid, roundness, sharpness, etc.) are now automatically excluded from the output table in DAOStarFinder, IRAFStarFinder, and StarFinder.

The sky keyword in DAOStarFinder, and IRAFStarFinder is now deprecated and will be removed in a future version. One should background subtract the image before calling the star finders.

Improvements to Radial Profile tools

The CurveOfGrowth class now has calc_ee_from_radius and calc_radius_from_ee methods to calculate the encircled energy (EE) at a given radius and vice versa using a cubic interpolator.

The CurveOfGrowth and RadialProfile classes now have a unnormalize method to return the profile to the state before any normalize calls were run.

Other changes

Please see the Changelog for the complete list of changes.