Spectrophotometric Background Extraction?


The newly released Spectrophotometric Color Calibration (SPCC) feature is simply amazing. I wonder if you have any plans to release a Spectrophotometric Background Extraction feature that can automatically generate sky background models using spectrophotometric data of stars
I have already answered similar questions before. Many people expect this to be doable, and I understand why. However, unfortunately, star catalogs provide no helpful information to compare local background levels measured on the image to the actual sky local background at each star location (e.g., from data acquired without atmosphere, such as Gaia catalogs). Stellar photometry alone does not provide useful information for effectively modeling additive gradients.

Gradient correction is a very complex and challenging task. For accurate and reliable gradient correction we need observational data; no purely software-based solution can provide it. We have developed a methodology that is being applied with great success:

We have been discussing with our development team the possibility of implementing this multiscale gradient correction method as a new automated process in PixInsight. This would require a database with accurate and robust sky background data available for the whole sky. After an initial exploration, we concluded that such data does not exist, or at least we cannot access it. So we'd need to acquire it ourselves. An extremely interesting project and certainly doable with the necessary human and technical resources... could be a collaborative effort.
Juan and I wrote about this some time ago. Since then, I found some further information: https://iopscience.iop.org/article/10.1086/648480#pasp_121_885_1180s3

The interesting part begins at section 3.4. The author of this project used space based photometry data of the background brightness as a reference for gradient reduction. I tried to access these data, but I was unable to find anything relevant about it. Maybe this could provide a solution for this problem?

At least for me, gradient reduction is the biggest remaining problem in image processing as of now, since deeper images make it almost impossible to find pure background data in a single image and an observational approach is often practically impossible, at least in my case.

CS Gerrit