Are there in-built functions for reading arrays from image interfaces? I have an interest in adding a 'save as .png' and 'save as .cur' to my deck, minart, but I'm not sure where to start when it comes to image conversion. Is this something that should be done externally?
That's a rather expansive question!
Exporting GIF images with write[] and asking users to convert them to other image formats as desired will generally be the simplest approach. Decker natively supports GIF because it's a simple and universally-supported format that is also a good match for Decker's paletted-color model.
If you decide to dig deeper, the image interface has a .pixels attribute which, when read, will give you the pixel values of that image as a list of lists of numbers (a matrix).
These numbers will be Decker pattern indices, so you'd need to do some work to "flatten out" 1-bit patterns and animated patterns (if applicable) and then look up the corresponding RGB colors from Decker's active palette. The internals of the PDF module might be a useful reference.
In principle, you could use an Array interface to construct your own PNG encoder in pure Lil, but this could be a significant amount of work! The CUR format is a bit simpler than PNG and (so far as I'm aware) builds on BMP, which is also relatively straightforward. An encoder is, at least, quite a bit simpler than a decoder, since you can avoid implementing all the features and options you don't need for your intended application.
If you only care about PNG export from web builds of your tools, you could also consider writing a module that uses the danger zone to call some JavaScript from Lil and use ordinary web APIs to perform the conversion. See The Forbidden Library for examples of doing this kind of JS/Lil interop.
Does any of that point you in the right direction?