how can I tell which images get downloaded from the site vs images requested by user?

which images the user requested to save

A typical browser saves all images to disk, to some directory dedicated to files for temporary use. So you cannot tell whether the user has just viewed an image or "saved" it in terms that they would intentionally press "save" and choose a folder where they would want to put the image file. The http request asking for the image file is exactly the same in both cases.

to differentiate between images that were on the site itself vs images that the user requested to be downloaded

You can, but not very reliably:

• you may extract the html code, which you can also see in the capture, into a text editor and find links to images in it, which you then compare to the next html requests' targets,
• you may compare the timestamps of the html requests requesting the html code and requesting the images:
  
• if they are close to each other, the images were likely downloaded because the html code contained links to them,
• isolated (in time) requests to download a single image indicate intentional user activity.

So it may be easier and less time consuming to visit the site and see by your own eyes.