Can confirm that robots.txt blocking access to the resource causes the problem. Unblocking the directory with the image in robots.txt fixed the problem for me.

Also, at first it didn't solve the problem, so after clicking [Scrape Again] dozens of times like a maniac, I thought that maybe Facebook was using a cached copy of robots.txt. So, I ran the url debugger on mydomain.com/robots.txt and then went back and scraped the home page again. The og:image updated immediately.