Here's a detailed example.
In Zim, I created a sample page and a subpage, both with images, with this structure:

The first page looks like this:
The subpage looks like this:
I export this "tree" using Zim > Export > Single page (with sub-pages), Markdown format
Zim creates a folder structure, I attach a zip file where with the folder that Zim creates, you can see both the ".md" and the images are all there, the images are in sub-folders that have the same name as the .md files. They seem to be properly "linked" from the .md files.
Importing with Joplin (importing the entire folder), I see Joplin picks up the structure:

It also picks up the first page, but misses all the images:
And the sub-page is completely empty:
Here's the Zim export folder:
Zim Export Test - for Joplin.zip (258.2 KB)
If I import in Joplin the .md files one by one (instead of the folder), I can see the content in both pages (i.e. no empty pages), but the images are missing in both cases.
So two problems I am facing to import my zim workbook: a) all inserted images are lost, and b) subpages are coming up with empty content (no content at all). Importing pages one by one at least gets the content of the page all right, but is not scalable (as a long-time zim user, I have around 1K notes in the workbook)
Thanks @joplin_team for looking into this