Just a few comments to the topic
(tl;dr: “it’s still unpredictale and unclear for me”)
Few mothns ago i did automation script to merge attachments from branch to main and hit ‘duplicates issue’(when you merge main’1,2’ with branch’3,4’, you receive ‘1,2,3,4’, but after branch update record(4=>5) it will merge 3,5 to 1-4, and 3 became duplicate. Then I changed it to ‘merge removing duplicates’ and discovered another problem because similar file in two different fields has different links, and defininition “same name=duplicate” is of course wrong. At least, should be name&size
Now i did some utility to backup/restore attachments including small script to display size of all attachments in all tables, and discovered that their actual downloaded number and size less than expected by script (link to attachment contain 8-digit code, which i append to filename for restore purposes, so i supposed 10 similar files must have 10 downloaded “versions”). And it’s not true, Airtable can detect such case and store only one file, with 10 links to it.
But when i tried to test it and uploaded to small table, 3 files repeatedly to 10 records in random order and number per 1 record (by usual upload through user GUI), i found that all of them has different link, none repeated. The method of storing attachments in Airtable is a mystery for me again.
Now i doubt that “Airtable can detect such case”. maybe we just cloned some records with sync or automation ot “horizontal to vertical” table transfer.