Because vMix lacks support in JSON data sources for using row 1 as a column name, I attempted to build support for it within my Companion Google Sheets module (as that appropriately uses the Google API, so can poll significantly faster than vMix could before running in to 429 ratelimit errors).

One thing I noticed though, is that if the values in Row 1 contain certain special characters (and these values would be the ones I'm attempting to make Column Names), then when those special characters are in keyes in the JSON objects they get replaced in vMix with "a:item". and if multiple keys have special characters vMix will drop all of them except the first.

In the following image, you can see a JSON document, that is valid JSON and has any special characters that need escaping appropriately escaped, but the result in vMix is just a table of foo, bar, and a:item.
vMixJSON.png (12kb) downloaded 0 time(s).

The exact same data, but this time encoded as a CSV does not encounter this issue and vMix has no trouble using those special characters as Column Names.
vMixCSV.png (13kb) downloaded 0 time(s).

So my questions are, why doesn't vMix natively support using Row 1 as Column Names in JSON data sources? and if the JSON data itself contains special characters as object keys why are they being changed to a:item and breaking any table that contains more than one column that gets converted to a:item?