Editing the field "Description:" in the metadata pane seems to prefix the text with the characters xEFxBBxBF.
Tested with .jpg files with no initial Exif data and adding new data. Tested with .jpg files with existing Exif and then editing the data. The results were viewed in other software in the Exif field 0x010e ImageDescription. Some software ignored the three characters, some displayed dots and some displayed non ASCII characters.
Opus 12.3 6183 x64 and updated Win 10 x64
If your talking about the extended properties description field, i can't confirm this. With which camera are the images taken with? Do you use the meta data panel, or the one in the image viewer?
The field is called "Description:" and is near the top in the "Document Properties" section. Opus populates the Exif field 0x010e ImageDescription with the text from there. "Description:" is the only field where I found this to occur.
I discovered this on test jpg files, stripped of Exif metadata, while investigating what metadata fields Opus populates. They were not files from a camera. But I just tested a file from a camera and got the same results.
Some software, including Opus, will not display the extra characters. IrfanView displays non ASCII characters and can be copied and pasted into a text editor to determine their hex values. JPEGsnoop shows dots in place of the characters. So it appears to me that they actually do exist in the metadata field. In testing other software's metadata populating behavior I do not see extra characters using my test methods.
This was using the metadata panel (F9) and I have not tried the from within the image viewer.
That is a UTF-8 BOM, added so that things reading the string can be sure it is UTF-8 and not some other encoding.
The BOM is uncommon with Linux/Unix software, where just about everything is always UTF-8, but very common in the Windows world where a much larger mix of encodings are common.
Are you seeing those characters when you read the field back in Opus, or are you only seeing them in other software? It's possible the other software doesn't understand UTF-8 or doesn't know how to handle the BOM marker.
Thanks for the reply Leo.
I thought it might be that but I don't speak Unicode so couldn't decipher it. A quick look at the Exif Version 2.31 document seems to say that the ImageDescription field can only contain ASCII. At least for that field, you might consider not adding the UTF-8 BOM just for compatibility with others.
I sure would like to use Opus more and other software less for maintaining metadata 
Some discussion from a related Stack Overflow answer:
Exif suffers from the common encoding misery, the Exif standard dictates that only 7-bit ASCII codes should be used but everybody ignores it. They have to, ASCII just can't properly encode text in many languages. Pretty remarkable btw, Exif comes from Japan, a country with a language that has very little use for ASCII and a rich history of encoding problems. So everybody just picks whatever encoding suits them, could be UTF8 or could be ANSI, whatever code page is in common use where the image was created.
Between a rock and a hard place, using UTF8Encoding is the best choice.
I'd expect any decent EXIF tool to support UTF-8 these days, as so much other software uses it, and basic ASCII simply cannot store all the strings people want to use.
Point well taken. Thanks for the education. I should be using newer imaging software.