And here is POD source for the new section I am considering:
=head1 CHARACTER ENCODINGS
Certain meta information formats allow coded character sets other than plain
ASCII. When reading, 8-bit encodings are passed straight through ExifTool
without re-coding unless specified otherwise below, and multi-byte encodings
are converted according to the L</Charset> option ('UTF8' by default). When
writing, the inverse conversions are performed. See the L</Charset> option
for a list of valid character sets.
More specific details are given below about how character coding is handled
for EXIF, IPTC, XMP, PNG, ID3, PDF and MIE information:
=head2 EXIF
Most textual information in EXIF is stored in ASCII format, and ExifTool
does not convert these tags. However it is not uncommon for applications to
write UTF-8 or other encodings where ASCII is expected, and ExifTool will
quite happily read/write any encoding without conversion. For a few EXIF
tags (UserComment, GPSProcessingMethod and GPSAreaInformation) the stored
text may be encoded either in ASCII, Unicode (UCS-2) or JIS. When reading
these tags, Unicode and JIS are converted to the character set specified by
the L</Charset> option. Other encodings are not converted. When writing,
text is stored as ASCII unless the string contains special characters, in
which case it is converted from the specified character set and stored as
Unicode. ExifTool writes Unicode in native EXIF byte ordering by default,
but this may be changed by setting the ExifUnicodeByteOrder tag. The EXIF
"XP" tags (XPTitle, XPComment, etc) are always stored as little-endian
Unicode, and are read and written using the specified character set.
=head2 IPTC
The value of the IPTC:CodedCharacterSet tag determines how the internal IPTC
string values are interpreted. If CodedCharacterSet exists and has a value
of 'UTF8' (or 'ESC % G') then string values are assumed to be stored as
UTF-8, otherwise Windows Latin1 (cp1252, 'Latin') coding is assumed by
default, but this can be changed with the L</CharsetIPTC> option. When
reading, these strings are converted to the character set specified by the
L</Charset> option. When writing, the inverse conversions are performed.
No conversion is done if the internal (IPTC) and external (ExifTool)
character sets are the same. Note that ISO 2022 character set shifting is
not supported. Instead, a warning is issued and the string is not converted
if an ISO 2022 shift code is found. See L<
http://www.iptc.org/IIM/> for the
official IPTC specification.
=head2 XMP
Exiftool reads XMP encoded as UTF-8, UTF-16 or UTF-32, and converts them all
to UTF-8 internally. Also, all XML character entity references and numeric
character references are converted. When writing, ExifTool always encodes
XMP as UTF-8, converting the following 5 characters to XML character
references: E<amp> E<lt> E<gt> E<39> E<quot>. By default no further
conversion is performed, however if the L</Charset> option is other than
'UTF8' then text is converted to/from a specified character set when
reading/writing.
=head2 PNG
L<PNG TextualData tags|Image::ExifTool::TagNames/"PNG TextualData Tags"> are
stored as tEXt, zTXt and iTXt chunks in PNG images. The tEXt and zTXt
chunks use ISO 8859-1 encoding, while iTXt uses UTF-8. When reading,
ExifTool converts all PNG textual data to the character set specified by the
L</Charset> option. When writing, ExifTool generates a tEXt chunk (or zTXt
with the L</Compress> option) if the text doesn't contain special characters
or if Latin encoding is specified; otherwise an iTXt chunk is used and the
text is converted from the specified character set and stored as UTF-8.
=head2 ID3
The ID3v1 specification officially supports only ISO 8859-1 encoding (a
subset of Windows Latin1), although some applications may incorrectly use
other character sets. By default ExifTool converts ID3v1 text from Latin to
the character set specified by the </Charset> option. However, the internal
ID3v1 charset may be specified with the L</CharsetID3> option. The encoding
for ID3v2 information is stored in the file, so ExifTool converts ID3v2 text
from this encoding to the character set specified by the L</Charset> option.
ExifTool does not currently write ID3 information.
=head2 PDF
PDF text strings are stored in either PDFDocEncoding (similar to Windows
Latin1) or Unicode (UCS-2). When reading, ExifTool converts to the
character set specified by the L</Charset> option. When writing, ExifTool
encodes input text from the specified character set as Unicode only if the
string contains special characters, otherwise PDFDocEncoding is used.
=head2 MIE
MIE strings are stored as either UTF-8 or ISO 8859-1. When reading, UTF-8
strings are converted according to the L</Charset> option, and ISO 8859-1
strings are never converted. When writing, input strings are converted from
the specified character set to UTF-8. The resulting strings are stored as
UTF-8 if they contain multi-byte UTF-8 character sequences, otherwise they
are stored as ISO 8859-1.
