Here is a quote:
–hex-blob
Dump binary columns using hexadecimal notation (for example, ‘abc’ becomes 0×616263). The affected data types are BINARY, VARBINARY, the BLOB types, and BIT.

( http://dev.mysql.com/doc/refman/5.1/en/mysqldump.html#option_mysqldump_hex-blob )

BTW,
Here are some improvements:
—–

curdate=`date +%Y-%m-%d`

mysqldump -u user -pXXXXX -h myhost.com –hex-blob –set-charset –default-character-set=latin1 databasename > “$curdate”.latin1.sql

sed -i -e “1 iSET NAMES utf8;” -e ‘s/^\([)] ENGINE=MyISAM \(AUTO_INCREMENT=[0-9]\+ \)\?DEFAULT CHARSET=\)latin1;$/\1utf8;/; s#^\(/\*![0-9]\+ SET NAMES \)latin1 \*/;$#\1utf8 \*/;#’ “$curdate”.latin1.sql

iconv -f latin1 -t utf8 “$curdate”.latin1.sql > “$curdate”.utf8.sql

rm “$curdate”.latin1.sql

mysql -u user -pXXXXX -h myhost.com databasename < "$curdate".utf8.sql

—–
Of course you should change the user, password, host and database names for your own.
$curdate just hold the current date in the format YYYY-MM-DD, and used for filenames, you could give it any value.
I'm using just one sed, so it should be faster for large files.
The sed instructions are much more accurate and should never conflict with the data.

In that case I would only run this conversion tool on the non binary data, as far as I am aware there is no encoding for the BLOB data type, running this conversion tool may incorrectly try to convert the data.