Please supply the shortest possible script that can be used 
to replicate the problem and the font file in question. 

<?php
$im = imagecreate(450,200);
	if ($im=="") { die("Error creating image"); }
$bg_color = imagecolorallocate ($im, 255, 255, 255);
$black = imagecolorallocate($im,0,0,0);

$ttf_tahoma = "[your_path]/ttf/tahoma.ttf"; // edit this

imagettftext($im, 11, 0, 10, 20, $black, $ttf_tahoma, "Example of chars are: ? and ? (Czech language)");

ImagePNG($im);
ImageDestroy($im);
?>

example output: http://ccres.net/char_test.php

Have you tried exporting Czech locale before creating the 
image? 

I'm not sure if I understand what you mean, I have added the following code:

setlocale(LC_ALL, 'cs_CZ');

Doesn't make any difference.

Provide the script downloadadble somewhere.

Some notes about using non-ASCII characters with GD 
extension:

1. Don't enable --enable-gd-jis-conv if you don't really 
understand what you are doing. The option is only 
meaningful when it comes to handling Japanese texts. 
Please refer to the relevant documentation included in 
the GD library distribiton.

2. The FreeType2 library which is used by GD to render a 
bitmap with a given TTF file always assumes the input 
string to be encoded in UTF-8. This means passing 
strings of another encoding (character set) doesn't make 
any sense and it will most likely end up with a bunch of 
square placeholders. Thus you have to encode strings by 
UTF-8 before passing it to ImageTTFText().

3. Make sure that you are using a correct TTF file whose 
glyphs are correctly mapped to the character set by 
which the string you want to rasterlize is represented. 
If it doesn't match, rendering will silently fail.

4. utf8_encode() can only handle the ISO-8859-1 
character set, where the character set that is most 
commonly used in Czech texts is ISO-8859-2. Use 
mb_convert_encoding() or iconv() instead.

 


 

And this might be helpful too:
http://www.voltex.jp/articles/imagettf_m10n.htm