TextMesh Pro TMP 2.0 and Arabic

Hello,

I've heard that TMP 2.0 in 2019.1 will bring more support to RTL languages like Arabic, so I gave it a try.

I made a new SDF using the Calibri font and the following Unicode ranges: 600-6FF,750-77F,0020-007F
This is the text I am trying to write: "هذا نص عربي". (Link on google translator).
This is the result:


The upper is using TMP normal text editor, and the lower is using the RTL editor. Both cases are not accurate, as the letters appear correctly but separately. In Arabic, the letters has to be connected, not shown separately.
The upper case is a bit more correct though, as it shows the letter ordered correctly from Right to Left. The lower (RTL editor) shows the letters in reverse order.

Now using an RTLTMP package, (Link to forum thread) this is the result:


Same letters, but in correct order, direction and letter connections.

I tried using the SDF attached with the RTLTMP package with TMPro, but it didn't work. I am using TMP 2.0 in a 2019.1 project, and RTLTMP in a 2018.3 project.
For some reason, I am getting random crashes (in both projects) whenever I try making a new Arabic SDF (same steps as I used before).

I don't know how Arabic and RTL support in TMP 2.0 was tested, but the results in the first image usually happen on PC's that do not have full Arabic support installed or have a partial support, so the letters will appear correctly (which may give the impression that Arabic is fully supported), but their direction and connections are actually incorrect.

@Stephan_B

Arabic support is a really huge help to thousands of devs. Please let me know if I missed something
Thanks.

 

we have a big unity arabic game (opensource) https://vgwb.github.io/Antura_arabic/ where i hooked the TMP engine to render perfectly the diacritics and everything.

now we have a ported the project to 2018.3 but still using the old TMPro lib.. it's not super easy to upgrade to 1.3.0 but we'll do soon.. i'll write here the results

 

That would be helpful, thanks

 

I used the instruction on your site to make arabic font asset well the result is as always since I found about Text Messh Pro 1.0 "absolute no support for arabic" I don't even know why the author kept the RTL check box? other so called RTL languages can be written then aligned to right not as complexe script language like Arabic where letter connection and other particular stuff comes to play needs to be really processed from right to left not mirrored. Still waiting for a buit in solution for now using basic arabic text display like a dinosaur with RTL TMPro at least user can read it.
I had high hopes when Unity acquired Text mesh Pro but seems only RTL support offered is for asian languages.

 

Adding full support for OpenType font features which is require to properly support non Latin based languages like those using RTL is still in the plan / in the works.

I still don't have a firm ETA but I am working in it and it will happen. Until then please take a look at the free RTL plugin.

 

Thanks for your replay, looking forward for it.

 

Me too

 

Stephan hi. our Antura game is still using TMPro 1.4 and we developed a custom hack to fix the rendering of diacritics (modifying the xy offsets)
did you develop a native solution for diacritics in following releases? i could not find anything related

thank you!

 

Support for Diacritical Marks will be in the next preview release which will be version 1.6.0 which I would like to have available in the next few weeks.

 

oh thank you! we have to improve the rendering.. but now i'll wait for the native solution
will it still work on unity 2018?

 

The new Mark to Base and Mark to Mark tables will be available in 2018.4 but will need to be manually populated as the ability to extract the data for these tables from the font file will only be available in 2019.4 or newer and require an updated version of Unity for each.

 

hallo. sorry to ask: any idea if 1.6.0 preview will be available soon? thx

 

Any update on that? I see that version of TMP that support Mark 2 Mark and Mark 2 Base are in Pre Release, but will that also bring native support for RTL language like arabic?

 

hallo @Stephan_B , we updated to 2020 and TMPro 3.2 beta to check the new diacritics system.. it works in the editor, but the runtime seems to be hardcoded to recognize just a few diacritics (so Arabic doesn't work!!)
how can we go on? thank you

Code (CSharp):

1. internal static bool IsDiacriticalMark(uint c)
2. {
3.     return c >= 0x300 && c <= 0x36F || c >= 0x1AB0 && c <= 0x1AFF || c >= 0x1DC0 && c <= 0x1DFF || c >= 0x20D0 && c <= 0x20FF || c >= 0xFE20 && c <= 0xFE2F;
4. }
5.  
6. internal static bool IsBaseGlyph(uint c)
7. {
8.     return !(c >= 0x300 && c <= 0x36F || c >= 0x1AB0 && c <= 0x1AFF || c >= 0x1DC0 && c <= 0x1DFF || c >= 0x20D0 && c <= 0x20FF || c >= 0xFE20 && c <= 0xFE2F ||
9.         // Thai
10.         c == 0xE31 || c >= 0xE34 && c <= 0xE3A || c >= 0xE47 && c <= 0xE4E ||
11.         // Hebrew
12.         c >= 0x591 && c <= 0x5BD || c == 0x5BF || c >= 0x5C1 && c <= 0x5C2 || c >= 0x5C4 && c <= 0x5C5 || c == 0x5C7
13.         );
14. }