Re: [DCP-o-matic] Chinese subitles - long tale

Am 31.03.2016 um 22:32 schrieb Carsten Kurz: Okay, Chinese subtitles did not have the highest priority for this filmmaker, but finally we both found time to get this straight at our cinema. It's always complicated if you want to deal with a language that you don't understand, and even more so if you can't even read the glyphs... First issue was that the default font that DCP-o-matic uses on the Mac is LiberationSans, which actually took care to render the subtitles properly in 8 languages, among them spanish with proper accents, greek and cyrillic. Now if these damned modern OS's wouldn't be so user friendly, he would have noticed immediately that this font he chose does not contain the actual chinese chars/glyphs he needed. The chinese SRT file was prepared by a girl from taiwan, using a translation system called https://en.wikipedia.org/wiki/Bopomofo . Actually I am not even sure wether this scheme will result in 'proper' chinese subtitles or which of the chinese folks will actually understand this version. But it's all we have for now. Creating and forwarding the chinese SRTs between different computers, language settings, editors, file formats etc. all showed no code translation issues. Which is a miracle in itself... Even in DOM, a 'universal' system font collection takes care that all UTF chars in the SRT are rendered properly for the preview. That is, DOMs preview does not use the actual font assigned to the subtitles under Content->Subtitle->Fonts. Means, all looks well in preview, until you load/play the DCP with the assigned font on an actual DCI projector. Even EasyDCP Player as his preview tool used that system font selection and not the actual font packaged in the DCP. Now when we played the DCP on a DCI projector, no subtitles showed up, because projectors do use the supplied font. After I understood this issue, I collected a couple of fonts which I suspected would contain the necessary chinese glyphs and would be free to use. Another technical issue would be that because of legacy/series 1 projection systems in the field, a single font file shell not exceed 640KByte in size. Even the mentioned Liberation Sans with many glyph systems for all european languages, even greek and cyrillic, is much smaller than that. However, all chinese fonts I found were much bigger, easily exceeding 5-10Mbytes. So I knew we would probably need TIs DLP Font Compressor utility at some point. So today we threw everything together and fixed it at our cinema so we could test the resulting DCPs immediately on our Sony projector. First, we created a VF file in DCP-o-matic using the OV and without specifying a specific font. - this simply converted the original SRT subtitle file to a DCI XML subtitle file. We trashed this VF, keeping only the XML subtitle file, because that is what Font Compressor needs, it can't work on SRT or other subtitle formats. We then installed TI DLP Font-Compressor on a Windows machine. This tool is straightforward to use, but initially complained about chinese TTFs from our Macs - there would be 'cmap table missing', etc. It did work however with the chinese font 'SIMHEI', which seems to be contained in all windows versions since XP. I'm not exactly sure wether this Microsoft supplied font is free to use. Googling for it shows multiple places suggesting free downloads of this font. Anyway, for now we went ahead with SIMHEI. Throwing the chinese subtitle XML and SIMHEI.TTF onto TIs Font Compressor results in a new TTF font file being generated, containing only the chinese glyphs the XML file actually uses, and discards thousands of others. Means, this new font is tailored to the actual XML subtitle file used for this DCP and should/can not be used for other subtitle files/DCPs. So it's a good idea to give the new shrinked font file a unique name to indicate this, e.g. SIMHEI_<your movie title here>. This brought the original 9.2MByte SIMHEI file down to a mere 30KByte. We then created a new VF from the OV, adding the chinese subtitle XML and customized font file. This actually went well! We did see the chinese glyphs on screen and compared them to those in DOMs preview, and they matched. We also took a video of the subtitled short playing, so we can show it to the translator for qc. Later, just for the fun of it, I generated another VF, using the original 9.2MByte SIMHEI font, bypassing the Font Compressor pass. This actually played well on our Sony 515, so it seems these systems can digest font files at least up to 10MBytes in size. I have never heard real numbers about DLP series II font file size limitations so far (the 640KByte limit is for series I). Anyone? So that's that sorted out... - Carsten