Advanced Subtitle Filter/Renderer

[Omni]

Since several month I'm working on a 3d subtitle/text renderer to replace my "own" old Mental Ray based text effect/renderer system. The old system was kinda slow but allowed almost all kinds of text/effects with very less scripting. Basically it was like vsfilter + scripting support but since i used a raytracer it was never meant for realtime.
For the past few months I've been working on a replacement which will work in realtime using either opengl or directx to render fonts including their outlines. Now i reached a point where i asked myself if it makes sense to continue this project and maybe develop something others could have a use for (i.e. as replacement for vsfilter) or just stop waisting time if no one needs it^^.
I wonder if there is a basic interest in such a filter/renderer and if someone would be interested to help in the development.
The raw speed is quite fast so far just the stroker still needs improvements but for now it's "usable". the text can be rendered 2d or real 3d extruded allowing textures and well whatever you know from 3d applications/games ^_^
Everything should be built quite modular so that it's "easy" to integrate where it's needed. i.e. as avisynth filter, for aegisub, or as directshow filter.
So anyone who'd be interested?

[Soulhunter]

GFX accelerated subtitle renderer? (http://soulhunter.chronocrossdev.com/index.html#006) :]


Bye

[Omni]

basically yes^^
but don't know anyone who's really working on it. at least i am and i can at least tell that this is quite promising, but a lot of work :)

[Soulhunter]

Well, its good to see someone is working on something like this! I guess there are a lot ppl out there that will have use for it... Maybe posting about it on some fansubber centered boards will give you more response!? I wish you good luck with this project... I'll stay tuned for news! ^^


Bye

[TheFluff]

Potentially GPU accelerated subtitle renderer with highly ambitious goals for replacing what is effectively the "standard" subtitle renderer on Windows? Why I've never heard (http://sourceforge.net/projects/libass) of such a thing (http://asa.diac24.net/).

Seriously though all developers I know of who would have even the faint interest in working on something like this all hang out in #aegisub, and hence would be more interested in contributing to something that already has a lot of the nontrivial parts (like SSA lexing, proper collision detection, animations) done, like, oh, Asa (http://asa.diac24.net/). Vaporware projects with ambitious goals but no actual code and very vague outlines of what's done and not may attract fanboys and wild speculations, but they rarely attract competent developers.

[Soulhunter]

For the last part of your post... I bet comments like this really motivates potential developers of ambitious projects n encourages em to progress to a point where actual code will be released! :) :rolleyes:


Bye

[TheFluff]

Yeah, well, sorry about that, not the best thing to say (I'm actually not trying to troll here), but still, the fact is that the OP hasn't posted any actual code nor any real description of what his renderer is capable of today. No matter how much you talk about it, vaporware does not magically turn into software. Why not contribute to an existing project instead? I'd think that would be more helpful to the community overall.

I might add that despite at least four attempts (that I know of) and two years, noone has actually succeeded in writing a complete SSA/ASS renderer that supports all the features in common use today, much less anything that could replace VSFilter once and for all. libass is pretty close to fulfilling the first requirement but it still lacks a bunch of rather important rendering features. Hence I reserve the right to stay skeptical. :V

Actual helpful tip: if you're really serious about this, you might want to start hanging out in #aegisub on irc.rizon.net.

[Omni]

Actually i've been in the aegisub channel for some time. I just had to take a break due to exams and well when i came back i heard that asa currently seems to be stalled. My plan was to integrate into the asa development but well currently it doesn't look like that'll happen.
Concerning the rest, i actually have to agree to TheFluff that talking about what can be done and actually showing a working code is quite a difference.
For now i've just several test code fragments which worked quite fine but well before i'm able to generate nice outlines (currently still bit bugged) i publish anything *sorry*
if anyone is interested visit the aegisub forum for further discussions :)
thanks for the feedback.