AMV Playback at Conventions... thoughts

[Quu]

I am thinking about changes to what i use to playback videos

I am insane
I am going to finish the alpha2 of ConPlayer as i have it... and then I am going to seriously rethink how I am doing this.

I am not sure if i should stick with a pure HTML 5 user interface, or make a "real" UI... Possibly keep both.

I am thinking of making a completely stand alone single executable file app with no dependencies that will use libav. If i make it a stand alone app... which UI tool kit do I use? GTK+, QT, wxWidgets. What do I use to output the video and audio, SDL, OpenGL, OpenAL, PortAudio?

Or another option is to start hosting my own MPlayer2 builds instead of using an external source... if the current version is appealing.

Mainly I need to figure out what works and what does not, and how it will work best... I need to rethink things now that I have more experience with the process.
http://quu.livejournal.com/832475.html

[Mister Hatt]

As opposed to just "use CCCP because it works"? MPlayer2 is nice and all but it's slow at some things and bad at others, and unless you specifically have a non-Windows machine, it's probably not the best thing to use at a con where performance is critical. Also, you're thinking too abstract at times (just use Qt) and then too low-level at others (who cares about sound servers, phonon says hi.)

[Quu]

i have a personal dislike of DirectShow due to experiences with dll hell and different filters trying to compete... it gets to be a pain, and is sometimes only solved with a reformat and a reinstall of windows. I can and have used DirectShow, but it is a pain. If it ends up being the best option, i can deal... but I feel like that is going to be a problem.

As far as non windows machines... that is actually something that allot of con directors have brought up to me... non linux requests, but more than a few mac requests. Can i make it work on a mac mini... can i make it work with a macbook pro... etc...

the importance of the sound server is that at a convention i would like to direct all of the default sounds from the OS to one device, and the audio from the AMV to another device.

and "just use QT"... is it the best? I have no UI experience outside of HTML based apps and java.

[Mister Hatt]

I wouldn't say Qt is the best but it's certainly the least annoying to do anything multiplatform, unless you want to look at stuff like Flex. Even without sound server hax, you can still direct things to different devices under QtMultimedia.

I'm not saying that dshow is perfect but rather, the libraries that you can use under it are a lot better. MPlayer2 still requires you to do stupid things with slave mode which makes it a royal pain to get MKV support that doesn't suck, let alone m2ts and friends. Haali's splitter "just works" and that's the main reason for dshow based setups, along with LAV filters and VSFilter I suppose. I've yet to see a proper setup for *nix that really works is all I'm saying.

[Quu]

i know about the stupid slave mode hax... trust me... i know

m,y only problem with linux setups is the audio configuration... being able to get the different audio devices configured... video works wonderfully.

[Mister Hatt]

MKV playback in non-MPlayer2 is broken (editions, segment linking, codec switches, although resolution switches work when they don't in Haali), transport stream playback/seeking is horrid, a lot of codecs aren't supported properly, etc. Of course, setting strict codec/format rules at conventions gets around this to a point. I've never had too much issue with audio in linux though, other than stupid latency because every soundserver sucks pretty much.

[Quu]

how well does libAV handle mkv?

[Mister Hatt]

Minimally at best. It's fine for single segment content mostly I suppose. There were patches for multi-segment (but not editions) however I forget if they were ever merged as they were for ffmpeg-mt back in the day. Certainly not merged into ffmpeg though.

[Phantasmagoriat]

Something I thought about the other day was how to maintain synchronization in large rooms because the speed of sound is 340.29meters/second or 0.00294seconds/meter.

...so @29.97frames/second x 0.00294seconds/meter = 0.0880895frames/meter
~~~> 1 frame unsynch for ever 12 meters. (Which can be pretty bad for people in the back row.)

I've never set up a contest before, but is there a certain speaker arrangement that works best?

[Mister Hatt]

You'd need a ~34m long hall (distance required for 100ms desync regardless of framerate) for that to become an issue, but generally halls designed for screening have the audio setup to take everything into account, in my experience at least. It's usually setup as controlled speaker latency in the hardware mixer, which is something the PC has no interaction with.