[mythtv-users] IPad and mythtv.

[Phil Linttell]

On 10-05-27 12:10 PM, mythtv-users-request at mythtv.org wrote:
> Date: Thu, 27 May 2010 08:55:23 -0400
> From: Raymond Wagner<raymond at wagnerrp.com>
> Subject: Re: [mythtv-users] IPad and mythtv.
> To: Discussion about mythtv<mythtv-users at mythtv.org>
>
> On 5/27/2010 08:33, Phil Linttell wrote:
>> >  Maybe what mythtv needs is a streaming plug-in that can be configured to
>> >  do on-the-fly transcoding according to different mobile device screen
>> >  resolutions, available bandwidth, and codec support (and potentially
>> >  control protocols.)
> The backend does not support plugins, only the frontend.  While there
> has been some discussion of duplicating the plugin code into the
> backend, such live transcoding (to something other than nupplevideo)
> would ideally be integrated into the backend.  Have some generic method
> that could be used in the backend, and proxied through MythProto, UPnP,
> MythWeb, or anything else you wanted.
>> >  There's a whole variety of mobile devices and
>> >  embedded TV front-ends these days, and it's simply not practical to do
>> >  static transcoding of all recorded content to each potential target.
>> >  
> Why not?  iPods, phones, PSPs, are all going to have similar
> capabilities in terms of their hardware decoders (and none in terms of
> software decoders).  TVs are likely able to playback anything you record
> directly.  That means its a single transcode to support most devices
> available.  A near term plan for MythTV by one of the devs is to rework
> the storage of recordings, to allow multiple files to be stored against
> a single recording.  This would significantly improve management of
> multiple transcodes.
I'm not suggesting the implementation at this stage...  a on-the-fly 
transcoding/streaming capability could be done as new back-end 
functionality, through some form of plug-in interface, or as a 
stand-alone server application that can talk to the database (via 
python, etc.)   I don't see any reason that it couldn't be done 
stand-alone at this stage, in a similar model to AirVideo Server but 
with closer integration with the database and more flexibility in the 
transcoding options / clients.

iPods, phones, PSPs, (PS3's, XBOX's, UPnP TV's) all have different 
codecs or codec profiles/options they support, and while with something 
like HandBrake you can make a least common denominator h264 profile come 
*close* to supporting multiple devices, it will not be optimised for 
differences in screen resolution or bandwith.  Nor would you want the 
same profile used for a smartphone that you use for a tablet.  Further, 
codec support is getting more complicated with HTML5.

In my case, I'd need to transcode to at least four different 
codecs/profiles/resolutions in order to support the mobile devices I'd 
like to be able to use in my house.    No way am I going to devote that 
amount of processing time, electricity, and storage to transcode each 
recording to only or two of them.   I'd rather have available processing 
capacity on the back-end server to transcode/stream on the fly, such 
that I can watch whatever I want from whatever device (and network) I'm on.

(BTW, if you're wondering why the difference in device form-factor is an 
issue.....  you don't want to be sending higher resolution to a device 
that can't display in that resolution, even if the device can scale down 
the video.   You're forcing a lot of extra processing, which directly 
translates to reduced battery time.  Further, it'll limit the network 
you can use.... you might manage wifi, but you might not managed 3G), 
make it more difficult to fast-forward/rewind, etc.

Today, myth can stream to a native front-end, via UPnP, and via 
transcoding in mythweb.  There are various, convoluted solutions for 
supporting specific mobile devices/platforms.   I'd argue that most of 
use would like to be able to watch Myth recordings from whatever mobile 
devices we use (some will prefer streaming, some will prefer 
downloading, but both can be achieved), and that "native" support for 
on-the-fly transcoding/streaming according to different device/network 
is a natural evolution for myth.