NextPVR Forums
  • ______
  • Home
  • New Posts
  • Wiki
  • Members
  • Help
  • Search
  • Register
  • Login
  • Home
  • Wiki
  • Members
  • Help
  • Search
NextPVR Forums Public Developers v
« Previous 1 5 6 7 8 9 … 93 Next »
Time to make Touch web Service less experimental

 
  • 0 Vote(s) - 0 Average
Time to make Touch web Service less experimental
sub
Offline

Administrator

NextPVR HQ, New Zealand
Posts: 106,705
Threads: 767
Joined: Nov 2003
#31
2013-09-16, 06:23 PM
whurlston Wrote:Are you talking about the progress bar on the new progress popup? I'll double check my NeedsRendering()
It's not a bug in your plugin. It's just a limitation in the way the touch stuff currently works. It pretty much only returns a screen to a web client each time it receives stimulus like button press. ie, it's not updating the screen automatically when stuff just happens by itself.
whurlston
Offline

Posting Freak

Posts: 7,885
Threads: 102
Joined: Nov 2006
#32
2013-09-16, 06:35 PM
Good to know. Thanks.
fred250
Offline

Senior Member

Posts: 457
Threads: 10
Joined: Jul 2006
#33
2013-09-16, 08:26 PM
Inaccurate seeking when streaming due to crappy ts-timelines seem to be a problem in most client players that can't benefit from using timing.info.
I've made some tests using my own streamer and I think I may have a solution.
It's quite straight forward.

From the client just before firing the seek request to the player.
Using a web service, prime NextPVR with the desired seek position in time and a GUID identifying the client. Store the values in a list for later use.

When the streamer gets the request look up in the list using GUID and just start streaming from the corresponding byte found in timing.info.

In my tests I left the response headers untouched and let the client think it got the requested range - only changed the position from where to start reading the file. I wouldn’t say the player ever noticed.
This way we can get ~1 sec seek precision in most players.

/Fred
mvallevand
Offline

Posting Freak

Ontario Canada
Posts: 52,926
Threads: 956
Joined: May 2006
#34
2013-09-16, 08:36 PM
IMO, a smart server combined with a smart or semi-smart client would be disaster.

Martin
fred250
Offline

Senior Member

Posts: 457
Threads: 10
Joined: Jul 2006
#35
2013-09-16, 08:45 PM
mvallevand Wrote:IMO, a smart server combined with a smart or semi-smart client would be disaster.

Martin

Please elaborate...

/Fred
mvallevand
Offline

Posting Freak

Ontario Canada
Posts: 52,926
Threads: 956
Joined: May 2006
#36
2013-09-16, 09:55 PM
Short answer,

the smart client seeks by bytes,
smart server says you must want this time and streams bytes from time
smart server assumes location is correct and makes subsequent skips base of the same offset.
if the demux reads the file and the time line is off it might do an incremental seeks trying to find a good time
repeat until demux is happy

It's better to have a timing.info aware player or a totally dumb player that responds to a smart streamer but not both.

Martin
reven
Offline

Posting Freak

Posts: 5,782
Threads: 396
Joined: Sep 2004
#37
2013-09-16, 10:54 PM
very interested in this, if you manage to fix some seeking issues, hopefully that problem on samsung tvs Smile
fred250
Offline

Senior Member

Posts: 457
Threads: 10
Joined: Jul 2006
#38
2013-09-16, 11:08 PM
mvallevand Wrote:Short answer,
the smart client seeks by bytes,
smart server says you must want this time and streams bytes from time
OK so far...

mvallevand Wrote:smart server assumes location is correct and makes subsequent skips base of the same offset.
I don't se why why there will be subsequent skips on the same offset.
Every skip triggered by user or the player itself is a new request as far as I know and the primed value is used once and discarded.


mvallevand Wrote:if the demux reads the file and the time line is off it might do an incremental seeks trying to find a good time
repeat until demux is happy
I see no problem with that, it's pretty much what happens today anyway with crappy time lines.
IMO it can only get better.

mvallevand Wrote:It's better to have a timing.info aware player or a totally dumb player that responds to a smart streamer but not both.

I totally agree. But I don’t think I'll ever be successful in persuading LG, Samsung and every other manufacturer to implement support for timing.info or even a seek by byte function.
I'd love to see a smart streamer capable of snappy and accurate seeking, preferably taking a .ts file transcoding it on the fly to some html5 playable format without lag or CPU-loading. For now that it's just a dream, and I'll stick with things I have some small clue about.

Using my suggested solution the bytes received is normally only seconds off from what the player expected and the playback position if time line is somewhat accurate will be spot on whereas without timeing.info correction the bytes are exactly the requested ones but the playback position may be way off from the requested time, and possibly cause the player to seek again in it's hunt for the playback position the user requested.

My suggested solution is not optimal but it's doable and dead accurate during the tests on my LG TV anyway.

/Fred
reven
Offline

Posting Freak

Posts: 5,782
Threads: 396
Joined: Sep 2004
#39
2013-09-16, 11:11 PM
Fred any chance I can get a hold of your server app to try out on my samsung client? I have two sample recordings from users that just dont seek at all well on samsung tvs, so could test out your solution on those.
fred250
Offline

Senior Member

Posts: 457
Threads: 10
Joined: Jul 2006
#40
2013-09-16, 11:53 PM
reven Wrote:Fred any chance I can get a hold of your server app to try out on my samsung client? I have two sample recordings from users that just dont seek at all well on samsung tvs, so could test out your solution on those.

Hi reven,

The streaming server isn't really in that state at the moment. I just today started to experiment with timing.info. The tests I made was by faking how timing.info is read from the .ts -file and more.
I will let you know when there is something you can try.

I do have a Samsung 2012 bluray. (E is it?) Any suggesting of simlpe tests I can make?
I don't have any developing tools up an running, and really no intenstion to install right now.

/fred
« Next Oldest | Next Newest »

Users browsing this thread: 1 Guest(s)

Pages (11): « Previous 1 2 3 4 5 6 … 11 Next »
Jump to page 


Possibly Related Threads…
Thread Author Replies Views Last Post
  is there a service?method which returns listings for multiple channels? reven 16 6,977 2022-04-11, 04:30 PM
Last Post: mandai
  Extending end time. mvallevand 5 4,083 2018-10-17, 01:21 PM
Last Post: jcole998
  NextPVR "Service" Backend? jcole998 5 4,795 2018-05-24, 01:17 PM
Last Post: jcole998
  Transcoding service mvallevand 27 12,346 2016-08-26, 07:43 PM
Last Post: sub
  Extending recording time mvallevand 2 2,456 2014-09-14, 01:04 AM
Last Post: sub
  Touch and http audio url's mvallevand 0 1,413 2013-11-03, 07:05 PM
Last Post: mvallevand
  <StartTime> and <EndTime> time formats? spinnaker 3 2,481 2013-10-25, 01:48 AM
Last Post: spinnaker
  New web service clarification needed bgowland 6 3,218 2013-10-19, 02:36 AM
Last Post: bgowland
  [REQ] Additional web service API calls reven 26 8,160 2013-08-31, 10:08 PM
Last Post: mvallevand
  getting recurring recordings from service?method API? reven 2 1,942 2013-08-03, 10:50 PM
Last Post: reven

  • View a Printable Version
  • Subscribe to this thread
Forum Jump:

© Designed by D&D, modified by NextPVR - Powered by MyBB

Linear Mode
Threaded Mode