Mailing List Archive

On 12/20/20 12:10 AM, Tom Harris wrote:
> My primary frontend is the shield mythfrontend app (Thank you Peter
> Bennett).
>
> I'm very happy with it, but one thing that could be optimized a bit is
> startup time.  In my previous x86 frontends I built with only the
> features I used - which for me is just viewing recordings.  I'm not
> really sure how much difference that makes, so I'm looking for input
> on whether it's worth the effort to set up a cross-compile environment
> to build the frontend.  Does a "light" build make a meaningful
> difference in startup time or other resource utilization?
>
Mythfrontend startup time is slow on any system, both linux and android.
I don't know of any way to speed it up.
> Also, I do know the leanfront app is effectively a lightweight Myth
> Frontend, and I use that sometimes.  But, I still end up using
> mythfrontend much of the time for the incredibly fast skip-ahead
> performance skipping over commercials.
>
>
leanfront does not use the seek table, so that is part of the reason it
is slower. However, some UK recordings seem to be much slower than US
recordings at skipping. My US recordings skip in a second or two, but
some UK samples take up to 15 seconds to skip. What recordings are you
using and how slow is your skip time? I may look into why the UK
recordings have such a slow skip time if there are affected users.

Peter


_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 20/12/2020 14:37, Peter Bennett wrote:
>
> On 12/20/20 12:10 AM, Tom Harris wrote:
>> My primary frontend is the shield mythfrontend app (Thank you Peter
>> Bennett).
>>
>> I'm very happy with it, but one thing that could be optimized a bit is
>> startup time.  In my previous x86 frontends I built with only the
>> features I used - which for me is just viewing recordings.  I'm not
>> really sure how much difference that makes, so I'm looking for input
>> on whether it's worth the effort to set up a cross-compile environment
>> to build the frontend.  Does a "light" build make a meaningful
>> difference in startup time or other resource utilization?
>>
> Mythfrontend startup time is slow on any system, both linux and android.
> I don't know of any way to speed it up.
>> Also, I do know the leanfront app is effectively a lightweight Myth
>> Frontend, and I use that sometimes.  But, I still end up using
>> mythfrontend much of the time for the incredibly fast skip-ahead
>> performance skipping over commercials.
>>
>>
> leanfront does not use the seek table, so that is part of the reason it
> is slower. However, some UK recordings seem to be much slower than US
> recordings at skipping. My US recordings skip in a second or two, but
> some UK samples take up to 15 seconds to skip. What recordings are you
> using and how slow is your skip time? I may look into why the UK
> recordings have such a slow skip time if there are affected users.
>
> Peter

I was not sure what was meant by the fast skip-ahead. The standard
frontend will recognise a cut in a defined cutlist and that skip is then
virtually undetectable during playback. ttbomk leanfront (or upnp)
ignores a cutlist, so skipping needs interactive real-time button pushing.

I now have scripts that work rather like the old 'lossless
mythtranscode' and will cut most of my UK DVB-T/T2 recordings to a
format that plays well in leanfront, as well as the other players I have
tried; all the button-pushing is done in advance, with a keyboard and
the cutlist editor of the standard x86_64 mythfrontend.

Cutting mpeg2 recordings is still quicker, easier and gives neater
transitions than with h264, but there is no re-encoding for either
format and the processes are quick.

Some of the problems that I have reported before in leanfront playback
seem to have been caused by my use of mplex (from mjpegtools) to do the
remultiplexing after ProjectX had selected, cut and concatenated the
video and audio streams; that was carried over from MythArchive, where
it worked well in DVD creation. Remultiplexing to mpegts with
mythffmpeg almost always gives files that work with leanfront, and those
files, unlike MKVs, are still myth-compatible.

John P



_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Sun, Dec 20, 2020 at 6:37 AM Peter Bennett <pb.mythtv@gmail.com> wrote:

>
> On 12/20/20 12:10 AM, Tom Harris wrote:
> > My primary frontend is the shield mythfrontend app (Thank you Peter
> > Bennett).
> >
> > I'm very happy with it, but one thing that could be optimized a bit is
> > startup time. In my previous x86 frontends I built with only the
> > features I used - which for me is just viewing recordings. I'm not
> > really sure how much difference that makes, so I'm looking for input
> > on whether it's worth the effort to set up a cross-compile environment
> > to build the frontend. Does a "light" build make a meaningful
> > difference in startup time or other resource utilization?
> >
> Mythfrontend startup time is slow on any system, both linux and android.
> I don't know of any way to speed it up.
> > Also, I do know the leanfront app is effectively a lightweight Myth
> > Frontend, and I use that sometimes. But, I still end up using
> > mythfrontend much of the time for the incredibly fast skip-ahead
> > performance skipping over commercials.
> >
> >
> leanfront does not use the seek table, so that is part of the reason it
> is slower. However, some UK recordings seem to be much slower than US
> recordings at skipping. My US recordings skip in a second or two, but
> some UK samples take up to 15 seconds to skip. What recordings are you
> using and how slow is your skip time? I may look into why the UK
> recordings have such a slow skip time if there are affected users.


I'm in the US, and the performance is good. A skip ahead on my Nvidia
Shield is about 1 second. It's better than most other media player players
I've tried.

So, it's not that leanfront is slow, it's just that mythfrontend is
incredibly fast. A skip ahead just jumps 30s ahead with almost no
perceptible interruption. When watching something like an NFL football
game I'm constantly skipping ahead between plays and past commercials and
the quick skips are really nice.

On 20/12/2020 23:45, Tom Harris wrote:
>
>
> On Sun, Dec 20, 2020 at 6:37 AM Peter Bennett <pb.mythtv@gmail.com
> <mailto:pb.mythtv@gmail.com>> wrote:
>
>
> On 12/20/20 12:10 AM, Tom Harris wrote:
> > My primary frontend is the shield mythfrontend app (Thank you Peter
> > Bennett).
> >
> > I'm very happy with it, but one thing that could be optimized a
> bit is
> > startup time.  In my previous x86 frontends I built with only the
> > features I used - which for me is just viewing recordings.  I'm not
> > really sure how much difference that makes, so I'm looking for input
> > on whether it's worth the effort to set up a cross-compile
> environment
> > to build the frontend.  Does a "light" build make a meaningful
> > difference in startup time or other resource utilization?
> >
> Mythfrontend startup time is slow on any system, both linux and
> android.
> I don't know of any way to speed it up.
> > Also, I do know the leanfront app is effectively a lightweight Myth
> > Frontend, and I use that sometimes.  But, I still end up using
> > mythfrontend much of the time for the incredibly fast skip-ahead
> > performance skipping over commercials.
> >
> >
> leanfront does not use the seek table, so that is part of the reason it
> is slower. However, some UK recordings seem to be much slower than US
> recordings at skipping. My US recordings skip in a second or two, but
> some UK samples take up to 15 seconds to skip. What recordings are you
> using and how slow is your skip time? I may look into why the UK
> recordings have such a slow skip time if there are affected users.
>
>
>  I'm in the US, and the performance is good.  A skip ahead on my Nvidia
> Shield is about 1 second.  It's better than most other media player
> players I've tried.
>
> So, it's not that leanfront is slow, it's just that mythfrontend is
> incredibly fast.  A skip ahead just jumps 30s ahead with almost no
> perceptible interruption.  When watching something like an NFL football
> game I'm constantly skipping ahead between plays and past commercials
> and the quick skips are really nice.

I have tried jumping around in leanfront in several of my recordings
including UK DVB-T2 HD. Short (ie single click) skips are typically
around 1 second, longer ones may bring up the buffering rotating arrow.

Perhaps initially more disconcerting is the very brief recalibration (?)
pause that follows. FireTVstick 4k, good wifi connection,

mythffmpeg -fflags +genpts -i infile -vcodec copy -acodec copy -f mpegts
outfile.ts

followed by a mythcommflag --rebuild, which iiuc leanfront won't use.

John P
_______________________________________________
mythtv-users mailing list
mythtv-users@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-users
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org