Request: TCPMP
 

Hi,

For those of you who don't know what it is: TCPMP is a open-source media player, originally designed fow pocket pc's (windows mobile).
For audio playback, besides the regular functions, this player has a preamplifier, and an equalizer.
For video playback, it has the option to lower the video quality, resulting in a smooth playback for high-bitrate videos.

This project has been closed for quite some time (1-2 years). Being open-source, the source is here: http://picard.exceed.hu/tcpmp/test/t....72RC1.tar.bz2

Older versions can be seen on the host site:
http://picard.exceed.hu/tcpmp/

Knowing that this code is compilable to arm (pocket pc), and that it is open-source, is a port possible?

Thank you

Ilia

Re: Request: TCPMP
 

Being ARM really doesn't mean very much as far as code compatibility. Especially with things like media playback. Really, there'd be a lot less work involved in putting together a front-end for mplayer that does what you want or fixing up one of the many good media playback applications already available to do what you want.

Re: Request: TCPMP
 

maybe.

the only feature i want is the capacity to lower the quality of a video, making high-bitrate videos display perfectly without any conversion.
Quite honestly, I doubt that there exists a MPlayer client capable of doing so. This is why I asked for the compilation of a player that already has has this feature.

Re: Request: TCPMP
 

I know that mplayer has the ability to skip frames, which will result in faster playback (automatically even, I believe). Also, of course, it can scale a high resolution video down on the fly. I do believe in some way mplayer can do the things you describe, the only problem is that it probably takes CPU time to do so just like it does to convert a high quality movie down to a lower quality on any PC. I'm not sure how this TCPMP software would solve this problem, unless it takes advantage of some ARM hardware ability that helps the process.

As far as porting TCPMP, the challenge there would be that it's made for Windows. Which of course makes the port work a lot more difficult than most other apps ported to Maemo. :(

In the meantime, check out an mplayer man page and see if the functionality exists in there. Mplayer does a LOT.

Re: Request: TCPMP
 

Adding whatever graceful degradation this has to mplayer would quite possibly be easier than porting the whole thing, really.

Re: Request: TCPMP
 

No, not "maybe". Fact. :)

MPlayer is perfectly capable of doing this. This will give you half-resolution decoding:

No, but the point is is that porting a whole new client is pointless, as it would be a trivial feature to add to one of the many existing mplayer front-ends. Especially when porting that client will likely involve more work to port than starting from scratch and leverages exactly none of the thousands of hours already put into optimizing media playback on the NITs with mplayer. . . .

Really, just make it easy on yourself and play your videos from xterm with the command I gave you above.

The whole idea between lowres decoding is to reduce the CPU impact by decoding less than the full frame. ;) What you're thinking of is called reencoding, but all we need to do is decode a portion of the frame, not decode the whole frame then reencode it at a lower resolution.

Actually, the fact that it's written for Windows is probably the easiest hurdle to jump. The real challenge is that it's designed for very different hardware and doesn't have any optimizations for the NITs.

Re: Request: TCPMP
 

It was also writen for palm os so you should be able to use it with the garnet vm

Re: Request: TCPMP
 

I was just sitting on a plane earlier today wishing that TCPMP was on the NIT. It had a really great feature that was an audio boost- that could help me hear the dialog over the jet engines.

On a lifedrive which clocked at no more 300mhz playback was smooth on xvids that I download from the news groups. Even with my NIT running max (390ish) it still skipped here and there.

TCPMP went from opensource to pay about a year ago - and had most of the codecs..

I'll second the request... nothing as flashy as canola - but it worked well!

Re: Request: TCPMP
 

Unfortunately, you can't, because gvm doesn't do ARM code.

Re: Request: TCPMP
 

I hope you're talking about your personal jet engine. I'd hate to sit next to someone blasting their PMP\PDA\RIM\WTF next to me just so they can listen to the dialogues.

PS: Earphones.

Re: Request: TCPMP
 

TCPMP was great but I'm not sure it's needed. mplayer plays a hell of a lot of files and it's more optimised for maemo.

Also. bad playback on N8*0 is gonna happen with any player.

Re: Request: TCPMP
 

Sorry - but it was Delta not my own personal jet - but I do have a pilot.

O'Yea. YES headphones...

;)

Re: Request: TCPMP
 

From what I read TCPMP was able to play large (high bitrate) files quite well. Whether this was down to the graceful degradation, video scaling, optimised codecs for ARM or support for video acceleration hardware, I don't know.

What might be useful is for people to list what features of TCPMP they are after, then people could have a go at adding them to one of the mplayer frontends.

Likewise it would be useful to see what sorts of videos (and I'm thinking mainly the large xvid/divx that mplayer has some troubles with) TCPMP was able to play without troubles (plus the spec of the machine you're playing on) vs. the performance on an N8x0/880 and we'll see if there is actually a significant difference.

P.S. I lack a machine that TCPMP will run on, so I can't do this myself, sorry.

Re: Request: TCPMP
 

TCPMP plays how large and how high bitrate files on what screen?

Re: Request: TCPMP
 

Please! Not another media player for the N8x0. I got media players coming out me A***e! For the love of Jebus no!

Proper integrated PIM and MS-compatible Word Processing, that's what we need.

mumble..mumble....'nother media player....mumble..mumble

Re: Request: TCPMP
 

I don't think it's the wanting another media player, but rather the features in said player. That's a perfectly reasonable thing to ask for, assuming the features are actually worth having.

Re: Request: TCPMP
 

just for the record, I have a [semi-broken] acer n311.
400mhz cpu, 64mb ram, 640x480 resolution. close enough to my n800 (OS2008).
I made a simple test: ran the same file on both computers:
1) on "normal" video quality, the n800 is *a bit* better; video extremely laggy on both.
2) when reducing the quality to "low" on the TCPMP, my acer plays the video smoothly (with a bit lower quality, but more than acceptable on the VGA screen). I haven't yet tried the command on my n800.

I will post the comparison on youtube for you to see.

Re: Request: TCPMP
 

Is it bad enough to tell the difference with youtube? :eek:

Re: Request: TCPMP
 

Not sure but it plays xvid fine on a pocket pc I once had.

Re: Request: TCPMP
 

And what about an equalizer? How is it possible that none of the available media players does not provide an equalizer?

Re: Request: TCPMP
 

How about upgrading the functionality of the real player that we already have? I just downloaded the latest version (11?) on my xp; it even playes flv files.

Re: Request: TCPMP
 

XMMS has a beautiful equalizer and happens to work on N8x0 too.

bun

Re: Request: TCPMP
 

That's great. I don't have any windows mobile PDA myself, so sharing your impressions about the performance and features of TCPMP is very much welcome.

It is interesting, so we might already have more or less the same proformance in mplayer/ffmpeg. This makes the decision where to put efforts quite obvious: it is better to continue improving actively developed project (mplayer/ffmpeg) than a dead one (TCPMP) :)

By the way, does TCPMP have any kind of built-in benchmarking? For example, you can easily get information about the number of frames dropped by mplayer. If the same statistics can be obtained for TCPMP, that would make the comparison more "scientific" ;)

Currently 'lowres' decoding mode just started to work. It does not have all the needed optimizations yet: http://www.internettablettalk.com/fo...4&postcount=27

If people are more or less satisfied with the image quality when using this mode, I may invest more efforts into making it really fast.

Good. You can also try some other videos to see if the results are consistent.

Re: Request: TCPMP
 

update on the comparison:

media (info from TCPMP):
duration: 2:59.879s
filesize: 53500kb
video: MPEG-4
codec: CoreASP
video size: 640x480
framerate:25.00
audio: MP3
format: 48000hz Stereo
audio bitrate: 128kb/s

Media info (From MPlayer):
Video: XVID, 640x480 24bpp 25.000 fps 2298.3kbps (280.6 kbyte/s)
clip info: software VirtualDubMod 1.5.4.1
video codec: ffodivx vfm: FFmpeg MPEG-4
audio: 48000hz, 2ch, s16le, 128.0 kbit/8.33%

stats from TCPMP, played in fullscreen, from SD card "normal quality" (no modifications):
frames played:4519
frames dropped:3697
displayed FPS:4.547
comments: low FPS - looks like 2-4 FPS; good quality - all details present; no audio lag; impossible to watch a movie due to low FPS.

stats from TCPMP, played in fullscreen, from SD card "low quality":
frames played: 4539
frames dropped: 1095
displayed FPS: 18.969
comment: perfect FPS - no lag noticed at all. acceptable quality - loss of minor details; still good to watch a movie

Stats from Mplayer, played in fullscreen, original quality:
notes: a) got "your system is too SLOW to play this!" [duh... i know that already]
b) in the middle of the playback (about 1:30), got a complete lag for ~10s. after this, i obtained 3 frames in the following 30s
c) audio - OK (with the exception of the audio lag) video - inexistent.
d) got [mpeg4 @ 0x46ade4]marker does not match f_code MANY times.
stats (if someone knows how to interpret!):
A: 178.6 V: 160.6 A-V: 18.079 ct: 15.915 4015/4015 120% 0% 13.7%

Stats from Mplayer, played in fullscreen, with -lavdopts lowres=1
comments: choppy video (like original quality on n311); the framerate actually seems worse.
comment2: quote:
MPlayer interrupted by signal 11 in module: decode_video
- MPlayer crashed by bad usage of CPU/FPU/RAM
Recompile MPlayer with --enable-debug and make 'gdb' backtrace and disassembly. Details in DOCS/HTML/en/bugreports.html#bugreports_crash.
- MPlayer crasher. This shouldn't happen.
It can be a bug in the MPlayer code _or_ in your drivers _or_ in your gcc version.

I have MPlayer 1.0rc1-maemo.27.n8x0

Stats from Mplayer, played in fullscreen, with -lavdopts lowres=2

comments: no lag at all; great FPS; worse resolution than the n311 with "low resolution", but more than acceptable playback.
crashed in the middle (about 1:00) got the same quote as before. The crash occurred at different sections of the clip.
stats:
A: 65.8 V: 68.3 A-V: 0.151 ct: 0.032 1709/1709 72% 1% 11.0% 3

bottom line:
although the screen is larger on the n800 (4.1" vs 3.7"), the resolution of the video is still the same since the borders are cut on the n800 (see video resolution)
on the original quality, the n311 works better - better framerate; no audio lag...
is there another command for optimizing MPlayer?

I rethought about the idea of youtube - you will not be able to see the difference in the quality. I will rather post the video on a host site and give you the link.

system comparison:
n311:
400mhz
Windows Mobile 5
64mb RAM; 53 user-available
video played from SD card (SDIO, 1gb)

n800:
400mhz
OS2008
128mb RAM, 128MB swap just in case
video played from internal SD: 16gb, class6.

in both cases, TCPMP and terminal+MPlayer were the only programs open.

Re: Request: TCPMP
 

If you are running mplayer from xterm, make sure to provide it with '-quiet' (otherwise output to console eats too much of the resources) and '-benchmark' options. Also if you run it as root, it may have problems getting configuration options, so you may want to add '-framedrop', '-noslices' and maybe some other options. Make sure that mplayer is using 'omapfb' driver for video output and 'sdl' for audio (with libmad decoder). Thanks.

edit: Anyway, better try to play this video in the 'standard' mplayer GUI till it finishes itself. After the playback finishes (I hope it will not crash :) ), press 'Report' button to get the screen with statistics. The information about the number of dropped frames will be at the very bottom. The default configuration should be theoretically best tuned.

Re: Request: TCPMP
 

could you give me the entire list?

I was running as user (will try root now). It was using omapfb and sdl (the problem isn't there)

# mplayer -quiet - benchmark -noslices
process 11110: D-Bus library appears to be incorrectly set up; failed to read machine uuid: Failed to open "/var/lib/dbus/machine-id": No such file or directory
D-Bus not built in with -rdynamic so unable to print a backtrace

Re: Request: TCPMP
 

Run "dbus-uuidgen --ensure=/var/lib/dbus/machine-id" as root ;)

Re: Request: TCPMP
 

$ mplayer -quiet - benchmark -noslices -framedrop
got a record FPS of about 0.75!!!

#dbus-uuidgen --ensure=/var/lib/dbus/machine-id

# mplayer -quiet - benchmark -noslices -framedrop
now, even the audio lags! and the framerate dropped down (to ~0.3)

other ideas?

Re: Request: TCPMP
 

What is 0.75? Can you provide a complete statistics? Also try running the same benchmark with additional '-lavdopts lowres=1' option. Running mplayer as user should be ok, you don't need to run it as root. Additional option you may want to try is '-hardframedrop' (but you may get artefacts on screen).

Sharing this video somewhere would be the best option. At least we know how TCPMP handles it on a CPU with a comparable clock frequency.

Re: Request: TCPMP
 

$ mplayer -quiet - benchmark -noslices -framedrop
I don't get any system stats due to the -quiet!

$ mplayer -quiet - benchmark -noslices -hardframedrop
get about 1fps video. no stats available yet

p.s. plz tell me how to interpret the stats

Re: Request: TCPMP
 

Hmmm... my 0.02:

TCPMP was an amazing program that really was the only thing besides EasyCalc that made me hold on to my Palm Zire72 as long as I did. I could download almost any movie/media file and put it on my media card and TCPMP would play it without (in most cases) breaking a sweat. Even high-bitrate/fast action sequences rendered beautifully. I never had to worry about optomizations or how large I coded my videos.

Despite all this, we definitely don't need to port over the entire application. But adding some of these features (equalizer, preamp, and automatic frame dropping/resizing) would make mplayer so much more amazing :) re-encoding videos just so I can watch them on my tablet is a pain in the butt.

@Serge - TCPMP has a really nice built in benchmarking tool :) On the Palm it's available from the "File" menu. It'll play back the current media file as fast as possible on your device and display the output :)

PS: I'm still looking for a good replacement for EasyCalc

Re: Request: TCPMP
 

SUCCESS!!!
was finally able to obtain a similar playback as for the n311, low resolution mode.

$ mplayer -quiet - benchmark -noslices -hardframedrop -lavdopts lowres=1
the quality of the video is similar; the framerate is perfect: no lag, no stall (continuous video playback, you can't see individual frames.

happens from time to time; does not disturb much

Re: Request: TCPMP
 

second that. I considered my n311 dead. just resuscitated it yesterday. still, the external speaker doesn't work, and if I use earphones, I must be really careful not to move them too much - this actually freezes the device :(

if you find one - tell me!
other calculators I miss: spacetime (great graphing tools, as well as the capacity to derivate, integrate, take limits and matrices)

Re: Request: TCPMP
 

You should get the stats when video playback finishes.

Statistics report should look like this:
http://www.internettablettalk.com/fo...&postcount=429

Pay attention to the lines:
How to interpret what stats? Looks like you are doing something wrong (probably stopping playback too early).

Re: Request: TCPMP
 

strange: now , when I run
$ mplayer -quiet - benchmark -noslices -hardframedrop -lavdopts lowres=1
, the program crashes with
MPlayer interrupted by signal 11 in module: decode_video
- MPlayer crashed by bad usage of CPU/FPU/RAM
Recompile MPlayer with --enable-debug and make 'gdb' backtrace and disassembly. Details in DOCS/HTML/en/bugreports.html#bugreports_crash.
- MPlayer crasher. This shouldn't happen.
It can be a bug in the MPlayer code _or_ in your drivers _or_ in your gcc version.

Re: Request: TCPMP
 

That's one more reason to upload this test video files somewhere, so that I could take a look at it ;)

Re: Request: TCPMP
 

tried another [high-bitrate] file.
link: http://adrive.com/public/740d3197f11...d8a7dd862.html
stats:
$ mplayer -quiet - benchmark -noslices -hardframedrop -lavdopts lowres=1
BENCHMARKs:VC: 167.870s VO: 7.662s A: 2.276s Sys: 22.319s = 200.127s
BENCHMARK%: VC: 83.8817% VO: 3.8286% A: 1.1374% sys:11.1522% =100.0000%
BENCHMARKn: disp: 3818 (19.08 fps) drop: 1138 (22%) total: 4956 (24.76fps)

$ mplayer -quiet – benchmark
BENCHMARKs:VC: 223.302s VO: 0.522s A: 2.487s Sys: 1.581s = 227.892s
BENCHMARK%: VC: 97.9860% VO: 0.2292% A: 1.0912% sys:0.6937% =100.0000%
BENCHMARKn: disp: 34 (0.15 fps) drop: 4305 (99%) total: 4339 (19.04fps)

TCPMP low resolution:
Average Speed: 69.32%
Video Frames: 3695
Audio Samples: 7347072
Amount of data: 48148

Bench. Time: 3:33.226
Bench. Frame rate: 17.33
Bench. Sample rate: 34457
Bench. Data rate: 1.8Mbit/s

Original time: 2.27.779
Original framerate: 25.00
Original sample rate: 48000
Original data rate: 2.6 Mbit/s

TCPMP normal resolution
Average Speed: 38.65%
Video Frames: 3695
Audio Samples: 7268246
Amount of data: 48148

Bench. Time: 6:22.408
Bench. Frame rate: 9.66
Bench. Sample rate: 19006
Bench. Data rate: 1.0Mbit/s

Original time: 2.27.779
Original framerate: 25.00
Original sample rate: 48000
Original data rate: 2.6 Mbit/s



NOTE: TCPMP displays EVERY SIMGLE frame, yet this increases the time of the video playback.
MPlayer kills the frames, yet has the playback time of the original video.

Re: Request: TCPMP
 

http://adrive.com/public/27488d8bcb1...7fc0fcd8f.html
this is the file I have been performing tests on previously (the one that crashed)

Re: Request: TCPMP
 

Can you provide MD5 for both these files? Looks one file is slightly corrupted in the middle (the one that crashes mplayer) and the other one is completely broken and does not even show video when played.

Re: Request: TCPMP
 

sorry, I don't know what MD5 is...

I'll try uploading the second one (Les rois du monde) again... would make no sence that I could play it and you couldn't