New release of Media Player 6.4

[gabest]

Each renderer has to be assigned to a window (mpc has only one ), or else the filter graph will create one for them when the playback starts (with that active movie something title).

[bond]

hm, i noticed that mpc98me6467 didnt open a new window for the second video stream, tough i think this was more a bug than a feature cause it isnt the case anymore with newer versions...

[JimiK]

Hi gabest,
please excuse me if this sounds dumb, but: does that mean that the embedded VobSub subs are not supposed to occur in the subtitles menue?
Best regards,
JimiK

[gabest]

That menu only lists subtitles opened by the player itself, dvd/ogm/dvobsub has to be queried and controlled through their own filter interfaces with some extra work for each.

[CiTay]

I have a question about VMR7 vs. VMR9. In the thread Zoomplayer VML9? (apparently started by a japanese), a Zoomplayer dev states that VMR7 is/is about to become obsolete. Is there any reason you keep both VMR methods in MPC?

Then, in the output options, the tip for "Use texture surfaces and render video in 2d" reads "Requires a video card that can...". If some of the requirements are not met, will it not play at all or will the image quality degrade? Furthermore, what combination do you recommend for a modern graphics card? I currently use VMR9 (renderless) with the 2D option.

Lastly, i found no way to change the mouse button associations. Reason i looked for them was that i use a doubleclick to make the video go fullscreen, but with the current setting, it always "blips" because it uses left-click for play/pause.

Thanks for a great player BTW.

[gabest]

Originally posted by CiTay
I have a question about VMR7 vs. VMR9. In the thread Zoomplayer VML9? (apparently started by a japanese), a Zoomplayer dev states that VMR7 is/is about to become obsolete. Is there any reason you keep both VMR methods in MPC?

VMR7 is more stable and doesn't suffer from a few annoying bugs still present in directx9 or the video drivers. In windowed mode it can even use overlay much smoother than the old renderer.

Then, in the output options, the tip for "Use texture surfaces and render video in 2d" reads "Requires a video card that can...". If some of the requirements are not met, will it not play at all or will the image quality degrade?

There are two levels of the requirement checkings, vmr9's own and mine in renderless mode. If the first fails the filter won't even create and mpc will fall back to the "default video renderer".

Furthermore, what combination do you recommend for a modern graphics card? I currently use VMR9 (renderless) with the 2D option.

Anything what works for you . texture+2d can be useful when the video is stretched in a smil script (StretchRect doesn't do stretching when the target is an offscreen surface), and with texture+3d I could create nice transition effects in 3d ... if I knew d3d better, hehe.

Lastly, i found no way to change the mouse button associations. ...

options/keys: full screen & "mouse" column -> nothing.

[CiTay]

Originally posted by gabest
options/keys: full screen & "mouse" column -> nothing.

Doh. In all this time, i never paid attention to the horizontal scrollbar.

[CiTay]

By the way, now that i turned off left click for play/pause, can i set it so that the window can be dragged when clicking anywhere in the MPC window and holding the left button? I admit it, i once was a long-time Zoomplayer user.

[Lobuz]

Originally posted by gabest
Lobuz:

From how many did you lower matrixmixer's merit? Didn't it autoload when it was just simply registered without being on the filter overrides' list?

It loaded but after one of vorbis decoders, not after audio switcher. I lowered merit of MatrixMixer from 00800001 to 00800000.

But I've found other strange behaviour. After clearing override list and adding only MatrixMixer and setting merit to 00800000 it didn't show up. Only after adding another one filter to override list (xvid, vorbis...) it started to work. And another strange after adding to that list ffdshow, to open divx3 files was used old Mpeg4 Video Decompresor instead of DivX5.1 decoder. Maybe it's normal behaviour I don't know.

When the ogg splitter looses sync, it begins to search for the "OggS" string, and that can take a while. For avi this is much faster because the file format is indexed and the splitter can just lookup the next valid chunk being after the last readed successfully. Btw, I forgot to reenable that warning dialog of the avi splitter in the latest mpc release. I commented it out for some reason and then it was left like that...

Couldn't it be possible to just stop that searching immediately if I click the seek bar in other place, or press stop or pause. Now it just freezes till it finds the good data. If I want to preview in eMule video and there is just e few chunks o actual good data it's hard to find that places. MPC doasn't show them

> But I've found an avi where the MPC doesn't want to fast jump over incompleted part just slowly "plays" it after "reindexing".

How big is it?

After muxing a small part of video with AC3 and filling a few MB of data with 00 with Hex editor it behaves thesame, so if You try that Yoursel it shoul be simulated. You can simulte like that unfinished ogm or mkv too.

Could it be possible with avi to skip playing that blank places after cliking at the seek bar on it. Now it only jumps over bad places after playing good parts. So to find good place I must try many times on seek bar to find it.

Regards
Lobuz

[Safai]

gabest: non-interleaved avi files are not working in 6471, don't get the beeps like in 6470, black video, no audio.

[LigH]

Non-interleaved AVIs? That means: Complete video at the beginning, complete audio at the end, and the read heads jumping forth and back like a sewing-machine. What is that good for, except hardware stress testing?

[Safai]

Originally posted by LigH
Non-interleaved AVIs? That means: Complete video at the beginning, complete audio at the end,

i think it is the other way, audio in front, video at the back...

and the read heads jumping forth and back like a sewing-machine. What is that good for, except hardware stress testing?

well... i changed them to interleaved after the download is completed... but now i can't really preview them anymore while downloading...

and it used to work... unless gabest "officially" dropped support for them...

[gabest]

CiTay:

This was one of the forgotten things ... finally done

Lobuz:

I can explain why the matrixmixer only loaded after changing its merit to 0x800000... To allow transform filters to load before the audio switcher (it was designed to be the last before the renderer), first I lookup all audio renders and set the switcher's merit to the highest plus a little (like 0x800000.1). This way the auto-loading filters having a higher merit than all audio renderers (that's why they auto-load after all) will be connected after the decoders and before the switcher always (the switcher tries to act as the renderer). This includes matrixmixer too, because its merit is 0x800001 by default. Now, when you change it to 0x800000, there is a chance that it will be enumerated _before_ the audio render having this very same merit, but there is also a chance it will be enumarated _after_ and then the render terminates the stream too eary Now, what is the solution? To make sure the order inside the filter override list is applied, I'm already adding those "plus littles" or fractions to theirs merit, but unfortunately I started counting from zero, so when the first on the list was matrixmixer it ended up with the same merit as the default audio renderer (0x800000+0), but was put after it on the internal sorted list.

> Couldn't it be possible to just stop that searching immediately if I click the seek bar in other place, or press stop or pause. Now it just freezes till it finds the good data.

Yea, a not very good thing it is. Unlike the demuxing loop, seeking must be a blocking command, it should not return without setting the pointers to the right place, that's how it was designed for now. There is also a difference how avi and ogg can be resynced. Because avi has an index I set the supposedly right file position in the seek function and care about its validiy later when it can be broken in the demuxing loop.

> it only jumps over bad places after playing good parts.

It should work like that, ... will be done sometime surely

Safai:

> non-interleaved avi files are not working in 6471, don't get the beeps like in 6470, black video, no audio.

Quoted from myself

Btw, I forgot to reenable that warning dialog of the avi splitter in the latest mpc release. I commented it out for some reason and then it was left like that...

LigH:

Non-interleaved avi means (in my interpretation), that one chunk from one stream doesn't necessarily follows the previous from another in time. My avi splitter can tolerate (by buffering) a distance of 500 chunks, but after that it shows that warnining dialog.

A little explanation: v - video, a - audio

Interleaved:

v:   0 40 80    120 160     200 ...
a: 0         90         180

Not properly interleaved:

v:   0 40 80 120    160 200 240 280     320 360 ...
a: 0             90                 180

After a while this skew between the streams gets so huge, that it becomes impossible to play it by reading the file sequentially.

[Lobuz]

These not interleaved avis are still common. And there are some with audio at the beginning and then video I think. Those are completly black.
But I've found another type of strange avi. It's audio stream is stretched to 95% (showed in AVIInfo)and I'm not sure but it's propably vbr mp3 muxed as cbr. I have only the beginning and end of the file but VDub shows the warning about vbr. It play but after a few seconds starts stuttering. But with default avisplitter it's ok. And I don't have it completed yet. Just my friend noticed me about that problem.

Regards
Lobuz

ps. Still oggDs splitter is noticable faster while seeking when I click on good ogg data in unfinished ogm. It jumps to I frames? Maybe borrowing some code from that splitter would be useful.

[gabest]

> ps. Still oggDs splitter is noticable faster while seeking when I click on good ogg data in unfinished ogm. It jumps to I frames? Maybe borrowing some code from that splitter would be useful.

Have you ever seen its settings? :P One of the few checkboxes there is making it seek onto keyframes only, turn it off and it will be much slower.

[Longinus]

Gabest, I was trying the web interface, and I'm having some problems.. Using the automatic page to send the requests works. But if I try any simple POST querys from PHP, it makes MPC crash completly.

Here is the post query i'm using.

POST / HTTP/1.1
Host: localhost
Content-Length: 13
Content-Type: application/x-www-form-urlencoded
Connection: Close
wmcommand=131

What I'm doing wrong?
Thanks! I can think of a LOT of uses for these interface thing. =D

[chros]

Originally posted by Longinus
Thanks! I can think of a LOT of uses for these interface thing. =D

Which uses do you mean ? I have been thinking about this a lot, but I don't see which practical situations will benefit from this ?
Please, give me some idea !

Thanx

[gabest]

chros: It's the same for me. But I always hoped people like Longinus will come once and tell me what to use it for

Longinus: I can't debug it now, but I believe mpc is expecting the terminating \r\n\r\n at the end of the header. The POST data should only come after that in the body, all in one line separated with & characters, just like a GET argument list.

[uneasy]

Gabest, have you checked this DropTarget thingy yet? Is it possible to implement it?

[gabest]

I did, but I got lost in that maze... I would be really pleased if someone could setup the registry the way it works for multiple selections (passed as one via the command line) and showed me how he did it.