MickJT-MPlayer

Not sure if its a bug, but in amigaos_stuff.c there is:
..
struct Library *GadToolsBase = NULL;
struct AslIFace *IGadTools = NULL;
..

Is "struct AslIFace" correct?!?! Or should it be "struct GadToolsIFace"

I've not looked at or touched that part of the code before, but yeah it seems wrong. I'll take a closer look later on.

Latest OS4depot version running great here on X1000.
I have a very taxing fullHD x.264 MP4 of Ex Machina that no previous player could keep A/V sync with.

Micks latest version can. Great work, thanks.

@ddni

If "lavdopts=skiploopfilter=all:skipframe=noref" is activated in the conf file, you'll be able to play it of course but quality will be ugly.

Except if you have very big def videos, this option should be commented otherwise you won't get optimal quality (and you'll think MPlayer is faster than it really is).

@MickJT

Thank you for this piece of software!

@K-L

"noref" is actually a typo. I corrected it in the config in 1.0.1, it's supposed to be "nonref" and skips more frames.

@ddni

It's quite possible that the reason it's keeping up is because of skiploopfilter=all which degrades the quality of the video a bit. Thanks for your compliment though :)

@benny

Thanks.

@K-L

ok thanks, I'll check that.

Does Mplayer GUI ignore the config file?

@jabirulo

Yes its wrong but GadTools is not used, it just some crap from some where, I guess, can be deleted.

@ddni

yes, don't think mplayer GUI knows about that file, mplayer overrides any options you might have in the GUI.

I am able to play 720p music videos on my X5k

Generally it is ok but sometimes I hear some cracks in sound replay (from NGFS partition).

I will check if there is something in serial output ?

This version of Mplayer works really fast with those "fast" settings on an X1000.

@SinanSam460

Serial output, debug kernel ? I think that slow things down, the serial port really slow.

@MickJT

The sound gets distorted with some videos. That doesn't happen with other Mplayer's.

@TSK

Any example videos I can try? Do you use playlists at all?

I know of 2 types of distortion issues. One is when switching between audio formats from FLOAT_BE to something else if you're using a playlist or specifying 2 or more files on the command line. The other is just when the audio plays too loud. If it's mp3 audio, try -ac ffmp3float.

@MickJT

No, I don't use playlists. The audio format is AAC in those videos.

Here's a video in YT.

I couldn't find keys to control playback volume, I don't have numpad on the keyboard if that matters.

@TSK

You can use the 9 and 0 keys to change the volume level, if you have input.conf in the conf directory.

I'm assuming you mean the audio is clipping, and that's what you mean by distorted.

I played the 640x360 MP4 video and hear audio clipping. It seems like it happens only when using FLOAT_BE output in AHI. Changing the audio levels in AHI or MPlayer itself doesn't have any effect. I will need to investigate. LiveForIt-MPlayer doesn't natively play this output to AHI, but converts it to a different format while playing back. The ahi_dev2 I'm using is from Kjetil but it's still experimental.

Edit: Can you try with -softvol -volume 50 (and other volume levels with softvol enabled) and tell me if that "fixes" the issue?

@MickJT

That -softvol -volume 50 helped.

I didn't know you can play floats through AHI ?! I'm using swresample to convert audio in my own video players.

@TSK

No i wrote a code optimizer, and conversion rotines

https://github.com/khval/mplayer-amiga ... /master/src/amigaos/asm.c

https://github.com/khval/mplayer-amiga ... /master/src/amigaos/asm.h


I think its scale factor, I need have look at code again I think.

on line 61 it inserts muls, using register 1.
question now is where is register 1 one set.
my guess its done in start of the code.

#define dRT (RT == 1 ? 0 : RT),(RT == 1 ? 0 : RT)

dRT replaces register 1 with register 0,
PowerPC normal takes 3 arguments,

normally format is
INSTRUCTION DST,INDEX,SRC

or

INSTRUCTION REG1,REG2,REG3

maybe the last part of dRT is wrong, maybe it should have been a index. but we need to lookup etch individual powerpc instruction to see, sadly I did not make comment in .h file, anyway as long as loaded and saved in same location it should not be a problem.


#define m_muls(a,b,c) ( 0xEC000032 | (a<<21) | (b<<16) | (c<<6) )


Look at this macro I wonder if Multiplications might be hardcoded into macro it self.

@LiveForIt

Thanks for the explanation. That explains the even worse distortion if I disable the JIT function entirely. A hack to decrease the volume in software would work, but it seems a little crude to do that. I'm clueless about how the AHI & JIT code works, so I'm not the right person to modify that code directly.

It's been pointed out to me that 1.0.1 is not an Amiga version number. The whole reason I called the first release "1.0" was to have a compliant version string. It slipped my mind. I'll retroactively call 1.0.1, 1.1, and the next version can be 1.2. Sorry about that everyone :)