January 2011 - MPlayer-dev-eng - lists.mplayerhq.hu

Best way to give fps to vo
by Dan Oscarsson 31 Jan '11

31 Jan '11

Hi I have added code to enable switching the screen to a frame rate close to the movie frame rate. Just like -vm but only changing the frame rate. This gives best results on many LCD TV sets. For this I need to get the fps of the movie to the vo layer. One way would be to change the "config" call definition to include fps, an other to use global variables. Currently I have added a vo_dfps global variable that is set in libmpcodecs/vd.c before config of vo is called. I see that vo_fps is set from mplayer.c and one comments indicates that it is intended to be used by vo init. I have compared my vo_dfps with vo_fps and they are mostly the same. But not in my .wmv samples which sets vo-fps to 1000 hz before vo init while my vo_dfps set in vd.c uses (more) correct fps value. So which way should I use to send fps to the vo layer? - change the "config" call definition to include fps. - use my vo_dfps value set in vd.c - change so vd.c updates vo_fps and use that. or some other way. Dan

1 0

[PATCH] add missing include headers
by Reinhard Tartler 31 Jan '11

31 Jan '11

with r32834, some memalign calls were replaced by av_malloc. This causes compiler warnings like: Function `av_malloc' implicitly converted to pointer at libmpcodecs/vf_perspective.c:111 Function `av_malloc' implicitly converted to pointer at libmpcodecs/vf_pp7.c:353 Function `av_malloc' implicitly converted to pointer at libmpcodecs/vf_perspective.c:111 Function `av_malloc' implicitly converted to pointer at libmpcodecs/vf_pp7.c:353 Function `av_malloc' implicitly converted to pointer at libmpcodecs/vf_perspective.c:111 Function `av_malloc' implicitly converted to pointer at libmpcodecs/vf_pp7.c:353 Function `av_malloc' implicitly converted to pointer at libmpcodecs/vf_perspective.c:111 Function `av_malloc' implicitly converted to pointer at libmpcodecs/vf_pp7.c:353 Moreover, the launchpad daily builders let this build fail with this error: ,---- | Our automated build log filter detected the problem(s) above that will | likely cause your package to segfault on architectures where the size of | a pointer is greater than the size of an integer, such as ia64 and amd64. | | This is often due to a missing function prototype definition. | | Since use of implicitly converted pointers is always fatal to the application | on ia64, they are errors. Please correct them for your next upload. | | More information can be found at: | http://wiki.debian.org/ImplicitPointerConversions `---- The following patch should fix the problem. OK to apply, or should I better include "libavutil/mem.h", which actually provides to prototype of av_malloc()? Index: libmpcodecs/vf_pp.c =================================================================== --- libmpcodecs/vf_pp.c (Revision 32834) +++ libmpcodecs/vf_pp.c (Arbeitskopie) @@ -34,6 +34,7 @@ #include "img_format.h" #include "mp_image.h" #include "vf.h" +#include "libavutil/common.h" #include "libavutil/internal.h" #include "libpostproc/postprocess.h" Index: libmpcodecs/vf_perspective.c =================================================================== --- libmpcodecs/vf_perspective.c (Revision 32834) +++ libmpcodecs/vf_perspective.c (Arbeitskopie) @@ -32,6 +32,8 @@ #include <malloc.h> #endif +#include "libavutil/common.h" + #include "img_format.h" #include "mp_image.h" #include "vf.h" -- Gruesse/greetings, Reinhard Tartler, KeyID 945348A4

3 3

CRC32 function needed
by Ingo Brückl 30 Jan '11

30 Jan '11

For improvements of the scrolling function (dynamic labels) I need to check whether the string to scroll has changed. In order not to waste memory and storing it for comparison, it seems simplest to calculate a CRC32 checksum and store this instead. I've found a CalcCRC32() function in libmpdemux/muxer_mpeg.c but am not sure whether this will always be compiled regardless of the user's configuration (it seems that this only goes in mencoder). Further on, this function calculates its table every time it is called. If called often, it might be better this table would be precalculated and stored. Some time ago I wrote a CRC32 function without table usage that I could use instead if CalcCRC32() isn't the one to go with but I don't want to needlessly increase MPlayer's size if not necessary either. Ingo

2 2

Remove needless fontid
by Ingo Brückl 29 Jan '11

29 Jan '11

Does anyone know about (the history of?) fontid? Actually, fontid isn't used anywhere (at least no fontid declared in the skin file; an internal id is assigned each font by its name, but this is totally internal and doesn't affect the user). The documentation is wrong and very confusing for skin designers. Ingo

3 2

Re: [MPlayer-dev-eng] [FFmpeg-cvslog] Remove useage of memalign() from libmpcodecs, not all platforms have memalign().
by Reimar Döffinger 29 Jan '11

29 Jan '11

On Fri, Jan 28, 2011 at 09:57:03PM +0100, Michael Niedermayer wrote: > ffmpeg | branch: master | Michael Niedermayer <michaelni(a)gmx.at> | Fri Jan 28 21:54:16 2011 +0100| [55fa97e215c40a1f7cdd2ad3ca4c098a3caa4c28] | committer: Michael Niedermayer > > Remove useage of memalign() from libmpcodecs, not all platforms have memalign(). If it helps you, I think it would be ok to do this chage to MPlayer's libmpcodecs as well.

1 0

MPlayer in Browser? How to get informed about security holes?
by Manuel Reimer 28 Jan '11

28 Jan '11

Hello, I'm using MPlayer embedded in my browser to view embedded video on websites. So the important question is: Do you think MPlayer is as secure as a plugin in browser requires it to be? Is there a mailing list, where I get informed about known security holes, so I can update my MPlayer installation as soon as possible? For me it seems like the announce mailing list isn't used any more? Thanks in advance Yours Manuel

2 1

Fractional part of time stamp in OSD
by Christian 28 Jan '11

28 Jan '11

Hi, currently, the OSD shows only the integral part of the time stamp. Attached patch allows to optionally also show the fractional part of the time stamp in the OSD. Kind regards, Christian P.S.: This is useful for example when trying to create EDLs. There, when trying to place a cut-mark just before a scene change, it typically occurs that one notices the scene change just after it had occurred (i.e.: one frame too late). Hence, one has to rewind the stream and go to the frame just before the scene-change. Doing so requires to constantly have an eye on the terminal (for the subsecond time stamp) and one on the movie itself (to assert the scene change did not yet occur). Allowing to show the fractional part of the time stamp in the OSD solves this issue. P.P.S.: The proposed changes of the same approach in 2007 [1] have been included in the attached patch. Another previous attempt in this direction has been ignored in 2008 [2]. [1] http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/2007-April/051032.html [2] http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/2008-April/056972.html

1 0

Missing reference picture
by Swapnil Jain 28 Jan '11

28 Jan '11

Hi, When i play a FullHD H.264 multicast stream (udp://235.1.1.1), i see some frame drops and get the following errors, is there any way i can avoid this. [h264_vdpau @ 0x8a8cc20]mmco: unref short failure V4L is 1 MPEG CMD use PCI! [h264_vdpau @ 0x8a8cc20]mmco: unref short failure [h264_vdpau @ 0x8a8cc20]mmco: unref short failure [h264_vdpau @ 0x8a8cc20]mmco: unref short failure [h264_vdpau @ 0x8a8cc20]reference picture missing during reorder [h264_vdpau @ 0x8a8cc20]Missing reference picture [h264_vdpau @ 0x8a8cc20]mmco: unref short failure [h264_vdpau @ 0x8a8cc20]reference picture missing during reorder [h264_vdpau @ 0x8a8cc20]Missing reference picture [h264_vdpau @ 0x8a8cc20]mmco: unref short failure [h264_vdpau @ 0x8a8cc20]reference picture missing during reorder [h264_vdpau @ 0x8a8cc20]Missing reference picture [h264_vdpau @ 0x8a8cc20]mmco: unref short failure [h264_vdpau @ 0x8a8cc20]reference picture missing during reorder [h264_vdpau @ 0x8a8cc20]Missing reference picture [h264_vdpau @ 0x8a8cc20]mmco: unref short failure -------------------------------- Swapnil Jain

1 0

Re: [MPlayer-dev-eng] EVR Renderer
by Reimar Döffinger 27 Jan '11

27 Jan '11

On Tue, Jan 25, 2011 at 06:09:41PM +0200, Georgi Petrov wrote: > Then when I try to implement DxVA on top of this EVR renderer, I will > have to create from scratch another 5000+ lines of header files > declarations? If this is the case, I will have to stop developing this > feature. In addition to the other suggestions there is no way I'll believe you need a 5000 line header to use EVR. Normally you shouldn't need more than a few lines per each feature you actually use (which for Microsoft APIs almost always is a tiny, tiny minority). If you mean dependencies: the wine headers usually work fine as replacements. Finally, modifying the header is an issue with concerns to copyright, and I have some doubts Microsoft allows modification in the license. However you can still use hacks like a different header you include first and that uses #defines to hack it to work as intended and similar stuff.

4 7

Re: [MPlayer-dev-eng] EVR Renderer
by Diego Biurrun 26 Jan '11

26 Jan '11

On Tue, Jan 25, 2011 at 07:03:32PM +0200, Georgi Petrov wrote: > > what is the license of the windows sdk? > > > > if its a permissable license we can include it. or maybe just host it > > like the directx7 headers? > > This would fix all my problems. The question is very good. I couldn't > find any explicit license, connected to the header files. The runtimes > and sample code examples are covered, but not the header files. They > are actually auto generated. Every file start with (for example): > > /* this ALWAYS GENERATED file contains the definitions for the interfaces */ > /* File created by MIDL compiler version 7.00.0555 */ > > I don't believe such files to be any different from the already hosted > DirectX 7 include files. Is it possible to do the same in case MinGW > decides not to include this file (and the other 3 or 4 required)? Yes. > BTW what is the problem with the DirectX 7 include files? Why aren't > they part of MinGW? Try submitting them. Diego

1 0