April 2002 - MPlayer-dev-eng - lists.mplayerhq.hu

docs: users-against-developers versus gcc-2.96-3.0
by Diego Biurrun 09 May '02

09 May '02

Hi! I'm almost finished reworking users-against-developers.html, which is quite out of date now that binary redistribution is no longer a problem. But that's not the point. While going over the document, I realized that users-against-developers.html and gcc-2.96-3.0.html have not only a lot in common but are in fact redundant. users-against-developers describes the whole situation, gcc-2.96-3.0 gives a few more technical details. Both documents reference each other by a "Also read this text!" link, emphasizing how closely related they are. I believe that one topic should not be covered in two different places. I would therefore like to merge gcc-2.96-3.0.html into users-against-developers.html. gcc-2.96-3.0.html is not very long and as some of it is redundant, I believe I can condense it into 20-30 additional lines of text in users-against-developers.html. Removing the file will of course break a few links in other parts of the docs, but I will find and correct those in the patch. Since this is more of a structural change than what I have done before, I would like to have your agreement before proceeding. Regards Diego

3 11

Re: mga_vid as a patch
by Dariusz Pietrzak 04 May '02

04 May '02

>> it via some other means. >it would be nice if it's included in the kernel :) >but i'm sure it will never be... :( I'll try to ask alternative kernel maintainers ( Alan Cox probably wouldn't be eager to do so, but WOLK or jp could do it, especially if we'd provide the same not only for mga but also radeon and sis). I have put the patch in cvs in drivers/ directory, it is also available for download at http://eyck.forumakad.pl/mplayer/ as mgavid.patch. hmm.. as I look at it now, it is 52k uncompressed, and probably shouldn't be put in cvs. Patch is against clean 2.4.18, mga_vid driver is then accessible via Console drivers -> framebuffer support -> Matrox acceleration -> Matrox mga_vid driver. -- Dariusz Pietrzak yck, ... zaklął książę pan po francusku

4 3

[ANNOUNCE] MPlayer prerelease RPM packages available for testing
by Dominik Mierzejewski 01 May '02

01 May '02

I'm happy to announce the RPM packages for MPlayer here: http://home.elka.pw.edu.pl/~dmierzej/mplayer.html Built from today's snapshot, they still need some minor tuning, but basically this is what they'll look like. We need to sort out this xanim issue. For now I just patched codecs.conf, but this is not a good solution. Testing experiences, comments, suggestions, etc. welcome. -- "The Universe doesn't give you any points for doing things that are easy." -- Sheridan to Garibaldi in Babylon 5:"The Geometry of Shadows" Dominik 'Rathann' Mierzejewski <rathann(at)rangers.eu.org>

7 29

[patch] passthrough mode for audio plugins, bugfix for pl_resample
by Christian Ohm 01 May '02

01 May '02

hi. the attached patch adds a passthrough mode to all audio plugins. i've tested only pl_resample and pl_volnorm, but all others should work, too. the pl_resample patch also fixes a bug where the volume would be decreased if fin == fout. (at least i can't see a reason for the right-shift of the data when fin == fout.) bye christian ohm -- Walking on water wasn't built in a day. -- Jack Kerouac

4 6

libmpdvdkit works on BSD/OS - patch included
by Steven M. Schultz 01 May '02

01 May '02

Greetings - If this does not break anything could it be included before rc3? The problem is that the directories "BSDI_dvdioctl" and "extras/BSDI_dvdioctl" only exist within the 'vlc' program from Videolan.Org - the pathnames are incorrect for MPlayer. The only other small change to ./configure was to make sure that DVD_STRUCT_IN_BSDI_DVDIOCTL_DVD_H is _not_ set unless the system is "bsdos" - that shouldn't break anything (and indeed might help prevent breaking something). With these changes I can build/use 'libmpdvdkit' on BSD/OS. Thanks! Cheers, Steven Schultz sms(a)2bsd.com ------------------------cut here------------------------ --- configure.dist Tue Apr 30 16:02:25 2002 +++ configure Tue Apr 30 16:17:06 2002 @@ -1612,13 +1612,13 @@ echores "$_dvd" -echocheck "extras/BSDI_dvdioctl/dvd.h" +echocheck "BSDI dvd.h" cat > $TMPC << EOF -#include <extras/BSDI_dvdioctl/dvd.h> +#include <dvd.h> int main(void) { return 0; } EOF _bsdi_dvd=no -cc_check && _bsdi_dvd=yes +cc_check && bsdos && _bsdi_dvd=yes if test "$_bsdi_dvd" = yes ; then _def_bsdi_dvd='#define DVD_STRUCT_IN_BSDI_DVDIOCTL_DVD_H 1' else --- libmpdvdkit/ioctl.c.dist Wed Apr 24 12:28:03 2002 +++ libmpdvdkit/ioctl.c Tue Apr 30 16:12:25 2002 @@ -53,7 +53,7 @@ # include <dvd.h> #endif #ifdef DVD_STRUCT_IN_BSDI_DVDIOCTL_DVD_H -# include <BSDI_dvdioctl/dvd.h> +# include <dvd.h> #endif #ifdef SYS_BEOS # include <malloc.h>

1 0

Patch to skip 3dnow downmixes in liba52
by Stephen Davies 01 May '02

01 May '02

Hi, Attached is a trivial patch to stop liba52 using the buggy 3dnow downmix routines. This at least gets multichannel ac3 output working again for AMD users until I or someone can fix the downmixes. I didn't yet commit it - rather have your comments first. But it does seem a shame to do a new release with this still broken. Steve

4 7

some cooperation -- mplayer & SnapStream
by Rakesh Agrawal 30 Apr '02

30 Apr '02

hi-- hetz (affiliated with slashdot.org) had contacted us last week and he sent us the way of your application. we make PVR and media server software for the Windows platform and we make heavy use of windows media (use microsoft's encoder/format SDKs, dshow for capture, use our own component to stream). hetz suggested we talk to you about doing some work together, that your component could possibly be a piece in a linux version of SnapStream. and on the other side, hetz indicated that you guys would be interested in the stuff we do in our application in terms of streaming windows media video files. so I thought I would initiate a conversation to see what kind of mutual interest there might be in our working together on something. thoughts? Regards, Rakesh Rakesh Agrawal CEO/Co-founder SnapStream Media +1-713-644-6240 office +1-713-446-7725 cell Watch TV... Anytime, Anywhere: www.snapstream.com

2 1

0.90pre2 is out
by Arpi 30 Apr '02

30 Apr '02

Hi, pre2 is out, with last-minute fix by Steven for ac3-downmix-3dnow disable and removing FrNo from codecs.conf by me A'rpi / Astral & ESP-team -- Developer of MPlayer, the Movie Player for Linux - http://www.MPlayerHQ.hu

3 4

[Patch] Subtitles colors problem.
by salvador 30 Apr '02

30 Apr '02

Hi All! This is a follow up for a thread I started in mplayer-users. I'm including the description of the patch again. Purpose of this patch: Make fonts visible when reading an MPEG stream and no VOB is available to get the palette. Long explanation: After talking with Felix Buenemann about it I looked into spudec.c and found the problem. As Felix told me the palette isn't in the stream and is different for each DVD/MPEG2 stream. But the problem is that mplayer uses an "old black palette" (memory allocated with calloc => all 0) and this will generate a bad palette all the time, black subtitles are impossible to read in most movies. The fact is that subtitles uses 4 colors from a palette of 16 colors. One of this colors usually (I think that's all the time but, who knows?) is a transparent color (achieved using the alpha channel). The other 3 colors are used to get a border and an "inside" part (sorry, I'm not good for english ;-). In many cases only 2 colors are really used (one repeated or made transparent with the alpha channel). What I did is the following: 1) In spudec_new_scaled when palette is NULL I enable a flag indicating we don't have a palette and we must generate the colors on the fly. Same for the new spudec_new_scaled_vobsub. 2) When we get the alpha information (in all the cases I tested it comes after the colors, should we assume that's the case for all?) I added: if (auto_palette) compute_palette(this); This function counts the used colors and assigns different gray levels to these colors so we get a good contrast. With this we get visible fonts. We have 50% of probability to get black fonts with white around and 50% to get the reverse. In any case the subtitles are ok. As a plus I added a function to pass the "font factor" (-ffactor) value to this routine so we can choose to get a very big contrast or just white fonts. If this patch is accepted I want to also add: 1) The possibility to specify a palette from the command line. 2) Some alpha channel control so we can make the subtitles translucent (?) SET -- Salvador Eduardo Tropea (SET). (Electronics Engineer) Visit my home page: http://welcome.to/SetSoft or http://www.geocities.com/SiliconValley/Vista/6552/ Alternative e-mail: set(a)computer.org set(a)ieee.org Address: Curapaligue 2124, Caseros, 3 de Febrero Buenos Aires, (1678), ARGENTINA Phone: +(5411) 4759 0013 Index: mplayer.c =================================================================== RCS file: /cvsroot/mplayer/main/mplayer.c,v retrieving revision 1.482 diff -u -r1.482 mplayer.c --- mplayer.c 25 Apr 2002 13:27:15 -0000 1.482 +++ mplayer.c 26 Apr 2002 13:30:30 -0000 @@ -1152,6 +1152,7 @@ if (vo_spudec==NULL) { current_module="spudec_init_normal"; vo_spudec=spudec_new_scaled(NULL, sh_video->disp_w, sh_video->disp_h); + spudec_set_font_factor(font_factor); } if (vo_spudec!=NULL) Index: spudec.c =================================================================== RCS file: /cvsroot/mplayer/main/spudec.c,v retrieving revision 1.25 diff -u -r1.25 spudec.c --- spudec.c 25 Apr 2002 18:46:44 -0000 1.25 +++ spudec.c 26 Apr 2002 13:30:31 -0000 @@ -65,6 +65,11 @@ unsigned char *scaled_aimage; } spudec_handle_t; +/* 1 if we lack a palette and must use an heuristic. */ +static int auto_palette = 0; +/* Darkest value used for the computed font */ +static int font_start_level = 0; + static inline unsigned int get_be16(const unsigned char *p) { return (p[0] << 8) + p[1]; @@ -191,6 +196,43 @@ } } + +/* + This function tries to create a usable palette. + Is searchs how many non-transparent colors are used and assigns different +gray scale values to each color. + I tested it with four streams and even got something readable. Half of the +times I got black characters with white around and half the reverse. +*/ +static void compute_palette(spudec_handle_t *this) +{ + int used[16],i,cused,start,step,color; + + memset(used, 0, sizeof(used)); + for (i=0; i<4; i++) + if (this->alpha[i]) /* !Transparent? */ + used[this->palette[i]] = 1; + for (cused=0, i=0; i<16; i++) + if (used[i]) cused++; + if (!cused) return; + if (cused == 1) { + start = 0x80; + step = 0; + } else { + start = font_start_level; + step = (0xF0-font_start_level)/(cused-1); + } + memset(used, 0, sizeof(used)); + for (i=0; i<4; i++) { + color = this->palette[i]; + if (this->alpha[i] && !used[color]) { /* not assigned? */ + used[color] = 1; + this->global_palette[color] = start<<16; + start += step; + } + } +} + static void spudec_process_control(spudec_handle_t *this, unsigned int pts100) { int a,b; /* Temporary vars */ @@ -244,6 +286,7 @@ this->alpha[1] = this->packet[off] & 0xf; this->alpha[2] = this->packet[off + 1] >> 4; this->alpha[3] = this->packet[off + 1] & 0xf; + if (auto_palette) compute_palette(this); mp_msg(MSGT_SPUDEC,MSGL_DBG2,"Alpha %d, %d, %d, %d\n", this->alpha[0], this->alpha[1], this->alpha[2], this->alpha[3]); off+=2; @@ -682,12 +725,23 @@ memcpy(spu->global_palette, palette, sizeof(spu->global_palette)); } +void spudec_set_font_factor(double factor) +{ + font_start_level = (int)(0xF0-(0xE0*factor)); +} + void *spudec_new_scaled(unsigned int *palette, unsigned int frame_width, unsigned int frame_height) { spudec_handle_t *this = calloc(1, sizeof(spudec_handle_t)); if (this) { - if (palette) + if (palette) { memcpy(this->global_palette, palette, sizeof(this->global_palette)); + auto_palette = 0; + } + else { + /* No palette, compute one */ + auto_palette = 1; + } this->packet = NULL; this->image = NULL; this->scaled_image = NULL; @@ -707,7 +761,10 @@ if (palette){ memcpy(this->global_palette, palette, sizeof(this->global_palette)); memcpy(this->cuspal, cuspal, sizeof(this->cuspal)); + auto_palette = 0; } + else + auto_palette = 1; //(fprintf(stderr,"VobSub Custom Palette: %d,%d,%d,%d", this->cuspal[0], this->cuspal[1], this->cuspal[2],this->cuspal[3]); this->packet = NULL; this->image = NULL; @@ -715,6 +772,19 @@ this->orig_frame_width = frame_width; this->orig_frame_height = frame_height; this->custom = custom; + #if 0 + /* SET: Shouldn't be like this? */ + auto_palette = 1; + if (palette){ + memcpy(this->global_palette, palette, sizeof(this->global_palette)); + auto_palette = 0; + } + this->custom = custom; + if (custom) { + memcpy(this->cuspal, cuspal, sizeof(this->cuspal)); + auto_palette = 0; + } + #endif } else perror("FATAL: spudec_init: calloc"); Index: spudec.h =================================================================== RCS file: /cvsroot/mplayer/main/spudec.h,v retrieving revision 1.10 diff -u -r1.10 spudec.h --- spudec.h 15 Apr 2002 19:17:12 -0000 1.10 +++ spudec.h 26 Apr 2002 13:30:33 -0000 @@ -11,6 +11,7 @@ void spudec_free(void *this); void spudec_reset(void *this); // called after seek int spudec_visible(void *this); // check if spu is visible +void spudec_set_font_factor(double factor); // sets the equivalent to ffactor #endif

2 3

Re: [MPlayer-dev-eng] vobsub and sdl vo
by 陆然 30 Apr '02

30 Apr '02

Hi >Currently my patch uses the dys parameter passed to spudec_draw_scaled as the >"vertical size of visible area", if this value includes this area then my >patch can move subtitles there. > >SET > i read you patch today and i think your code can solve the problem, maybe can go a little bit further: move the subtile to the black area just as text sub do in sdl_vo, i guess it can be used in all vo(maybe need some effort). that will be a great thing. ---------------------- Best Regard LR _________________________________________________________________ 与联机的朋友进行交流，请使用 MSN Messenger: http://messenger.microsoft.com/cn

1 0