August 2003 - MPlayer-G2-dev - lists.mplayerhq.hu

[?] hit 3 flies - aspect ratio, resize, query_format
by Arpi 15 Sep '03

15 Sep '03

Hi, I'm trying to solve several design issues/problems/ideas around the image size, including the bypassing buffer size to vo driver in query_format, aspect ratio handling in vf layer, cleanup of VFCTRL_RESIZE. After several hours of analyzing ideas, i'm still in doubt, but here is what i found out the best: feel free to tell me better solutions or point out possible problems with this method! I would extend vf's query_format() by a int p[6] parameter. (actually int* size, which points to an array of 6 integers) these 6 values are: buff_w, buff_h - w/h of the image buffer (real pixels) disp_w, disp_h - pre-scaled w/h (recommended display size) [for startup] want_w, want_h - wanted output size [for window resizing] query_format() of 'normal' filters (which dont alter aspect ratio nor buffer size) would just pass thru the pointer to next filter. other filters shoudl implement it this way: query_format(...){ - change buff_w/h (only filters which chaneg buffer dimensios) - change disp_w/h (only filters which change aspect ratio) - call next filter's query_format() - change want_w/h (only filters which change buffer dimensios) } also, the scaling flags of vfcap.h shoudl be reviewed: merging HSWSCALE_UP and HWSCALE_DOWN, it ha sno sence to keep them separated. query_format() implementations can now check source and dest resolution so can decide if sw/hw scaling is possible or not. if they can do the scaling (or resize), they should change the want_w/h values. otherwise left unchanged. actually buff_* and disp_* are the current w/h pairs passed in vf->config(). buff-* is the real (phisical) size of the buffer, ignoring any kind of apect ratios. it's originally the width/height of the movie, altered by filters which change the buffer dimensions (so scale, crop, expand, rotate). disp_* are the scaled (by movie aspect, and aspect changes by filters) dimensions. this can be used to get the final aspect ratio at the end (vo). if we put the prescale dmovei size into it at the top, we'll get the recommended display window size. it's useful for startup, before the window created. want_* are the values from current VFCTRL_RESIZE. it's the wanted output size (values of disp_* at the end of chain, ie. at vo). it's useful when the user resized the window by mouse, or switched to fullscreen. we can calculate dimenstions backward. value of want_* may be 0 (or -1?), in this case the final values of disp_* will be copied at the end of chain (vf_vo2). I would also change config() of vf layer. The second w/h pair (d_width/h_height) will be the wanted w/h. config() functions shouldn't change it. But the filters which are able to do some kind of resizing, should call query_format of next filter from config(), to check if the rest of the filter chain is able to fulfill the resizing request (to wanted w/h), and do it otherwise. Ok, I see (know) this sounds very complex. But at least quite easy to implement (but a big work, mostly testing and discovering hidden side-effects and tricks used in current filters). Also it affects special (size/aspect changig) filters only, so only a few. (the rest should be changed to byupass new parameter of query_format, but it has rto be done anyway to solve vo->query_format problem (some vo supports some colorspaces only at limited size range, same for scaling) Comments? A'rpi / Astral & ESP-team -- Developer of MPlayer G2, the Movie Framework for all - http://www.MPlayerHQ.hu

3 3

pre41 released + aspect + fullscreen
by Arpi 22 Aug '03

22 Aug '03

Hi, pre41: - libvo2: split VOCTRL_RESIZE_DEST to _RESIZE_DEST+_MOVE_DEST+_REDRAW - event.c: RESIZE now takes care of aspect ratio (vo drivers needs to be fixed to clear unused (out of buffer) area!!! patches wanted!) - dir c0nfigure renamed to confiG2ure :) - applied stream_ftp patch v3, ported by Fabian, original by Albeu So nothing interesting. For the aspect stuff: i've checked Fabian's fs patch, and the aspect.[ch] from g1 is something we NEVER ever want to see in g2. What it does is saving various parameters (orig, scaled, display w/h) to static vars via various functions, and then let s1 to read them back. Instead, we should have all the needed info in the vo struct, and do aspect calculation (hey, it's just a few lines, doesnt worth a new library) on-the-fly, ie. in the default event handler (event.c) or in the Gui. Now i've added a few lines to event.c to maintain aspect ratio when resizing, including 'maximize'. It's not final (may be inaccurate), we should add aspect w/h to vo struct, and let vf_vo2 set it at config(). Actually we should review/rework the aspect stuff of vf layer anyway, but it's another story, and it can be delayed. About the fullscreen switching: Fabian's patch "ported" most of the x11 fs code from g1 as-is. The main idea behind the 2 windows fs switching is to avoid all that mess. At least it works in xine, i remember seeing a half screen long fs switching code there using 2 windows. No messing with WM hints, etc, just create a window with some special flags. I'm not an X11 guru, not even a X11 newbie, so i can't decide hwo can it be done, but I think (at least hope) that porting that mess from g1 is not the Right Way. Note, that gtk and SDL also uses 2 window method for fs switch, and they both work since ages, without the need for that fs mess and commandline switches and WM patches we hacked together in g1. A'rpi / Astral & ESP-team -- Developer of MPlayer G2, the Movie Framework for all - http://www.MPlayerHQ.hu

1 1

[PATCH][TESTSUITE] fullscreen-code from g1
by Fabian Franz 20 Aug '03

20 Aug '03

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, attached the new fs patch, especially for Gabucino to test it ... There are some issues: - - First: Design, this patch simply sucks. Its just a test suite for added x11_fs.c and for two window-system. So no resize, etc. ... This file is for guis and every code, that wants to use fullscreen (if with two windows or XReparentWindow). As it would be really bloat if each gui implemented this file itself ... I plan to make more functions static, all variables are already static and a header-file is missing, yet. This patch is NOT for inclusion. Its just for testers and people, who want to have fullscreen with g2 ;) - - Second issue is, that its all about timing, if you press 'f' at the right time, you get a good fullscreen-window else its not on top or not even there, just vanished ... Thats strange behaviour, but I guess there is somewhere a 10l I missed, yet ;) After you pressed f at leat 3 times it then does work: So the following should bring you always to good fullscreen-mode ;) : echo fff | ./test-play <filename> cu Fabian -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.0 (GNU/Linux) iD8DBQE/QAMpI0lSH7CXz7MRAj36AJ0WUP6d7CBLmMk+A90JVonpIOeBLwCeMkW0 wppYZjkdWqYY1IcOYdfnaJ4= =Klkz -----END PGP SIGNATURE-----

2 6

[PATCH][TESTSUITE2] fullscreen again ...
by Fabian Franz 20 Aug '03

20 Aug '03

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, much cleaner, but not clean anough for A'rpi. ;) Still I don't understand the architecture enough, to change it accordingly ... So for now aspect-code is in x11_helper.c. However there are just 3 new global static vars... (and they _are_ needed) and 2 new EVENTS - FULLSCREEN & MOVIE, where the last sets the MOVIE-aspect, as I could not get this from vf_vo2 in any other way ... I can't see how to include aspect nicely with vf_layer from x11_helpers point of view ... Also you say, vo should not handle fullscreen directly and so on, but it has flags for vm, zoom, fs and flip ?! That seems to me a bit confusing ... cu Fabian -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.0 (GNU/Linux) iD8DBQE/QrfyI0lSH7CXz7MRAs0hAJ4kk9GIBlXPAPyyrc/gOMpbHUoN9gCfQENP AROVD2Ilxz612q1uT2iliKg= =Qtwm -----END PGP SIGNATURE-----

2 1

[PATCH] Fullscreen issues - g2 is great :-)
by Fabian Franz 17 Aug '03

17 Aug '03

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, I wanted to gain some experience in the field of fullscreen :-). (And g1 was sooooooooo slow in compiling that it was a pain to develop in it :)) ) So I started with g2 and it was astounishingly easy ... :) I create a second window on the fly if fullscreen, which I hacked in as I couldn't find the function, where events are parsed, is selected ... This window still needs to be removed of any decorations and so on, but one can guess what it will look like ... For that I had to "port" the aspect-code (copy & paste, and comment something out) ... That however gives some questions: - - Are two static variables that are global ok for the fullscreen and backup-window ? (only x11_helper.c affected) - - How can vo_xv, vo_x11 give the aspect code the correct screensize (vo_screenwidth, vo_screenheight) if it does not know about it ... -> The same goes for centering the picture (btw. I think this should go to aspect.c as it always the same and aspect.c renamed to helper or tools or similar ...) Of course, x11_helper.c could include aspect.h and set the aspect, when it knows about screenwidth, screenheight, but this gives the problems that gui-programmers NEED to do the same ... Which is bad, as they could get undesired behaviour and for fullscreen and don't know why ... Also for fullscreen I'm not satisfied with this code in x11_helper.c, as perhaps it would be nice to have this functions available for gui, but I dunno ... Should guis also use the helper-code ? I added screenwidth,height as global (I know its ugly) variables for now ... - - vo_x11 is always fullscreened to the widht, height, the window had ... Why is the RESIZE_CODE uncommented, it would be easier if it was handled that way ... In fullscreen resize-commands are ignored ... :) Btw. thanks for adding fullscreen flag to vo_instance_s :)). - - zoom parameter not yet accepted by vo_x11 (as there is no wide-used config layer yet ) So, I really like the new clean design ... Its really nice, with alle the events and callbacks handled nice and allowing finally to communicate with vo->layer :). cu Fabian -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.0 (GNU/Linux) iD8DBQE/P3SbI0lSH7CXz7MRAl8CAJ9dBZJgEewnP/LXo3AdF68diknc1QCeNUp4 lmFJGfXKw+MGEaSOUabvSNM= =D0Nb -----END PGP SIGNATURE-----

3 5

module_list vs. gui
by Arpi 16 Aug '03

16 Aug '03

Hi, As I already found it, and tdoay Albeu asked too, there is a problem with the current config element type MODULE_LIST. The current code uses a function to find modules by name, but it's not useful for guis where you want to list the available modules. Let's see what are our options: - find_module_by_name() (current) - not good for gui (can't list modules) - pointer to module_t array - not good for code, there may be multiple lists, dinamically loadable modules (and as we discussed earlier, a CLI shouldnt load all those, only the explicitly named modules), and special hidden modules (for internal use, like null or vf_vo2/vf_vd wrappers) - enumeration func: a function which returns a single module at a call, and can be called until you get null, to get all available modules (you should pass the previous return value (or null at 1st call) as parameter, to make it thread safe / reentrant) I would go for a mix of the 1st and 3rd option, ie. an enumeration function which can also do searching. Something like the good old findfirst/findnext of DOS: module_t* enum_modules(module_t* prev, char* mod_name, int type) prev: the result of previous call, NULL for 1st call mod_name: name of the module we're searching for (when explicitly named), or NULL (when we list all modules) type: some type value, like 0=static(builtin modules), 1=dinamic -1=internal (note: this field shouldn't be changed between the first and other calls, and implementations may read it only at 1st call) Mayeb it could be flags instead, ie 1, 2, 4, so they can be OR'ed So listing public (not hidden/internal) static and dinamic modules: for(t=0;t<=1;t++){ module_t* prev=NULL; while((prev=enum_modules(prev,NULL,t))){ printf("[%s] %s\n",t?"dinamic":"static", prev->name); } } And searching for a named module: mod=enum_modules(NULL,"vf_vo",0); You will ask why don't make 2 functions instead, one for search, and one for list/enumerate. The answer is simple: there is only one 'priv' (pointer) parameter in the config_t struct, and adding one more only for this is silly. A'rpi / Astral & ESP-team -- Developer of MPlayer G2, the Movie Framework for all - http://www.MPlayerHQ.hu

3 2

[PATCH] ftp port from g1 to g2 :)
by Fabian Franz 14 Aug '03

14 Aug '03

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 hi, attached is the port from ftplib in g1 to g2 ... what still strange is that g2 needs much more time, before it displays the video/audio, but then it works perfect (at least for me ;)). Perhaps some cache-problem ? cu Fabian -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.0 (GNU/Linux) iD8DBQE/O/veI0lSH7CXz7MRAoBBAJ9KHgJcqoJ95woJh6KJY082SaJp1wCfcDII CpagrGESpA24EuN1VG4I0yQ= =yh9q -----END PGP SIGNATURE-----

1 2

release: pre-40
by Arpi 12 Aug '03

12 Aug '03

Hi, pre40: - applied mingw/win32 patch from Sascha Sommer, shmem.[ch] updated from main - vf_remcirc, lame unfinished circle removal filter - first try at modular configure (c0nfigure/, name inspired by autom4te :)) - vo drivers reordered, now: directx, xv, xover, x11, fbdev, mga, tdfx_vid - fixed some bugs in vd_ffmpeg and vf_vd PTS handing code (GoneNutty.avi) - demuxer: type of absolute PTS changed from float to rate_m-based integer!! now you can get time in seconds: pts/rate_m, and frame number: pts/rate_d :) - vd_ffmpeg updated to CVS/main version, to fix slices func (width=>type) - vf_vo2: check buffer_h>=mpi->height too (fixes green line / sig11 bug) A'rpi / Astral & ESP-team -- Developer of MPlayer G2, the Movie Framework for all - http://www.MPlayerHQ.hu

3 3

MainActor
by Fabian Franz 11 Aug '03

11 Aug '03

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 hi, today mainconcepts released its video editing software MainActor for Linux ... As we are speking about GUI's and necessary compontents, we should also take a look, what the others do: http://www.mainconcept.com/mainactor_v5.shtml#linux Its for now in a preview version and can be downloaded ... Its for an SuSE Linux 8.2 / Mandrake 9.1 System and has as major dependancy qt3.1.1 ... On Systems, where qt3 is splitted to plugins it seems necessary to do: export LD_PRELOAD=/usr/lib/qt3/plugins/styles/libqwindowsstyle.so (Repeat for echt plugin that is expected to be in the binary or recompile libqt3 from source with threads) (Thats the case on debian-based systems for example) Another major dependancy is Pentium III / Athlon with 700 Mhz. (They use sse2/mmx2 optimized code for the codecs) However with deleting or renaming libmcdvd_32.so it was possible to start it, with just an AMD K6 II and one can still get an impression how the program works ... (DnD and context-menu seem to be two major parts of the ui-design) I'm just telling you, so that MPlayer G2 and its guis can compete with commercial projects :-). A'rpi, what would be nice was, if you could give a small status report about what is finished and what still needs to be done. (Somewhat more detailed than the TODO-List) ... Was just an idea :-) cu Fabian -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.0 (GNU/Linux) iD8DBQE/N5e5I0lSH7CXz7MRAkqvAJsHJC8wuq9uo5DY4PNDitWpYbSxGgCeJkn0 cRc1vljXOX2XCDa5I5x3kVs= =iacj -----END PGP SIGNATURE-----

1 0

libaf
by Anders Johansson 11 Aug '03

11 Aug '03

Hi, I will try to convert libaf to the new general config system this weekend. Is there anything else you want me to change in the audio filter layer while I am at it? How is it going with the inclusion of my chages to libao? //Anders

3 5