From danapple Wed Mar 8 00:41:51 1989 Received: by FLOTSAM.MIT.EDU (5.45/4.8) id AA00825; Wed, 8 Mar 89 00:41:51 EST From: Daniel I. Applebaum Message-Id: <8903080541.AA00825@FLOTSAM.MIT.EDU> To: galatea-users, video-mail@athena Subject: Big switchers Date: Wed, 08 Mar 89 00:41:47 EST I have a devised a system by which I can extend Galatea to handle switchers with multiple outputs. The method I've been thinking of will require no changes to the protocol. Unfortunately, to make this method work efficiently will require the server final output numbers be restricted. Server output numbers used to be any integer. Now, they will have to be positive, and there will be a maximum integer, which will be compile-time changeable. I don't believe either of these restrictions should cause any problem, but if anybody thinks of a serious reason why these restrictions are burdensome, let me know. I'm thinking of making the default number of outputs 10, numbered 0-9. For SIGGRAPH, this number will be at least 20. Dan. From rubin@media-lab.media.mit.edu Wed Mar 8 14:17:39 1989 Received: by FLOTSAM.MIT.EDU (5.45/4.8) id AA04737; Wed, 8 Mar 89 14:17:39 EST Received: by media-lab.media.mit.edu (5.61/4.8) id AA08549; Wed, 8 Mar 89 14:16:47 -0500 Date: Wed, 8 Mar 89 14:16:47 -0500 From: Ben Rubin Message-Id: <8903081916.AA08549@media-lab.media.mit.edu> To: danapple@FLOTSAM.MIT.EDU Subject: Re: Big switchers Cc: galatea-users@FLOTSAM.MIT.EDU, video-mail@ATHENA.MIT.EDU As long as the maximum allowable outputs will accomodate our 16x16 matrix, it sounds good to me -b From hkbirke@MOVIES.MIT.EDU Sun Apr 23 22:29:33 1989 Received: by FLOTSAM.MIT.EDU (5.45/4.8) id AA16009; Sun, 23 Apr 89 22:29:33 EDT Received: by MOVIES.MIT.EDU (5.45/4.8) id AA01179; Sun, 23 Apr 89 22:29:55 EDT From: Hal Birkeland Message-Id: <8904240229.AA01179@MOVIES.MIT.EDU> To: danapple@FLOTSAM.MIT.EDU Cc: rubin@MEDIA-LAB.MEDIA.MIT.EDU, hkbirke@MOVIES.MIT.EDU, galatea-users@FLOTSAM.MIT.EDU, bgardner@ATHENA.MIT.EDU Subject: Galatea Seamless Playback (new version proposal) Date: Sun, 23 Apr 89 21:29:52 EST Seamless playback routine modification (grammar -- text format): List ::= Clips : Switches Clips ::= Clip ; Clips | Clip Switches::= Switch ; Switches | Switch Clip ::= Time, Volume , InFrame , OutFrame [, Speed] Switch ::= Time, VolumeFrom, VolumeTo [, Mask] Time ::= Frame InFrame ::= Frame OutFrame::= Frame Volume ::= VolumeName VolumeFrom ::= VolumeName VolumeTo ::= VolumeName Frame ::= Integer Speed ::= Integer Mask ::= Integer VolumeName ::= String What all this means (as if you really care Dan): A seamless playback is now composed of two distinct segments: the clip and the switch. Clips are the physical disc motion commands, the rolls and releases. Switches are the appearance changes on the client's output. They are being separated after a long discussion with Ben where we talked about the hassles of hierarchies and the current assignment methods. It is very easy to determine where the switches occur in a playback, but not always where to put them in the current structured list. Also this will allow for better optimization of the generated list and improved resource allocation. We will be producing a set of library routines (client side) to facilitate using these commands... here is an initial spec for it: /* structures */ typedef structure _clip { int frame; char *volumename; int in,out,speed; } Clip; typedef structure _switch { int frame; Clip *in, *out; int mask; } Switch; typedef structure _list { Clip *clip[MAX_CLIPS]; Switch *switch[MAX_SWITCHES]; } List; /* create a new list to work with */ List *SeamlessCreate() /* free a current list */ void SeamlessFree(list) List *list; /* add a clip */ int SeamlessAddClip(list,clip) List *list; Clip *clip; /* add a switch */ int SeamlessAddSwitch(list,swtch) List *list; Switch *swtch; /* switch is a reserved word */ /* create a clip */ Clip *SeamlessCreateClip(time, volumename, in, out, speed) char *volumename; int time, in, out, speed; /* time is frames from start */ /* create a switch */ Switch *SeamlessCreateSwitch(time, in, out, mask) Clip *in, *out; /* in and out clips */ int time, mask; /* time is frames from start */ This left to be decided... There needs to be an improved interface here to isolate the programmer from all of the nastiness associated with overlaying shots... but there doesn't seem to be a nice way of doing it... here are a few of my suggestions: /* Return "top" clip at a given point in time */ Clip *SeamlessTopClip (list, time) List *list; int time; /* Optimize List -- Whee!!! */ SeamlessOptimize (list) List *list; /* Find Current List */ List *SeamlessCurrent (list) /* make a new current list */ SeamlessMakeCurrent (list) List *list; /* play current list, downloading if needed */ SeamlessPlayback () --hal From danapple Mon Apr 24 00:15:10 1989 Received: by FLOTSAM.MIT.EDU (5.45/4.8) id AA16140; Mon, 24 Apr 89 00:15:10 EDT From: Daniel I. Applebaum Message-Id: <8904240415.AA16140@FLOTSAM.MIT.EDU> To: Hal Birkeland Cc: danapple@FLOTSAM.MIT.EDU, rubin@MEDIA-LAB.MEDIA.MIT.EDU, galatea-users@FLOTSAM.MIT.EDU, bgardner@ATHENA.MIT.EDU, danapple Subject: Re: Galatea Seamless Playback (new version proposal) In-Reply-To: Your message of Sun, 23 Apr 89 21:29:52 EST. <8904240229.AA01179@MOVIES.MIT.EDU> Date: Sun, 23 Apr 89 23:15:05 EST Great, wonderful! Except: check for naming conflicts (ie. I already have a #define for Switch. Dan. From hkbirke@MOVIES.MIT.EDU Tue May 9 20:25:43 1989 Received: by FLOTSAM.MIT.EDU (5.45/4.8) id AA20508; Tue, 9 May 89 20:25:43 EDT Received: by MOVIES.MIT.EDU (5.45/4.8) id AA01459; Tue, 9 May 89 20:26:04 EDT Date: Tue, 9 May 89 20:26:04 EDT From: Hal Birkeland Message-Id: <8905100026.AA01459@MOVIES.MIT.EDU> To: galatea-users@FLOTSAM.MIT.EDU Subject: multiple client problem: seamless MAJOR PROBLEM WITH CURRENT SEAMLESS PLAYBACK IMPLEMENTATION when running with multiple clients... a client is not guaranteed that his playback list will be the one he requests to playback. Scenario: Client #1 downloads a 15 second playlist Client #2 downloads a 2 hour playlist Client #1 play list. Client #1 will play the 2 hour and not the fifteen second list... sigh. This will be repaired in Seamless Part 2, due out in 10 days or so. --hal hkbirke@movies.mit.edu From danapple Thu Mar 15 17:28:15 1990 Received: by flotsam.MIT.EDU (5.61/4.8) id AA06141; Thu, 15 Mar 90 17:28:15 -0500 From: Daniel I. Applebaum Message-Id: <9003152228.AA06141@flotsam.MIT.EDU> To: galatea-users Subject: New Galatea Date: Thu, 15 Mar 90 17:28:14 EST Sender: danapple Galatea 2.4 is now available. This newest version of the Galatea Network Video Device Control System has many new features compared to the previous release, 2.2. Some of the enhancements include the capability to record onto write-once video devices and run-time remounting of video discs. To get the new release, you may ftp the tar or cpio files from flotsam.mit.edu using anonymous ftp. The files galatea.2.4.tar and galatea.2.4.cpio are available under the galatea subdirectory. Or, if you are unable to retrieve the files that way, I'd be happy to mail you a uuencoded version of either file. Dan.