it appears your encoder/player do not yet reflece the changes rillian introduced wrt theora headers, which have been split into three parts -- header proper, comment, and tables (see toplevel.c). It's strange that your code works at all linked against the new libs; presumably since the tables still have the hard-coded values, if you encode & decode from the same lib compile you won't see the problem -- the tables just never get loaded & use default values. You can test this by trying to decode with NEW_FREQS (hufftables.h) set to 1 (use a file encoded with it set to zero, or vice versa). You should see obvious artifacts that get progressively worse after each keyframe. Right now I think I have the encoder working properly, but I'm having trouble with the player. Should be debugged in a couple days. Clearly the code needs to be cleaned up so it errors or at a minimum doesn't initialize these tables to any default values. Hopefully then we can do a real merge & test/ checkin. Final bitstream change would be the arbitrary size fix, with a cropping rectangle in the header. Anyone working on that? ___ Dan Miller (++,) Founder, On2 Technologies <p>> -----Original Message-----> From: Mauricio Piacentini [mailto:mauricio@tabuleiro.com] > Sent: Friday, May 23, 2003 4:51 PM > To: theora-dev@xiph.org > Subject: [theora-dev] Win32 experimental player/encoder > > > This one is ust to let you all know that Windows versions of > a Theora simple > player and encoder are checked into the "win32/experimental" > subdirectory of > the CVS tree. This should allow the guys working on Windows > to test the > library more easily. I have also implemented simple audio latency > compensation, so audio and video playback should be > synchronized well (at > least they are on my tests.) The required portaudio/SDL includes are > supplied, so just checkout the whole tree and rebuild. > > Regards, > Mauricio > > --- >8 ---- > List archives: http://www.xiph.org/archives/ > Ogg project homepage: http://www.xiph.org/ogg/ > To unsubscribe from this list, send a message to > 'theora-dev-request@xiph.org' > containing only the word 'unsubscribe' in the body. No > subject is needed. > Unsubscribe messages sent to the list will be ignored/filtered. >--- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'theora-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
umm -- to clarify -- I don't particularly care about Win32 per se, I'm just using it to test. The changes I'm making are further bitstream mods to the headers, so they should in theory only affect the libs. ___ Dan Miller (++,) Founder, On2 Technologies> -----Original Message----- > From: Dan Miller > Sent: Sunday, May 25, 2003 1:19 AM > To: theora-dev@xiph.org > Subject: RE: [theora-dev] Win32 experimental player/encoder > > > it appears your encoder/player do not yet reflece the changes > rillian introduced wrt theora headers, which have been split > into three parts -- header proper, comment, and tables (see > toplevel.c). > > It's strange that your code works at all linked against the > new libs; presumably since the tables still have the > hard-coded values, if you encode & decode from the same lib > compile you won't see the problem -- the tables just never > get loaded & use default values. You can test this by trying > to decode with NEW_FREQS (hufftables.h) set to 1 (use a file > encoded with it set to zero, or vice versa). You should see > obvious artifacts that get progressively worse after each keyframe. > > Right now I think I have the encoder working properly, but > I'm having trouble with the player. Should be debugged in a > couple days. Clearly the code needs to be cleaned up so it > errors or at a minimum doesn't initialize these tables to any > default values. > > Hopefully then we can do a real merge & test/ checkin. Final > bitstream change would be the arbitrary size fix, with a > cropping rectangle in the header. Anyone working on that? > > ___ Dan Miller > (++,) Founder, On2 Technologies > > > > -----Original Message----- > > From: Mauricio Piacentini [mailto:mauricio@tabuleiro.com] > > Sent: Friday, May 23, 2003 4:51 PM > > To: theora-dev@xiph.org > > Subject: [theora-dev] Win32 experimental player/encoder > > > > > > This one is ust to let you all know that Windows versions of > > a Theora simple > > player and encoder are checked into the "win32/experimental" > > subdirectory of > > the CVS tree. This should allow the guys working on Windows > > to test the > > library more easily. I have also implemented simple audio latency > > compensation, so audio and video playback should be > > synchronized well (at > > least they are on my tests.) The required portaudio/SDL includes are > > supplied, so just checkout the whole tree and rebuild. > > > > Regards, > > Mauricio > > > > --- >8 ---- > > List archives: http://www.xiph.org/archives/ > > Ogg project homepage: http://www.xiph.org/ogg/ > > To unsubscribe from this list, send a message to > > 'theora-dev-request@xiph.org' > > containing only the word 'unsubscribe' in the body. No > > subject is needed. > > Unsubscribe messages sent to the list will be ignored/filtered. > > > --- >8 ---- > List archives: http://www.xiph.org/archives/ > Ogg project homepage: http://www.xiph.org/ogg/ > To unsubscribe from this list, send a message to > 'theora-dev-request@xiph.org' > containing only the word 'unsubscribe' in the body. No > subject is needed. > Unsubscribe messages sent to the list will be ignored/filtered. >--- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'theora-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
On Sunday, May 25, 2003, at 06:19 am, Dan Miller wrote:> Final bitstream change would be the arbitrary size fix, with a > cropping rectangle in the header. Anyone working on that?I've done some work in that direction already, but as before it might be a week or two before I'm done. -r --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'theora-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
Actually Dan I am using the current lib as it is in the tree, so the updated toplevel.c is being used. Did you rebuild your libraries first (using the theora static project?) The experimentatal player/encoder project is not forcing a rebuild of the libraries, so make sure you are rebuilding them first. But I see what you mean regarding the comments header and tables, gilles has updated the encoder/player as well on the 19th. I will have a look at it and update the samples. Regards, Mauricio ----- Original Message ----- From: "Dan Miller" <dan@on2.com> To: <theora-dev@xiph.org> Sent: Sunday, May 25, 2003 2:19 AM Subject: RE: [theora-dev] Win32 experimental player/encoder <p>> it appears your encoder/player do not yet reflece the changes rillian introduced wrt theora headers, which have been split into three parts -- header proper, comment, and tables (see toplevel.c).> > It's strange that your code works at all linked against the new libs;presumably since the tables still have the hard-coded values, if you encode & decode from the same lib compile you won't see the problem -- the tables just never get loaded & use default values. You can test this by trying to decode with NEW_FREQS (hufftables.h) set to 1 (use a file encoded with it set to zero, or vice versa). You should see obvious artifacts that get progressively worse after each keyframe. --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'theora-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
Corrected now, Dan, both encoder and player, to use the 3 header structure. Fixes are checked into the tree. I can now see the difference when setting NEW_FREQS, etc. Just update your files and it should be ok. One minor concern is that I am getting these results (filesize increase) using the same source file, 720X480, default compression settings: 1. old theora alpha format/encoder - 7126KB 2. new format (with comments/tables) - 7591 KB 3. new format, using NEW_FREQ = 1 in the lib - 7,643 KB This is a large file (4 minutes, fullscreen). It is copyrighted unfortunately, otherwise I could post it so you could test on the Linux players. Regards, Mauricio Piacentini Tabuleiro ----- Original Message ----- From: "Dan Miller" <dan@on2.com> To: <theora-dev@xiph.org> Sent: Sunday, May 25, 2003 2:19 AM Subject: RE: [theora-dev] Win32 experimental player/encoder> Right now I think I have the encoder working properly, but I'm havingtrouble with the player. Should be debugged in a couple days. Clearly the code needs to be cleaned up so it errors or at a minimum doesn't initialize these tables to any default values. --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'theora-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
> One minor concern is that I am getting these results > (filesize increase) > using the same source file, 720X480, default compression settings: > > 1. old theora alpha format/encoder - 7126KB > 2. new format (with comments/tables) - 7591 KB > 3. new format, using NEW_FREQ = 1 in the lib - 7,643 KBHmmm -- well, this is what I've been seeing -- random changes in file size. The encoder is clearly not initializing something in at least some situations. The header is only around 6K (my new fix will reduce it to 3K) -- you've got half a meg of difference.> > This is a large file (4 minutes, fullscreen). It is copyrighted > unfortunately, otherwise I could post it so you could test on > the Linux > players. > > Regards, > Mauricio Piacentini > Tabuleiro > > ----- Original Message ----- > From: "Dan Miller" <dan@on2.com> > To: <theora-dev@xiph.org> > Sent: Sunday, May 25, 2003 2:19 AM > Subject: RE: [theora-dev] Win32 experimental player/encoder > > > Right now I think I have the encoder working properly, but > I'm having > trouble with the player. Should be debugged in a couple > days. Clearly the > code needs to be cleaned up so it errors or at a minimum > doesn't initialize > these tables to any default values. > > --- >8 ---- > List archives: http://www.xiph.org/archives/ > Ogg project homepage: http://www.xiph.org/ogg/ > To unsubscribe from this list, send a message to > 'theora-dev-request@xiph.org' > containing only the word 'unsubscribe' in the body. No > subject is needed. > Unsubscribe messages sent to the list will be ignored/filtered. >--- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'theora-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.