#gemrb@irc.freenode.net logs for 10 Feb 2011 (GMT)

Archive Today Yesterday Tomorrow
GemRB homepage


[00:03:52] <-- Textmode has left IRC (Ping timeout: 246 seconds)
[00:04:54] --> Textmode has joined #GemRb
[00:24:03] --> |Cable| has joined #GemRb
[00:39:40] <Lightkey> 0.6.3 on the linux game tome nao
[01:27:41] <-- barra_home has left IRC (Ping timeout: 255 seconds)
[01:28:14] <-- Maighstir has left IRC (Quit: ~ Trillian Astra - www.trillian.im ~)
[02:28:02] <-- edheldil_ has left IRC (Ping timeout: 240 seconds)
[04:32:17] --> jschall has joined #GemRb
[05:30:34] <-- jschall has left IRC (Remote host closed the connection)
[05:40:41] --> jschall has joined #GemRb
[05:56:43] --- Minalien is now known as Min|Oceanside
[06:40:26] <-- devurandom has left IRC (Quit: No Ping reply in 180 seconds.)
[06:42:32] --> devurandom has joined #GemRb
[07:21:51] --> Bo_Thomsen has joined #GemRb
[07:33:32] --> edheldil_ has joined #GemRb
[07:49:58] --> lynxlynxlynx has joined #GemRb
[07:49:58] <-- lynxlynxlynx has left IRC (Changing host)
[07:49:58] --> lynxlynxlynx has joined #GemRb
[07:49:58] --- ChanServ gives channel operator status to lynxlynxlynx
[07:57:23] <lynxlynxlynx> thanks for the heads up Lightkey
[08:10:41] <-- Bo_Thomsen has left IRC (Quit: Leaving.)
[08:21:38] <-- edheldil_ has left IRC (Ping timeout: 240 seconds)
[08:31:14] <-- Textmode has left IRC (Ping timeout: 240 seconds)
[08:44:39] --> lubos has joined #GemRb
[08:57:32] --> edheldil_ has joined #GemRb
[09:06:40] <-- edheldil_ has left IRC (Ping timeout: 272 seconds)
[09:41:27] <-- |Cable| has left IRC (Remote host closed the connection)
[09:52:59] --> Demitar has joined #GemRb
[10:40:11] --> SiENcE has joined #GemRb
[11:52:15] --> Maighstir has joined #GemRb
[14:13:31] --> edheldil_ has joined #GemRb
[16:08:09] <-- Demitar has left IRC (Quit: Ex-Chat)
[16:46:06] <-- lubos has left IRC (Quit: Leaving.)
[17:02:54] <-- SiENcE has left IRC (Quit: @all: cya)
[17:16:09] --> Bo_Thomsen has joined #GemRb
[17:31:26] <-- edheldil_ has left IRC (Ping timeout: 272 seconds)
[18:07:42] --> Avenger has joined #GemRb
[18:07:42] --- ChanServ gives channel operator status to Avenger
[18:08:20] <Avenger> hi
[18:12:14] <wjp> hi
[18:16:16] <Avenger> wjp: there is another ugly graphic bug in gemrb: the bink player has some artifacts i couldn't find
[18:16:50] <Avenger> the black isle logo has only a few bits of difference, but the intro movie is terrible
[18:19:32] <wjp> ugh, I keep forgetting all those arcane cmake options to use
[18:20:01] <wjp> isn't there some cmake --help or something that lists the relevant ones to us?
[18:21:19] <Avenger> i have no idea :)
[18:22:05] <Avenger> the make system is really unmotivating, eh?
[18:22:15] <fuzzie> It usually is, on any project.
[18:22:37] <Avenger> yes, i meant in general
[18:24:28] <Avenger> i couldn't make that fireplace animation pretty even when i set it as ground anim :(
[18:25:05] <Avenger> it is really puzzling me how it is supposed to work
[18:25:41] <wjp> hm, did button click sounds always have a delay of a second?
[18:25:42] <Avenger> there is a wallgroup on it, and it has a massive part of ground on top of it
[18:26:06] <Avenger> no, this is fuzzie's doing, i think :)
[18:26:19] <Avenger> hmm wait, i think it is only in the sdl sound
[18:26:38] <fuzzie> you have to manually specify sdl sound in config, anyway, and it works fine for me :P
[18:26:49] <Avenger> as in no delay?
[18:26:54] <fuzzie> yes
[18:26:58] <fuzzie> so difficult to fix :-)
[18:27:12] <wjp> hm, there are some nasty minor artifacts in the intro in scrolling transitions?
[18:27:26] <wjp> and still this audio thing?
[18:27:27] <Avenger> iwd2 intro movie? yes
[18:27:38] <Avenger> play with the black isle logo
[18:27:46] <Avenger> that one has only very minor problems
[18:27:53] <Avenger> and it is smaller too
[18:28:26] <Avenger> i tried to modify ffmpeg to output the bits at reading and decoding
[18:28:27] <wjp> hm, a few vertical stripes on top of the horizontal stripe in the logo at the end?
[18:28:29] <Avenger> it was fine
[18:28:38] <Avenger> yes vertical lines
[18:29:17] <Avenger> the wotc logo seems to be exact
[18:30:48] <-- jschall has left IRC (Remote host closed the connection)
[18:34:56] <wjp> vertical lines are spaced at exactly 8x8 unsurprisingly
[18:35:36] <Avenger> yep, i wish i know which block op is rendering them
[18:36:34] <wjp> pattern
[18:36:54] <wjp> I think
[18:38:36] <wjp> j and i are flipped in the output
[18:39:14] <wjp> ok, that fixed it
[18:39:36] <Avenger> oh hell
[18:39:38] <Avenger> :)
[18:39:46] <Avenger> you are great
[18:40:30] <lynxlynxlynx> :)
[18:40:38] <Avenger> if you want you can make the colors into an array, if the a?b:c construct is slower
[18:40:52] <wjp> I see that's what ffmpeg does
[18:41:04] <Avenger> yes
[18:42:19] <Avenger> this didn't fix much of the intro movie :(
[18:42:32] <Avenger> but still, a step in the right direction
[18:42:49] <wjp> that looked more like motion
[18:43:06] <wjp> which is unfortunately a bit more complicated :-)
[18:43:10] <Avenger> cool, the stripes are gone
[18:43:14] --> |Cable| has joined #GemRb
[18:43:24] <lynxlynxlynx> no sendmail still, huh
[18:44:20] <wjp> people are complaining about over in #sourceforge from time to time, so SF is aware of the issue
[18:44:56] * wjp pushes this pattern fix in the meantime
[18:45:13] <wjp> wow, this is my first commit in ages
[18:45:41] * Avenger [CIA-00] wjp committed a patch ;P
[18:45:50] <wjp> :-)
[18:48:29] --> edheldil_ has joined #GemRb
[18:49:01] <Avenger> wjp, could it be more like residue block?
[18:49:06] <Avenger> that's motion + something
[18:49:12] <wjp> 19:47 < meonkeys> Also, anyone know if sendmail is working again from git hooks?
[18:49:12] <wjp> 19:47 <@burley-sf> meonkeys: we're deprecating that stuff, announcement of it pending
[18:49:15] <wjp> 19:48 <@burley-sf> we're going to allow http outbound to use curl to hit a webpage or cgi to then do whatever you want
[18:49:18] <wjp> 19:48 <@burley-sf> for hook scripts
[18:49:20] <wjp> 19:48 <@burley-sf> we don't have all those bits in place, but the new web allows outbound email
[18:49:47] <Avenger> that means, we need to rewrite our own git hook, or it is central stuff?
[18:51:17] <lynxlynxlynx> yep
[18:51:37] <lynxlynxlynx> and wait enough for someone else to write a mail layer
[18:51:57] <wjp> annoying
[18:52:34] <wjp> Avenger: yes, likely a residue block or similar
[18:52:46] <wjp> INTER_ is similar
[18:52:50] --> Demitar has joined #GemRb
[18:53:29] <Avenger> meh, yeah. So there are many possibilities
[18:53:55] <Avenger> i changed read_residue somewhat, tried to be clever :)
[18:54:26] <Avenger> basically, using a single variable instead of two, for mode list
[18:54:27] <edheldil_> lynxlynxlynx: "yes" we need to rewrite hooks or "yes" we need to wait?
[18:55:02] <lynxlynxlynx> both, but i doubt a big hook change will be needed
[18:56:02] <lynxlynxlynx> i'm confident other people will be overannoyed by this and it will be addressed fast :)
[18:56:18] <lynxlynxlynx> then it's just about leveraging the open source model
[18:57:38] <Avenger> haha
[18:59:02] <wjp> I looked over the read_residue differences some time ago without finding any direct issues
[18:59:29] <lynxlynxlynx> iirc Avenger did the same, trying also with that reverted
[19:00:22] <Avenger> wjp, did you see that dst[j + i*stride] was also swapped?
[19:00:41] <wjp> where?
[19:01:28] <Avenger> meh, the pattern block, i don't know what i did there ;)
[19:01:58] <Avenger> i have no idea why did i moved those variables around
[19:03:38] --> jschall has joined #GemRb
[19:04:50] <Avenger> in our code, i see no place where we use put|add_pixel_clamped
[19:10:32] --> barra_home has joined #GemRb
[19:17:44] <Avenger> scaled_block.intra seems to be totally different
[19:21:43] <wjp> also the treatment of bink_intra_quant/bink_inter_quant
[19:24:43] <Avenger> i used an earlier source of ffmpeg i think
[19:30:03] <-- edheldil_ has left IRC (Ping timeout: 246 seconds)
[19:33:50] <Avenger> wjp, i see the scaled block code is setting up ublock first, then adding it to the buffer
[19:34:15] <Avenger> maybe modifying the buffer directly messes up the motion
[19:40:17] <-- jschall has left IRC (Remote host closed the connection)
[19:42:46] --> jschall has joined #GemRb
[19:55:37] <wjp> moved the newer bink_intra_quant/bink_inter_quant stuf over
[19:55:43] <wjp> seems pretty decent now
[19:59:32] <wjp> ran it side-by-side with ffmpeg now; no glaring differences left as far as I can tell
[20:01:07] <Avenger> wow, that's great!
[20:02:02] <Avenger> wait, that is some constant array
[20:02:18] <wjp> and corresponding code that uses it
[20:02:23] <Avenger> ah
[20:02:26] <wjp> that's significantly different now
[20:02:38] <Avenger> ok, THAT was not my fault then :)
[20:03:40] <Avenger> i messed a lot with trying to optimise it, but i didn't imagine the original was wrong too
[20:04:18] <Avenger> i'm glad you took this over, you are much more effective than i was
[20:05:11] <wjp> hm, although now I'm having trouble finding any artifacts even without updating this
[20:06:11] <wjp> do you still see any problems in intro after my pattern commit?
[20:06:59] <Avenger> i didn't check it, i just changed mine locally after you said what's wrong with the pattenrn
[20:07:05] <Avenger> lemme reboot into linux ;)
[20:07:07] <-- Avenger has left IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203075014])
[20:10:10] --> Avenger has joined #GemRb
[20:10:11] --- ChanServ gives channel operator status to Avenger
[20:11:12] <Avenger> the intro video has still the motion problems, since you didn't commit the arrays
[20:11:38] <wjp> can you pinpoint one of those?
[20:11:50] <Avenger> the falling leaves leave snail trail
[20:12:02] <Avenger> it is very noticeable
[20:12:18] <wjp> in the first seconds?
[20:12:22] <Avenger> yes
[20:13:03] <wjp> against the black background?
[20:13:05] <Avenger> hmm, you got a 64 bit machine?
[20:13:07] <Avenger> yes
[20:13:08] <wjp> yes
[20:13:19] <wjp> because I don't see that even without the quant changes
[20:13:26] <Avenger> that's ODD
[20:13:47] <Avenger> are you sure you you recompiled it? ;)
[20:14:08] <wjp> yes
[20:14:08] <Avenger> did you see the problem earlier?
[20:14:25] <Avenger> all leaves left garbage behind them
[20:14:33] <Avenger> and the whole movie was noisy
[20:18:10] <wjp> no, didn't see any trouble with the leaves
[20:18:33] <wjp> you don't have any local changes left? (Maybe disabling certain blocks as a test?)
[20:19:13] <Avenger> no
[20:19:34] <wjp> hm
[20:19:44] <wjp> maybe I'll try a clean build
[20:22:39] <wjp> no problems with the leaves even before the pattern patch
[20:22:43] <wjp> confusing
[20:28:07] <Avenger> really confusing
[20:28:19] <Avenger> how could you see it earlier then
[20:28:33] <wjp> I haven't see any problem with the leaves today
[20:28:55] <Avenger> ok
[20:28:57] <wjp> just some weird movement things during scrolling, but I'm starting to think those are just part of the video
[20:29:10] <Avenger> maybe i should just download gemrb from the scratch
[20:33:07] <Avenger> hmm, what to type when you start a local sandbox for the first time?
[20:34:47] <wjp> git clone ssh://username@gemrb.git.sourceforge.net/gitroot/gemrb/gemrb
[20:34:57] <wjp> (which will create a subdirectory called gemrb)
[20:35:54] <Avenger> hmm, what git fetch will do....
[20:36:04] <Avenger> i did git init + git fetch ssh:..... :)
[20:38:07] <wjp> that might not create the 'origin' remote
[20:38:47] <wjp> what happens if you say 'git remote' ?
[20:40:08] <Avenger> i restarted with clone
[20:41:41] <Avenger> doing make now
[20:43:32] <Maighstir> I can't see a difference between GemRB playing the IwD2 intro now, and the original. Both on Windows.
[20:44:12] <Avenger> strange, because i saw the ugly artifacts on both system
[20:44:27] <Maighstir> I can hear a difference though.
[20:44:28] <Avenger> the audio is still bad?
[20:44:35] <wjp> audio is a disaster
[20:44:38] <Avenger> yep, the intro audio is
[20:44:54] <Avenger> it is like it cuts the decoded sample
[20:45:24] <Maighstir> Not too much, but GemRb seems "coarser" or something, less natural.
[20:45:54] <Avenger> the movie is still wrong for me
[20:47:15] <Avenger> and i can play the SAME movie without problems in ffmpeg
[20:47:37] <Avenger> so, i wonder what's different with me
[20:47:55] <Avenger> i just downloaded/recompiled gemrb
[20:48:13] <fuzzie> are you using the right plugin?
[20:48:20] <Avenger> sure :P
[20:48:37] <Avenger> it is not crashing, just got heavy artifacts
[20:48:47] <Avenger> and two other movies play flawlessly now
[20:49:22] <Maighstir> Of course, the movie has heavy artifacts through the original as well.
[20:49:35] <fuzzie> well, often people come in and report problems and it turns out that they have their plugin path pointing somewhere with old stuff :P
[20:49:39] <wjp> ff49e18f91f778e668dba4d921b44644 CD2/Data/Intro.mve
[20:50:47] <Maighstir> ff49e18f91f778e668dba4d921b44644 *Intro.mve
[20:53:01] <Avenger> ff49e18f91f778e668dba4d921b44644 Intro.mve
[20:53:24] <Avenger> btw, the other ingame movies are also bad
[20:53:45] <Avenger> e06dff18a160bf0847091e8a8e1ab0b2 END.mve
[20:54:17] <Avenger> you can get them in the jukebox by adding END=1 into the movies section of your icewind2.ini
[20:55:23] <Avenger> so, i guess this is some kind of cpu bug, but our ffmpeg implementation doesn't use any acceleration
[20:55:42] <Avenger> especially... since my original ffmpeg plays the movie happily
[20:56:01] <Avenger> so, what else, my sdl?
[20:56:02] <wjp> I don't see any obvious problems with END either :-(
[20:56:27] --> Textmode has joined #GemRb
[20:57:01] <Maighstir> Neither do I, except for the robotic voice.
[20:57:25] <Avenger> meh
[20:57:39] <Textmode> meep?
[20:58:31] <Avenger> well, wjp, if you can, try to fix the voice too :)
[20:59:01] <Avenger> i will try gemrb in my workplace
[21:00:20] <wjp> I'm toying with audio now. No progress yet :-)
[21:00:59] <wjp> but it's not bits of silence
[21:04:54] <Avenger> bye!
[21:04:56] <-- Avenger has left IRC (Quit: bye!)
[21:16:31] <-- Bo_Thomsen has left IRC (Quit: Leaving.)
[21:59:20] <wjp> hurray
[21:59:48] <wjp> the dct invocation in ffmpeg seems quite different from that in gemrb git somehow
[22:00:04] <wjp> changing ours to match ffmpeg's seems to fix the audio
[22:04:44] <wjp> hard to say what exactly caused it... I'm not familiar enough with the relation between this specific dct and idct to easily see what changed
[22:07:21] <wjp> we went from DCT_II (dct) to DCT_III (idct) now, in any case, and some scale factors changed
[22:11:16] <wjp> I don't see any videos which use the rdft branch (as opposed to the dct one) so I can't test that
[22:16:33] <wjp> ok, pushed
[22:16:49] <wjp> Maighstir: could you confirm audio is ok now for you too?
[22:24:24] --> Bo_Thomsen has joined #GemRb
[22:39:05] <-- lynxlynxlynx has left IRC (Remote host closed the connection)
[22:40:28] --> lynxlynxlynx has joined #GemRb
[22:40:28] --- ChanServ gives channel operator status to lynxlynxlynx
[22:41:40] <lynxlynxlynx> huh, what was that
[22:41:55] <lynxlynxlynx> anyway, got iwd2 copied over, will go try the videos
[22:42:08] <wjp> I hope everything's ok now
[22:49:21] <lynxlynxlynx> don't see any graphical glitches
[22:49:33] <lynxlynxlynx> iirc the original intro was also a bit blurry
[22:50:01] <lynxlynxlynx> the sound stopped almost immediately after the start though
[22:50:09] <lynxlynxlynx> no problem in the bis logo movie
[22:51:22] <lynxlynxlynx> second time it plays normally
[22:52:47] <lynxlynxlynx> reproducible with MIDDLE.mve :s
[22:54:11] <wjp> when does it stop?
[22:55:40] <lynxlynxlynx> after a second or so
[22:55:54] <lynxlynxlynx> i just hear a buzt
[22:56:45] <lynxlynxlynx> restarted gemrb and middle works normally too
[22:57:24] <wjp> weird, never saw/heard that
[22:58:00] <lynxlynxlynx> try playing a lot of tem
[22:58:12] <lynxlynxlynx> play/stop
[22:58:27] <lynxlynxlynx> maybe it's a race
[22:59:00] <wjp> openal/sdl?
[22:59:37] <lynxlynxlynx> openal
[23:00:07] <wjp> tried stopping/starting 10 times without issues
[23:00:07] <lynxlynxlynx> i'll go try sdl - i wonder if it is laggy on this machine too
[23:00:18] <wjp> tried both openal and sdl now
[23:02:51] <lynxlynxlynx> cool, no sdl problem in videos or ingame
[23:07:12] <lynxlynxlynx> really nice job btw
[23:08:21] <-- Bo_Thomsen has left IRC (Quit: Leaving.)
[23:12:07] <wjp> thanks :-)
[23:12:53] <wjp> hm, will leave those strange audio things you see with openal or avenger's strange video artifacts for now...
[23:12:58] <wjp> good night
[23:13:28] <lynxlynxlynx> night
[23:16:40] <lynxlynxlynx> i hope i would find an empty cycle in the iwd2 states bams like for bg1, but it really doesn't have any :(
[23:18:34] <-- lynxlynxlynx has left IRC (Remote host closed the connection)