Instrumental Anime Project
- rose4emily
- Joined: Fri Jan 23, 2004 1:36 am
- Location: Rochester, NY
- Contact:
I should note that I still have the original submission versions of every segment submitted to me. It's possible that some of the authors have even better copies, but I can send over anything else in the earliest generation I've had a chance to see.
I definately don't have to (and wouldn't, anyway) send over the copies that I've produced in my own transcoding process.
---
About the transcoding - good news:
I've installed XviD, seems to work without problems, and actually looks even better than the libAVCodec encodings (at the cost of much longer encode times - I think most of the added quality came from some of XviD's ability to use Global Motion Compensation, which predicts motion over an entire frame before predicting motion for individual macroblocks, saving a lot of bits on panning/rotating/zooming scenes that can then be spent on sharper lines and richer textures).
I haven't been able to fix the AVI indices in such a way (using the old libAVCodecs, but the audio dropout probably has nothing to do with the choice of video codec implementation) that the merged AVI files play properly in WMP. I have had plenty of success with muxing the streams into Matroska - which is demuxed at playback by the same library on all platforms (at least for now...). I'm thinking that this means Matroska will be this project's container of choice.
Re-doing the transcoding, merging, and muxing for this project will take several hours, especially with XviD's slower encoding rates (given the settings I've found to be best in terms of quality within reasonable size limitations - setting a constant Q of 1 and using only intra-encoded frames actually codes pretty fast, and looks absolutely pristine when comparing the encoded picture to its source, but also makes for files a bit largish for even broadband users to practically download. I instead am resting much of the picture quality on the use of light, but effective filtering and a few smart settings on the motion compensation and quanization behavior).
Since the next two weeks are vacation for me, and this week is an incredibly busy one as I try to get ahead on some project work before my team splits up to go home, I'm afraid I won't be doing the new encode until Sunday, when I'm back in New Hampshire. I'll have direct 24-hour access to a Windows machine there, however, so testing the newest encode should be no problem at all.
I've replaced the background audio for the "Tomoe" narrative, at Chris's request. It is now a snippet of "Reverie", as recorded by Jacques Loussier.
I think I'll be leaving the aspects of everything untouched. So far as I can tell, there's been a lot of confusion - but no really strong opinions on the subject. Changing the two videos in question would also seriously throw off the balance of the two sections, upset a track order that seems to work pretty well, and require a re-recording of those two narratives (which I think Song did beautifully).
The end credits still have to be rendered. This isn't going to take too long, however, as I've already figured out all the details of how they need to be composited from frame to frame. I am sill accepting images for a couple more days, but I'm going to have to say that once I've rendered everything, that's the version we're keeping.
I've only rendered a test segment of the widescreen narratives so far, but doing the rest is entirely a matter of sitting around and pressing a couple of keys every five minutes or so. The test looks nice - and actually looks much "smoother" on the lip-syching than the Fullscreen narratives, probably due to the widescreen narrator's smaller mouth and the fact that the entire narrator fills a smaller portion of the screen to keep him cropped in about the same manner as Yukino was for the Fullscreen section, given the Acadamy aspect.
---
Hardware problems suck. I've had my share of those myslef, over the past couple years. Unfortunately, there isn't always all that much you can do about them.
---
You will all have working copies of the entire film by Christmas.
Oh yes.... you will.
I definately don't have to (and wouldn't, anyway) send over the copies that I've produced in my own transcoding process.
---
About the transcoding - good news:
I've installed XviD, seems to work without problems, and actually looks even better than the libAVCodec encodings (at the cost of much longer encode times - I think most of the added quality came from some of XviD's ability to use Global Motion Compensation, which predicts motion over an entire frame before predicting motion for individual macroblocks, saving a lot of bits on panning/rotating/zooming scenes that can then be spent on sharper lines and richer textures).
I haven't been able to fix the AVI indices in such a way (using the old libAVCodecs, but the audio dropout probably has nothing to do with the choice of video codec implementation) that the merged AVI files play properly in WMP. I have had plenty of success with muxing the streams into Matroska - which is demuxed at playback by the same library on all platforms (at least for now...). I'm thinking that this means Matroska will be this project's container of choice.
Re-doing the transcoding, merging, and muxing for this project will take several hours, especially with XviD's slower encoding rates (given the settings I've found to be best in terms of quality within reasonable size limitations - setting a constant Q of 1 and using only intra-encoded frames actually codes pretty fast, and looks absolutely pristine when comparing the encoded picture to its source, but also makes for files a bit largish for even broadband users to practically download. I instead am resting much of the picture quality on the use of light, but effective filtering and a few smart settings on the motion compensation and quanization behavior).
Since the next two weeks are vacation for me, and this week is an incredibly busy one as I try to get ahead on some project work before my team splits up to go home, I'm afraid I won't be doing the new encode until Sunday, when I'm back in New Hampshire. I'll have direct 24-hour access to a Windows machine there, however, so testing the newest encode should be no problem at all.
I've replaced the background audio for the "Tomoe" narrative, at Chris's request. It is now a snippet of "Reverie", as recorded by Jacques Loussier.
I think I'll be leaving the aspects of everything untouched. So far as I can tell, there's been a lot of confusion - but no really strong opinions on the subject. Changing the two videos in question would also seriously throw off the balance of the two sections, upset a track order that seems to work pretty well, and require a re-recording of those two narratives (which I think Song did beautifully).
The end credits still have to be rendered. This isn't going to take too long, however, as I've already figured out all the details of how they need to be composited from frame to frame. I am sill accepting images for a couple more days, but I'm going to have to say that once I've rendered everything, that's the version we're keeping.
I've only rendered a test segment of the widescreen narratives so far, but doing the rest is entirely a matter of sitting around and pressing a couple of keys every five minutes or so. The test looks nice - and actually looks much "smoother" on the lip-syching than the Fullscreen narratives, probably due to the widescreen narrator's smaller mouth and the fact that the entire narrator fills a smaller portion of the screen to keep him cropped in about the same manner as Yukino was for the Fullscreen section, given the Acadamy aspect.
---
Hardware problems suck. I've had my share of those myslef, over the past couple years. Unfortunately, there isn't always all that much you can do about them.
---
You will all have working copies of the entire film by Christmas.
Oh yes.... you will.
may seeds of dreams fall from my hands -
and by yours be pressed into the ground.
and by yours be pressed into the ground.
- downwithpants
- BIG PICTURE person
- Joined: Tue Dec 03, 2002 1:28 am
- Status: out of service
- Location: storrs, ct
if the xvid comes out too large, you may want to try setting the minimum quantizers to 2, which results in a good saving in space, and usually little loss in quality.
i'd prefer you use avi if you can rather than matroska, for the reason that many aren't familiar with matroska.
unfortunately i'm going to be on dialup after i leave for home today. but i maybe i'll ask a friend at home with cable to download the vids for me when they're finished.
rose if you wouldn't mind sending bakadeshi an mpeg2 render of my vid, i'd prefer that over me sending bakadeshi a cd via snailmail
i'd prefer you use avi if you can rather than matroska, for the reason that many aren't familiar with matroska.
unfortunately i'm going to be on dialup after i leave for home today. but i maybe i'll ask a friend at home with cable to download the vids for me when they're finished.
rose if you wouldn't mind sending bakadeshi an mpeg2 render of my vid, i'd prefer that over me sending bakadeshi a cd via snailmail
maskandlayer()|My Guide to WMM 2.x
a-m-v.org Last.fm|<a href="http://www.frappr.com/animemusicvideosdotorg">Animemusicvideos.org Frappr</a>|<a href="http://tinyurl.com/2lryta"> Editors and fans against the misattribution of AMVs</a>
a-m-v.org Last.fm|<a href="http://www.frappr.com/animemusicvideosdotorg">Animemusicvideos.org Frappr</a>|<a href="http://tinyurl.com/2lryta"> Editors and fans against the misattribution of AMVs</a>
- rose4emily
- Joined: Fri Jan 23, 2004 1:36 am
- Location: Rochester, NY
- Contact:
I am using 2 as a min quantizer. That's why I have spend all the extra CPU cycles on the super-heavy-duty motion estimation and smart filtering to keep the encoding quality as close to DVD standards as possible. If I was using the constant Q of 1 (which I do on occasion, when I borrow a DVD from a friend or a library and don't have a chance to watch it before I can return it - so I can still see a high-quality copy, which isn't a problem for temporary purposes as all of that disk space is freed after I finish watching the film), I could get a transfer indistinguishable from DVD quality from the Huffman-submitted segments. Unfortunately, I also get a video stream in the range of 5-6 Mbps for 2D animation, and 8-9 Mbps for live-action footage, which is really too much to realisticly distribute or keep in long-term storage.downwithpants wrote:if the xvid comes out too large, you may want to try setting the minimum quantizers to 2, which results in a good saving in space, and usually little loss in quality.
i'd prefer you use avi if you can rather than matroska, for the reason that many aren't familiar with matroska.
unfortunately i'm going to be on dialup after i leave for home today. but i maybe i'll ask a friend at home with cable to download the vids for me when they're finished.
rose if you wouldn't mind sending bakadeshi an mpeg2 render of my vid, i'd prefer that over me sending bakadeshi a cd via snailmail
I will do a test run of installing Matroska on Windows to see if it's something that most of our audience would be able to easily do. I know it's a breeze in Linux, especially if you use the urpmi package manager or install MPlayer from current sources, and am hoping that Windows installation is a simple "click to download, double-click for the install dialog" process. If installing Matroska is more complex than that, I can always mux to Matroska, then demux to AVI, which would rebuild the streams and their indices from scratch, eliminating any inconsistencies caused by the file-merge process that might be missed by an index repair utility that simply fixes what it recognizes to be errors. If I'm going to do all that, however, it would be nice to just keep it in Matroska - it's a better format, and also a bit more space efficient than AVI, due to the nature of its index structure.
I'll probably end up sending Huffman copies of any videos Bakadeshi doesn't get directly from their authors, since I'm not so sure it's a great idea to try writing a DVD encoded using multiple versions of the MPEG2 codec - and equally unsure of how well the one MPEG2 encoder implementation I've even managed to successfully install on my system works. If the network transfer turns out to be too much, I'll burn the Huffmans to couple of data DVDs and send them through the mail.
may seeds of dreams fall from my hands -
and by yours be pressed into the ground.
and by yours be pressed into the ground.
- jasper-isis
- P. Y. T.
- Joined: Tue Aug 13, 2002 11:02 am
- Status: catching all the lights
The thing is that our separate files should all be usable as long as we all use the DVD-compatible settings posted by Bakadeshi above.rose4emily wrote:...since I'm not so sure it's a great idea to try writing a DVD encoded using multiple versions of the MPEG2 codec - and equally unsure of how well the one MPEG2 encoder implementation I've even managed to successfully install on my system works.
To everybody: remember to leave out your own titles and credits! I forgot to do this at first and had to reencode.
To Bakadeshi: Okay, I've got my two m2v files and their corresponding wav files. How can I get these to you?
____
A release before Christmas would be really, really sweet.
![Very Happy :D](./images/smilies/icon_biggrin.gif)
- Bakadeshi [AuN Studios]
- Joined: Wed Mar 24, 2004 7:59 pm
- Location: Georgia / S. FL WIP: ROS2, VG3, AR2
- Contact:
Yep, it should't be a problem no matter what codec brand you use, as long as it is mpeg2 compliant, and use the DVD compatibility settings I posted earlier. However, Huffyuv wouldn't be bad, I don;t mind doing the encode myself, the problem is their size...Jasper-Isis wrote:The thing is that our separate files should all be usable as long as we all use the DVD-compatible settings posted by Bakadeshi above.rose4emily wrote:...since I'm not so sure it's a great idea to try writing a DVD encoded using multiple versions of the MPEG2 codec - and equally unsure of how well the one MPEG2 encoder implementation I've even managed to successfully install on my system works.
How to get them to me now, I'm still trying to think of a good method since I don;t have FTP space available right now. I wonder if Kalium would mind us using his ftp space to upload to and I grab them all from there...
I'll also try to set up a ftp server on my PC, but I've had limited success with that, even with setting up port forwarding, probably something to do with the 2 firewalls I have to go through to get to the internet (cable modem and router).
THe other method would be to snail mail me them on CDR, PM me if you want to use that method and I'll give you an address.
- jasper-isis
- P. Y. T.
- Joined: Tue Aug 13, 2002 11:02 am
- Status: catching all the lights
- rose4emily
- Joined: Fri Jan 23, 2004 1:36 am
- Location: Rochester, NY
- Contact:
The question on whether different MPEG2 implementations should work together properly is, unfortunately a different question from whether they will work together properly - much like what we've just witnessed with different (yet all supposedly complient) MPEG4 implementations.
1.13 GiB * 1024 = 1,157 MiB
1157 MiB * 1024 = 1,184,890 KiB
1,184,890 KiB * 8 = 9,479,127 Kb (1 Byte = 8 bits)
9,479,127 Kb / (60 Kb/s) = 157,985 seconds = 2633 minutes = 43.88 hours
Even assuming you mean MPEG2 copies, instead of the HuffYUV versions, that only increases the compression by a factor of 4-6, giving you a 7 to 10 hour upload at that rate.
I think CD submission really might be the way to go for the DVD compilation, if AIM transfer is the fastest (and most reliable - that's the really scary part) internet transfer mechanism available.
---
Then the really strange part - this proves that it's sometimes possible for information to be carried more quickly by a USPS worker driving a mail truck than an electronic deleviery network operating at half the speed of light.
---
I've actually gone and packed up my desktop computer so I can finish the XviD encoding version while I use my laptop to do the coursework I need to get out of the way over the next week. I was also concerned about whether my laptop - which is only an inch thick and poorly equipped for heat dissapation (as it demonstrates by making a lot of fan noise, its backplate becoming about as hot as a freshly boiled pot of tea, and something near the memory module emitting some strange, loud, and really irritating high-pitched squealing noise any time I run an extended CPU-intensive process on it) - is really up to the task of eight hours of continuous video encoding, and another eight hours of raytracing for the end credits (I'm going to produce more frames of the 3D render, to allow for smoother motion).
Now I just have to deal with carrying a 110 pound box (Desktop tower, plus a large number of books that I need for my Sofware Engineering project) onto a train and placing it on the overhead luggage shelf. Those shelves are easily strong enough, but it's been two months since I've had time for the gym, so I've unfortunately found that lifting that sort of stuff overhead isn't quite as easy as I had recalled it to be. I just love it when this sort of thing reminds me of the wonderful effects this intellectual lifestyle has upon physical health and fitness. That, the sleep deficiency, the seventy percent male population at this school (ninety-something percent in my major, which is especially scary when combined with the effect the computer geek subculture has in disrupting the kind of normal social interaction I grew fond of back in high school), the lovely upstate NY weather, the mounting piles of debt waiting for me when I graduate - all the little things that make me want to get out of bed each day.
Ah well, at least I have two great housemates, and will probably enter a wonderful career in a few years. That's two good things about this place, at least.
The Huffman copies you sent to me? The two files, together, are about 1.13 GiB.Jasper-Isis wrote:Hmm... care to do an old-fashioned AIM/MSN file transfer?
I can upload at ~60kb/s, so my two files should transfer in under an hour and a half.
1.13 GiB * 1024 = 1,157 MiB
1157 MiB * 1024 = 1,184,890 KiB
1,184,890 KiB * 8 = 9,479,127 Kb (1 Byte = 8 bits)
9,479,127 Kb / (60 Kb/s) = 157,985 seconds = 2633 minutes = 43.88 hours
Even assuming you mean MPEG2 copies, instead of the HuffYUV versions, that only increases the compression by a factor of 4-6, giving you a 7 to 10 hour upload at that rate.
I think CD submission really might be the way to go for the DVD compilation, if AIM transfer is the fastest (and most reliable - that's the really scary part) internet transfer mechanism available.
---
Then the really strange part - this proves that it's sometimes possible for information to be carried more quickly by a USPS worker driving a mail truck than an electronic deleviery network operating at half the speed of light.
---
I've actually gone and packed up my desktop computer so I can finish the XviD encoding version while I use my laptop to do the coursework I need to get out of the way over the next week. I was also concerned about whether my laptop - which is only an inch thick and poorly equipped for heat dissapation (as it demonstrates by making a lot of fan noise, its backplate becoming about as hot as a freshly boiled pot of tea, and something near the memory module emitting some strange, loud, and really irritating high-pitched squealing noise any time I run an extended CPU-intensive process on it) - is really up to the task of eight hours of continuous video encoding, and another eight hours of raytracing for the end credits (I'm going to produce more frames of the 3D render, to allow for smoother motion).
Now I just have to deal with carrying a 110 pound box (Desktop tower, plus a large number of books that I need for my Sofware Engineering project) onto a train and placing it on the overhead luggage shelf. Those shelves are easily strong enough, but it's been two months since I've had time for the gym, so I've unfortunately found that lifting that sort of stuff overhead isn't quite as easy as I had recalled it to be. I just love it when this sort of thing reminds me of the wonderful effects this intellectual lifestyle has upon physical health and fitness. That, the sleep deficiency, the seventy percent male population at this school (ninety-something percent in my major, which is especially scary when combined with the effect the computer geek subculture has in disrupting the kind of normal social interaction I grew fond of back in high school), the lovely upstate NY weather, the mounting piles of debt waiting for me when I graduate - all the little things that make me want to get out of bed each day.
Ah well, at least I have two great housemates, and will probably enter a wonderful career in a few years. That's two good things about this place, at least.
may seeds of dreams fall from my hands -
and by yours be pressed into the ground.
and by yours be pressed into the ground.
- jasper-isis
- P. Y. T.
- Joined: Tue Aug 13, 2002 11:02 am
- Status: catching all the lights
Nope, I've uploaded a 64mb file in hardly more than 20 minutes. So my DVD submissions, which are 236mb total, should come to just under an hour an a half. Can't say that for some of your behemoth videos though. ![Razz :P](./images/smilies/icon_razz.gif)
I don't know about anybody else, but my Huffyuv-to-MPEG2 conversion compressed by about a factor of 10.
Anyway, I'm asking my friend (who owns the server that I use) if I can temporarily upload these files and let Bakadeshi download them directly off the server.
![Razz :P](./images/smilies/icon_razz.gif)
I don't know about anybody else, but my Huffyuv-to-MPEG2 conversion compressed by about a factor of 10.
Anyway, I'm asking my friend (who owns the server that I use) if I can temporarily upload these files and let Bakadeshi download them directly off the server.
- Bakadeshi [AuN Studios]
- Joined: Wed Mar 24, 2004 7:59 pm
- Location: Georgia / S. FL WIP: ROS2, VG3, AR2
- Contact:
Actually because of the many diferent settings in the many different implementations of the mpeg4 codec, and the method used to encode and decode mpeg4 implementations, you are bound to find incompatibility issues in mpeg4 codecs much more so than in mpeg2 codecs. WHile you have many diferent mpeg4 implementations, all of wich arn't even totally compatible with each other, (divx, xvid, 3vix, various freeware options, Miscorosft mpeg4) theres, to my knowledge, only 1 Mpeg2 standard that everyone follows.rose4emily wrote:The question on whether different MPEG2 implementations should work together properly is, unfortunately a different question from whether they will work together properly - much like what we've just witnessed with different (yet all supposedly complient) MPEG4 implementations.
.
Anyway, to totally disolve that issue, we could always just require tempgenc be used to make the mpeg2 encodes. The program(s) is freely available online for a free 15 day trail, if $30-$60 is too much for someone to buy them for. They have 3 diferent programs all capable of making compatible mpeg2 files that I've tested to work in my DVD authoring software.
I Have to agree on the Aim transfer thing, since it can't resume, we'd hate to have it crap out at 90% transfer frate and have to start all over again ;p I had this problem trying to send my video to Kalium over AIM, ended up just ftping it ;p
-
- Joined: Sat Aug 16, 2003 10:16 pm
- Location: cleveland
- Contact:
Would FTPing to a fast server be an option? If you could try to give me a maxium amounth of bandwidth it would take to transfer the files to the server then off it I can see if i can set something up. It would be much quicker if we did that b/c then it wouldn't be like standing in line to transfer your video. Just a Thought.
http://www.ooshna.com <--- click here b/c I can't make banners
http://www.animemusicvideos.org/members ... hp?v=38520
http://www.animemusicvideos.org/members ... hp?v=38520