QuickTime • 1:04:22
The MPEG-4 standard promises to revolutionize digital media production and delivery, today and into the future. In this session you'll learn how simple it is to encode with MPEG-4 using QuickTime, and achieve interoperable playback in industry-standard, ISO-compliant clients. You'll also learn how to optimize your production process to get the most out the MPEG-4 video and audio codecs using QuickTime.
Speakers: Dennis Backus, John Howell, Andrew Beach
Unlisted on Apple Developer site
Transcript
This transcript was generated using Whisper, it may have transcription errors.
Okay, it's just about time, so let's go ahead and get started. Everybody should be here for session 705, The Art of MPEG-4 Encoding. My name is Dennis Backus. I work on the QuickTime team at Apple Computer. Thank you. We're going to talk today about the hows of MPEG-4 encoding, not so much the whys, but the hows. Encoding in general is somewhat of a difficult process. To a certain extent, it's a science, but good encoding is actually an art, hence the title of this session. And today we're very lucky to have two, I think, some of the premier artists in the encoding world, and especially in the MPEG-4 encoding world. Andy Beach, right here to my left, and then John Howell from an organization. Thank you. from an organization called Last Exit from the East. So without any further ado, let's get started. Andy? Thank you very much.
Hi, everyone. I am Andy Beach. I'm from Last Exit. This is John Howell. Last Exit is a-- started as a web design and technology company but John and I are both from a film and video background. Quickly we turned to video on the web and interactive video as being some of the primary work that we wanted to do.
Dennis asked that I try and explain the last exit, naming some. Basically we were working for a rather large global company, got a little tired of the way everything was being run and the way things were being handled so we made our last exit out of a out of the big world and into our own company because if we were going to work for assholes we could work for ourselves So what is MPEG-4? What can it do for you? Apple has a lot that they've talked about it, and that's what we're all here for. Basically, what we want to talk about, though, is that MPEG-4 means standardizations, and it does mean interoperability, the two most important things for us.
What we're going to and what I'm going to do is I'm going to try very hard today to intro everything but then rapidly hand off so that we can actually get to some meat and potatoes demonstration of MPEG-4 encoding. And then lots of time for you guys to ask questions because I'm sure that you've had encoding nightmare question problems coming up. So we want to make sure that there's plenty of opportunity to get into that today. We are going to discuss, though, a basic understanding of the MPEG-4 spec. I don't feel necessarily qualified to fully speak to it, but I'm certainly going to do my best, given the people that are in the room today. How to produce the MPEG-4 content that you want, different ways to deliver that, you know, the different appliances out there, and demonstrate a workflow for creating your MPEG-4. Okay.
So, and again, I'll try to make this as brief as possible so that you don't have too many history lessons today. MPEG was introduced as a standard. QuickTime 6 brought MPEG-4 to us, the last QuickTime Live, 2002. We're now seeing support for AAC audio, MPEG-4 video, and with the latest DOT3, we're now getting the 3GP as well.
If we look at the two file formats, you're going to notice something very similar in the two graphics. Obviously, they are very similar because the MPEG-4 is based off of the architecture of the QuickTime. Right now, we're just dealing with the video and the audio aspects, but there is a great deal more that's there and things to come. We're patient. We'll wait. Okay.
Scalability is by far the most important aspect. This is our friend Terry Tate. I don't know how many of you got to see some of the Terry Tate content that came out last year. The scalability, designing for playback on a variety of devices, flexibility in the bit rates and the frame sizes that it's going to handle. This is the key to the MPEG-4, moving past something like an MPEG-2 broadcast architecture.
So this is the standards-based, going back to a wide array of platforms, developing for one format but being able to play back everywhere. The Internet Streaming and Media Alliance, if there's an analogy out there, we certainly look for the perfect analogy. MPEG-4 is really the ingredients, and the ISMA is the cookbook, giving you the recipes for all the various profiles. So it's going to tell you how everything should be structured and put together to meet those profiles that you're trying to. MPEG-4 has obviously got a great deal more behind it and under the hood. So it's that raw source that you're going to be able to tap into.
The two that we're concerning ourselves with with the encoding right now, even though it will do all of these great things But to get this started off it really is the video and the audio that are particularly important to us There's a number of ISO compliant MPEG-4 codecs out there in addition to the into the MPEG-4 There's a lot of interoperable plugins as well like in VIVIO and the impeccable to to get stuff out The audio, the AAC stuff, it's a lot less processor intensive than it has been on other things in the past, so you get a lot better playback on some lower platform systems. There's a great deal more audio support, and you get a lot higher audio in a lot smaller file now. Thank you.
For content creation, I want to ask John to step in. Okay, this I guess is my part of the speech. Apple wanted to make it really easy for everybody to be able to encode MPEG-4. They really threw themselves behind the standard and pretty much want to transition content creation and viewing on the Internet into MPEG-4.
And so they gave us a bunch of different ways to create MPEG-4 content using iMovie, Final Cut Pro, QuickTime Pro, you name it. They want you to do MPEG-4 coming out of it. So I'm going to outline sort of how easy it is to do MPEG-4 content, sort of go over an overview of how the QuickTime exporters work, and show you some other options for encoding. Thank you.
Okay, so basically to create an MPEG-4 file, you can open a video in Final Cut, or sorry, QuickTime Pro. You hit Export, Apple E, Command E. You choose MPEG-4. You choose a default setting, and you watch the progress bar go across the screen, and you end up with an MPEG-4 file on the other end. It's basically that easy. So I don't know. How many people out here are actively doing MPEG-4 content creation?
Most everybody? Cool. Those of you who are there, a bunch of you who are doing content creation and encoding that you don't want to go to MPEG-4, you're not gone to MPEG-4 yet, are those people here? Yeah, a couple? Okay. So hopefully we can sort of convince you to switch over. So if you look at the two screens, the one on the left, I guess, is export to MPEG-4. The one on the right is the default setting. But you move ahead to the actual QuickTime exporter, and it has a really neat feature in that when you open a program that knows how to use QuickTime, like Final Cut Pro, like After Effects, like iMovie, they basically all export video using the same MPEG-4 exporter, the Apple ISO-compliant MPEG-4 exporter. You create the.mp4 file out of that, and it'll play back on any ISO-compliant MPEG-4 player. You can use the Nvivo plug-in to play it back in real player on a PC. I've been told it works in Windows Media, but I try not to look at Windows Media, so I don't know. The Apple QuickTime encoder is a kind of neat thing. They've made it do one-pass VBR encoding, so it tries, you set the bitrate you want, You set it to be 300 kilobits a second, and it gets everything it can at 300 kilobits a second. But then you have areas that don't need that, so it scales it down some. And you get a smaller file size, which for an end user is a good thing. The encoder is optimized for a G4 processor. It doesn't fly, but it's pretty fast. Interesting to see what it does on the dual 2 gigahertz G5. Looking forward to that. So this is the QuickTime 6 exporter. Just like in the previous slide, you go Movie to MPEG-4.
And then you get another window that lets you choose your profile if you want to use simple, advanced, simple, complex, whatever the profiles are. You go in and do that. You set your video bit rate. You can set your frame size. You set key frames. You can do your audio bit rates. You can choose whether to do high quality encoding. You can choose whether-- well, actually, you can't choose whether to use the speech codec or the music codec. You can use the music codec. And then they have at the bottom, it tells you sort of what kind of file you're going to produce. And if you come up with settings that don't fit the MPEG-4 specification, it'll tell you. And it'll tell you how much you need to lower it by to get it to fit in the specifications, which is really handy, especially when you're starting out learning to do MPEG-4. So that's QuickTime Exporter going out of QuickTime 6 Player. This is the QuickTime Explorer Exporter going out of Final Cut 3. It's pretty much the exact same thing. You go Export, Movie to MPEG-4, and you have the same settings window. So the same way you did it in QuickTime Pro the first time you did it, you can now do it in Final Cut Pro.
In After Effects, export, movie to MPEG-4, same window. Seeing a pattern here, iMovie, export, movie to MPEG-4. Everybody ends up with MP4 files, and everybody has happy MPEG-4 compliant files. All right, with Cleaner and Squeeze, you get a little more advanced control over your video. Cleaner is sort of the industry standard compression program. It's been around for years. Squeeze is becoming, it was introduced a few years ago, and it's becoming sort of more widely adopted. Sorensen introduced their own MPEG-4 codec, so you're encoding video using their MPEG-4 exporter, not Apple's MPEG-4 exporter, and you get different results. Cleaner and Squeeze allow you to do cropping of your video, so if you want to take out the black bars at the top of the letterbox video, or you want to make something smaller, make something a really strange frame size like 100 by 4. You can use that to crop it out. You can do color corrections. Sometimes video gets muddy when you bring it into the computer, so you can brighten it up some. Bring it into the computer video color space so that it looks good on a desktop, just about as good as it would look on a television. You can do deinterlacing, which, again, makes it look good on the desktop because a desktop computer isn't interlaced video.
It's progressive. It does filtering to allow you to do more advanced color correction, batch processing, which comes in handy when you're doing 300 files that have to be due by Tuesday. It's a lot easier than going into QuickTime Pro and going export, export, and having to come back. You can set it to go. And if it doesn't crash, then you're good to go. For playback and delivery, basically you can create an MPEG-4 file that plays back on everything. You can create an MPEG-4 file that will only play back in QuickTime. You can create MPEG-4 files that will play back on phones that will also play back in QuickTime. And they all have different extensions. They have different characteristics, and they have sort of different ways to go about making them, and I'll go over a few of those. So you have file types. You've got MP4, which is an MPEG-4 movie. It's the native MPEG-4 file extension..mov, which is the QuickTime file extension, you can create an MPEG-4 movie with the.mov extension that is still technically an MPEG-4 movie..m4p is the iTunes Music Store file, but that's an AAC audio file, which means it's an MPEG-4 movie..3gp, playback on 3GPP phones.
One day they'll come to this country and we can enjoy that. But again, it's an MPEG-4 file. So you can play it back in MPEG-4 on your desktop. You can play back in QuickTime 6.3. All right,.mp4. It's the native MPEG-4 movie extension. It'll play back on all ISMA-compliant devices. The advantages to producing.mp4 files are its interoperability. That's why it's a standard. That's why we have it. That's why we encode for it. The disadvantage of it, or is it, specifically in QuickTime, it's not a fast-start movie. So if you do progressive content, progressive download, when it starts coming in and there's enough of it to play, it buffers in, it'll start playing back. Dot mp4 movies don't do that. You have to actually hit play. And that's kind of a bummer, because you can't always trust your user to hit play.
Sometimes they're not that smart. Sorry, users. But they don't always know they have to hit play. And they go, the movie doesn't work. Just hit play. MPEG-4 also has limited annotations on what you can do. You don't have the full range of QuickTime annotations with go to this URL for more stuff. I encoded the file. This is the copyright. All that jazz..mov files, it's the QuickTime extension.
It's meant for playback in the QuickTime player or the QuickTime plugin. The advantage of this, obviously, Fast Start. You do a MP4 file that you then save as a QuickTime movie. You can author it as a Fast Start movie on the web. It's a.mov file. Everybody that has QuickTime installed is going to be able to play it as long as they're using QuickTime 6 with the MPEG-4 codec. You have the full range of the QuickTime annotations, so you can do encoded by, you know, full name, copyright holder, all that jazz. You can also put in other QuickTime media types. So you can take advantage of things like still images because MPEG-4 spec under ISMA as implemented in QuickTime doesn't allow us to do the still image stuff and all that, but you can do sprites using an MPEG-4 movie, or based sort of around an MPEG-4 movie. You can make a media skin and put your MPEG-4 movie inside it. Unfortunately, it's not a.mp4. It's now a.mov, but you still used MPEG-4 as your codec, so it's an MPEG-4 movie. With me? Okay. The disadvantage is you lose interoperability. So it's an MPEG-4 movie, but it's not an MPEG-4 movie. Does that make sense? Yes? No? Sorry. It's at m4p. You have the... mp4p.
iTunes music store format. It's an iTunes protected file. It's locked. You can't take the ones off my laptop and play them on yours. They're all delivered through the Apple Music Store. The advantages of it are that it's digital rights management. So you can't take it off my laptop and play it on yours. It's got 128 kilobit AAC audio, so very high quality audio playback. The stuff about Dolby Labs did the testing, and you can't distinguish this from the source audio.
So you get good quality audio. You can also take that, and if you legitimately bought the iTunes file, you can import it into iMovie. You can import it into iDVD and use it for those projects, which is kind of cool because you bought the file, so you might as well use it wherever you want. The disadvantages, depending on how you butter your bread, are digital rights management, and some people don't like that. Yep.
The Apple Music Store files? I believe so, yes. The preview files I don't think are 128 kilobit per second, but the files you buy from the iTunes Music Store I believe are. The 3GP,.3GP files, they're QuickTime movies that will play back on 3G phones. They also play back in QuickTime 6.3 on a Mac.
There's the standard for playback on wireless devices. They've got audio, video, and 3GP text as implemented in QuickTime, which is really cool because you can actually stream text with video. And a lot of people need to do that for people who use closed captioning. If you want to send an English file to a foreign language, you can have subtitles. It'll stream wirelessly to a phone. It will download to a phone and play. But unfortunately, 3GP isn't really available in the US, so none of us get to play with that. As Frank Casanova said, "Go to Japan and sit in the airport." All right, demo. May I please have the demo machine? Any questions so far before we carry on with John's demonstrations? Everything's pretty straightforward so far.
It's a full application. Sorenson sells a version of Squeeze. It's Squeeze for MPEG-4. And it comes with the Sorenson MPEG-4 Pro codec. And that is the only way at this time to encode using the Sorenson codec. You can't just buy it like you could buy the old Sorenson video codecs. You have to buy Squeeze to get the codec to encode the files. So that's how they kind of get you to buy their soup-and-nuts product. OK.
You ready for a demo here? OK, on with the demo. Basically, I'm going to try and walk you through start to finish capture file to produce a file that can be uploaded to the internet as a streaming MPEG-4 file. So you start out, and you get a source tape from-- well, I get them from clients.
And the first thing to do is you have to import it into the computer. We use Final Cut Pro, Apple's gold standard video importing and editing application. It's a great program. We've been using it for years. So the acquisition stage is you go into Final Cut. Fortunately, I have a file that's-- yeah, yeah, yeah, reconnect.
Unfortunately, I have a file already here. We don't have to capture it. But getting the file into your computer is kind of the easiest thing to mess up. If you bring bad quality source file into your system, you're not going to be able to encode a high quality file. The sort of mantra has always been garbage in, garbage out. So always try and get the files or the source tapes from your clients or produce your files using the best quality things you can get. Use beta SP, use digi beta. I really like to try and get things exported directly out of either After Effects or Final Cut on a hard drive because there's no degradation of the image quality there. So we've got this file, which is a-- this came to us on a DV tape from a client. It's kind of a music video promo. Making L prints.
steps is committing. Something you may notice, lots of red. DV doesn't like red. So I kind of chose this clip for a reason in that it doesn't encode very well. And so instead of showing you the perfect file and this is great and look how great it is, I want to show you that not everything looks great.
And a DV file with lots of red encoded to MPEG-4 isn't always going to look great. And at the end of this, we'll see that. But this lots of red has kind of pixelation and blurring at the edges of it on the source file. So I'm definitely going to end up seeing some of that in my MPEG-4 file. So there's really not a whole lot you can do to stop that kind of stuff, especially if your clients give you stuff on DV tape. So we've got this. Lots of high movement.
Mr. Local Hero Sammy Sosa. So yeah, lots of high movement, lots of strong red colors. So basically, you're encoding "Nightmare" file, right? So this was brought in, final cut. and captured to the hard drive. So, final cut. You then open-- for this sake of this demonstration, we use Cleaner 6, which is able to encode a.mp4 file. So I take my file, bring it in to Cleaner, You can see the whole file. You can set in and out points. Unfortunately, this clip has actually got the in and out points right where I want them, so I don't have to set in and out points.
Then you open up, you get a setting. Since this is a streaming file, when you do streaming, you don't really have a whole lot of flexibility as to what bit rates you can use. It has to be under a certain bit rate or else it won't stream. So, you know, say for a 56K modem, you know, you need to encode at, you know, kind of around 30 kilobits per second for video because you tack your audio on there, you tack your 10% bandwidth tacks on there because when you encode a streaming file, you always kind of leave 10% for the hinting and for-- I guess network trouble. So you're always kind of locked into similar bit rates. Doing an ISDN file, you need to come in under 100 kilobits per second.
Typically, for doing broadband files, you want to come in at or around 300 kilobits per second. A lot of broadband connections get faster, but to sort of be nice to the slower connections, you do around 300. You can do a special 700 one sometimes, or even megabit streaming for the right connections. But for this demo, we just do around 300. So we have a setting. I'm outputting to MPEG-4. It's a.mp4 file. And if I click on this, we get our handy-dandy Apple thing. So Cleaner actually uses the Apple encoder.
image i'm cropping this is a 300k file i'm cropping the image to kind of near tv safe by taking 32 pixels off the left and right 24 off the top and bottom it you know kind of gets rid of any nastiness that you may have at the top and bottom of your video file sometimes when you bring in files from tape it kind of distorts around the bottom or the top and so by cropping your video, you're reducing that. You're also kind of reducing the size of the image so that when you encode it at a smaller size like 320 by 240 or 240 by 180, you're getting more of the actual real video image into the encoded file than you would at this size. And you're not really cropping out much more than people lose when they watch television. So it's a kind of safe way to crop files. Encoding this one to 240 by 180, it's a kind in a fast moving file like I showed you. So I chose a smaller frame size because it-- It kind of behaves better as a smaller frame size. As a larger frame size, you see some more pixels on some faster-moving places. All the footage was shot on television, so I'm deinterlacing, and Adapted Noise Reduce is just kind of something I leave on. Adjusts, you know, typical. You chuck the gamma up. You adjust the brightness and contrast to kind of make it look better on a computer, basically. And these brightness and contrast settings, If I go and apply them to this file and then do a preview of it, They kind of make it a little bright and washed out. I'm a Red Sox fan, so we get Pedro. And see, he gets kind of washed out and kind of nasty.
you can go in, bring this settings down some. You get a little richer color. He's a little less washed out. The crowd's a lot less washed out. I don't know what that looks like on here, but it looks better on my screen. You know, encode. You know, 15 frames a second for a 240 by 180 file at, you know, 300 kilobits a second is generally going to look pretty good.
You know, keyframe every basically 100 frames. So every 150 frames. Data rate to 300 kilobits a second. Audio. You notice there's no way to choose your audio bit rates here. But you can say 16-bit stereo, 16-bit. 44 kilohertz stereo. The way you choose your audio bit rate is you have to go back into the QuickTime encoder. You go to the output thing, go to options.
Audio. Now I can pick my thing. So stereo. It won't let me go any higher than 22. If I go to mono, I can bump it up some. And that generally actually sounds better than stereo. Sometimes. Depends on, again, the source file. Depends on whether it's speech or music. Encoding quality, I always choose best. 32 kilobits a second on top of my, just about 300 kilobits per second file will give me a file that will stream over almost every broadband connection in the US. So you choose apply for that setting. You kick it out.
And you, again, watch the progress bar creep across the screen. Favorite part of my day because it means I get to go play Xbox for about seven minutes. For the 15 frames per second, particularly for the broadband stuff, will work very well for all of the video footage that you have. But if you're encoding any animation, you may find that you're going to have to start bumping that up because on any animations, particularly things out of After Effects, you'll start finding that 15 just won't preserve the full range of motion. So bringing it up to as high as 24 helps. Anything above that, you tend to just start-- you're just tacking on extra frames that are probably going to get lost by the end user anyway as they're playing it. Like every good cooking show, I already have one prepared. So this is that file encoded using the setting that I had.
Once it started getting exciting and everybody started hitting home runs, it started to pixelate and kind of look nasty. And unfortunately that's because the bit rate that it was limited to for streaming, it did that. With this particular job, I talked the client into letting us deliver a downloadable file because it's only a minute long. It didn't produce a very big file. And by being able to up the bit rate, I was able to make a file that I was happier with in the thing. But, you know, as far as streaming though, it doesn't look all that bad for, you know, a good 90% of the time.
on a computer screen is still pretty fun to look at. Yes? I'm sorry, can you please go into the microphone so that translators-- The fraction between the audio-- Do you have some kind of standard setting? Kind of standard setting for that? No, honestly, I don't have a standard setting for anything I do.
Basically, every video is going to be different. So something that is a sort of spoken word, just talking piece, you can take the audio data rate down and you can bump the video data rate up some to get a little bit more quality. But generally, no. I don't have a specific way I set where I do it. Yes?
For the translators, he asked if I went with mono sound on that instead of stereo to get an audio quality. No, that was actually stereo sound. I didn't go with mono on that one. I did that this morning just playing around, and it was generally good sounding. I mean, whenever I do files, I wear headphones so I can sort of really hear what the audio sounds like. And then I take headphones off and I play it back over my regular Harman Kardon sound speakers. And-- Listening to it on the speakers forgives a lot more sins than listening to it on the headphones. So if you're really nitpicky, you wear headphones.
But generally, people aren't going to be doing it over headphones, so you can get away with mono. Do I have a 100K version of this file, is the question. No, but we can make one very quickly. The other reason I chose this clip is because it is short. Take these down to get the washed out out of it. This one I'll do mono.
Any other questions while we are watching the progress bar slowly creep across the screen? Or does anyone have an Xbox? He asked about input formats, what's a good input format. Like I said, this file came from DV, so I imported it via FireWire using Apple's DV codec. But generally, when we get files in on beta tape or digi-beta tape, we bring it in using Kona SD card.
Yes. So it's their either 8 or 10-bit uncompressed codec. And I've had a lot of happy results with that card. It's a nice product. They just put out this year a Firewire interface with SDI, and it uses the Kona SD and the 10-bit and the 8-bit Kona uncompressed codec. So I'm hoping to get my hands on that at some point to play with that because if I could do that on my laptop instead of having to go sit at a station for digitizing, that'll make life a lot more fun. But yeah, I like the Kona stuff with the uncompressed. With Final Cut 4, Apple introduced an uncompressed 8 and 10-bit kind of beginning codec. So people will start using that and will start seeing results and how that pans out. Is that-- Mm-hmm. I've also done MPEG-4 encoding where instead of doing-- I chose to just take the source file and do the 300 and the 100 out of Cleaner here because it's faster, and it produces a kind of fast, pretty decent workflow, pretty decent product. A lot of people will take and do all their preprocessing to one file before they encode it. And we've done that for a lot of files. We especially did that before Cleaner supported MPEG-4. So we would go to something like PhotoJPEG as an intermediate file, or go to-- I did PNG files, because they're pretty good quality, pretty smallish file size. So that's kind of an intermediate from where you capture to where you go to MPEG-4. I've done PhotoJPEG and PNG as-- - Yeah, if you start with a PhotoJPEG at the largest size that you want the overall media to be, So if it's a 320 by 240 or a 400 by 300, and you just start with that as your intermediary file, you'll get pretty good results. We were very pleased with that last year. We did experiment with the PNG as well, and it was sort of a situation of, to us it looked better, but it was also a little slower to export. - Okay, this is the 100K version. It's a 180 by 132, I think. Mono sound. This one I believe is 10 frames per second, so it's gonna be a little jerkier, but this one actually looks pretty good.
I'm making imprints In two steps It's commitment In three steps I'm not done yet Draw my other leg up And the pace is set And the... Considering the time we put into encoding this file, all of us here this afternoon just now, it's not a bad result. If you sit down with the file and tweak it and play with it, you're going to come up with a pretty good looking result. Also always keep in mind that at lower bit rates you're not going to get Apple movie trailer quality because this is streaming and that's not. And at 56K you're lucky to be getting video. Do the encoders get that, or the translators get that question? No? - Yeah, we need you guys to get up to the microphone to ask the questions. The translators can't hear the whole thing. Thank you.
When DVDs are being professionally mastered, they'll obviously do different trade-offs between audio and video, variable bitrate, and then size of the bitrate at various times in the file. I'm asking whether you do for streaming or whether these tools support for MPEG-4. You know, if I have the matrix, you know, action sequence, and then it goes to a dialogue portion and with music where I can trade off between audio video or even do more bit rate at one part and then less bit rate at another? The simple answer to that question is yes but the involved answer that question is still yes but kind of a Not why would you do that, but it's gonna take a lot more work.
For streaming, you pretty much wouldn't want to do that because you're always gonna be at a certain bitrate, and you -- you know, if you can have your talking parts looking as good as your, you know, high-motion parts, then, you know, that's just a benefit. You're not saving anything by going at a lower bitrate. For downloadable content, yeah, you will save.
especially on the file size, which is what is the sort of holy grail for downloadable content. So yeah, you could break it up and have the talking bit encoded at this bit rate with this sort of setting, and you can have the fast action bit done at this bit rate with this setting, and then you basically can combine them all in Final Cut Pro and export a.mov out. I don't think you can do it and produce a.mp4, a legal mpeg-4 file, but you can do it all using... You can?
Oh, yeah, set it to pass-through. Thank you. You don't get transcoding. No, you don't get transcoding. Actually, that's how-- say you do an MP4 file out of Cleaner, it doesn't hint it. They have a setting for hinting, but it actually doesn't hint the file. So this file-- that I just did is not hinted. There's no hint track in there.
So what I do when I have to hint these files then is I go into QuickTime Player. You open it in that-- You export movie to MPEG-4, options. Bless you. Like Mr. Waggoner gratefully volunteered, pass through, audio, video. So you're not getting any transcoding. Streaming, optimize for server. OK, save.
This very quickly kicks out a file that is now-- whoops, got hinted track. So this is now a streaming file that you put on your streaming server and it delivers. So yeah, to answer your question, yeah, you can do it as an MP4 file. You can do it as a.mov file. So yeah, you can save while doing your-- your bit rates. One of the questions you said, we just did that real quick.
Obviously, if you had more time, On making this better? Probably taking the audio down a notch or two with the bit rate, upping the bit rate on the video. Maybe trying to mess with the frame rate to -- because obviously more frames per second at a bit rate gives you a less good-looking file than a less frames per second at a given bitrate because you're throwing the same amount of bits at less frames than you're throwing at eight frames a second then you're able to throw more at it than at 12 so you know changing frame rates does it uh if you get really kind of in a jam especially with like the reds on that file i had a huge huge problem with and uh you know you can go in and try and take the red down you know which will give you a little less blockiness when those things come up. You can go in and change the brightness and contrast a little to try and give yourself a clearer-looking file as it goes out. So those are kind of the next steps.
And then the step after that is you just kind of go, hey, it looks good enough, and then you deliver it. I'd probably add to that maybe aggressively cropping more. Depending on the footage, the ability to, if the action's in the center of the screen or if it's focused in a given area, I'll start aggressively cropping the outside frame more, scrub through it quickly and see if I'm losing the action or the focus of it. So I'm devoting all of my data to the core action of the clip. I've not had a situation where somebody's come back and said, "Man, that was really cropped." Instead, it's been, "Man, that looked really good." More questions now, please.
Are there any more questions specifically? Sorry. If you have a question, please just feel free to hit a microphone. There's a couple. There's one there, there's one there, and there's one over there. share any experiences you've had with some of the other MPEG-4 codecs like Thrivex, DIVX? Maybe if you've used Fink, you could try some of the Xvid implementations that have been out the last few months. I honestly can't comment on that. My experience with DIVX has been watching things that people have showed me on the computer that they downloaded from nasty sources. So I've not encoded to DIVX. That's just my theory, but it's very useful.
Yeah. I've not encoded to DivX for client delivery files because my clients either go through Apple with their delivery mechanism and we have to deliver the files that Apple needs. you know, clients that I, you know, work with on our own without Apple who, you know, we try and steer them in a certain direction that will try and keep them sort of in line with what Apple is doing so that the content is all going to kind of look the same, behave the same, and act the same way. So I'm sorry, but I can't. man: I would not have that, um, 80-person, different impact for implementation in the next year perhaps might work with each other? Yeah, it's what I'm hoping. Yeah. Yes, sir. Hey, way to go. We love it. Thank you. Can you please go to the microphone? It's for the sake of the translation. I really want to find out what I sound like in Japanese. I think we all should.
on JPEG and using the 8-bit and 10-bit codec. Do you see any extraordinary differences when you're knocking things down in the encoding world when you start with an 8-bit file versus a 10-bit file? That's the first question. And then the second question is, have you started experimenting with HD source files yet? And what encoder choices-- obviously, there's Cleaner and Compressor and a couple other things-- can even look at HD files and accept them? The HD we haven't really had experience with yet, though.
we can't wait for the client that wants us to play with it because that's how we most often get to experiment with it. - That's how we get new hardware. Oh, we have to buy something for that? - The, as for the 8-bit versus the 10-bit, the, The main difference that we've seen is on high contrast stuff. So we have one place that's a production company, Eyeball, that we do a lot of work with. They do a lot of graphics-based 3D stuff. It's all flat graphics, but it's also really high contrast ratios. They like to push the envelope. And you start seeing a muddy quality to your black levels that you don't see. You see a muddy on the 8-bit that you don't see with the 10-bit at times. So there definitely is a value difference that we've seen. And I think my personal opinion is some of it's perceptual on the client's part, but I have actually seen it on even just metered looks at it. So there is some slight differences for that. But again, it's really got to be high contrast differences, and it's really mostly when it's graphics, so that there's really a far-field color value difference. With video, captured video, you can't tell that much of a difference. The human eye is sensitive, but it's not that sensitive. And when you're going from either 8 or 10-bit source to MPEG-4 or Sorensen or MPEG-1, 8 or 10-bit doesn't make all that much of a difference in kind of a real-world implementation. Yeah, I was just curious if there was any validity at all. There's sometimes, yeah, there's obviously going to be validity to almost everything. So yeah, but go 10-bit. OK, are there more questions? Yes. Oh, I'm sorry.
It's the name of the session, man. I do quite a bit of compression, I found that the quality of the MPEG-4 code in QuickTime doesn't seem to quite match up to Sorenson 3, that professional edition that's, you know, using QuickTime 5. I still use Sorenson 3. Professional edition. Is Apple's MPEG-4 implementation going to be improving in the future? Well, I think the, if I may, the important part of that was it was Sorenson Video 3, and a lot of times we consider this MPEG, Apple's MPEG-4 one. So, I mean, it's really, it's got lots of room to grow, and, yeah, there's room for improvement, But I think that improvement's coming. And I don't see a reason not to support a standard.
We still do a lot of Sorenson video as well for given situations. I think right now we're at a point where it's, depending on the type of content that it's going to be, dictates somewhat what we might choose. But that's the only thing. We don't just pick Sorenson video because we think it looks better. We think it might look better in some situations right now, but I wouldn't say that in 12 months or 18 months that that's necessarily going to be true.
You have a movie trailer site that I think Xtor still uses, starting with 3. Yes. Right. I haven't quite seen, you know, the great big MPEG-4 downfall from Apple. Well, if you look at a lot of the content on the actual QuickTime What's On page, the music videos, those sort of things, a lot of those are done in MPEG-4. That's a good question. They're done in streaming MPEG-4. And I know that because I do a lot of them. And so there is a transition. The movie trailers and things like that, you know, one day MPEG-4, we'll see.
That's the kind of question that you can bring up at the QuickTime feedback session. Which I believe is exactly. We actually are in transition. And on the MPEG, I mean on the MovieTrader site, stay tuned. Yeah, one thing actually that was brought up this morning. Sorry, Dennis. There's a standard for the QuickTime or the MPEG-4 decoder. There's not a standard for an encoder. So as people write better and more efficient and kind of... Nicer I guess encoders you're gonna be seeing better quality and peg for files You know because they'll have to play back in the same decoders that we have but as people sort of get used to writing for the Format, you know you are gonna get better encoder. So yeah, it's a transition. It's gonna happen Everything is gonna happen eventually to move along the sort of mpeg-4 thing because we all you know, it's it's like it's a standard That's that's what we want to get behind and push The important thing to remember about a standard is 10 years from now, the MPEG-4 files on Apple's site will still play. 10 years from now, you can't guarantee that the Sorenson 3 file is still going to play. So we think that going forward, it's a good idea to move to MPEG-4.
Thanks. Any other questions? Yes, sir. If you could use the microphone. With fees involved? If that committee has its way, I might be behind the times, but is that still an issue? With fees involved? In PEG-4. There is actually a fee structure in PEG-4, and we're actually lucky here at the WWDC to have Larry Horn. He's in one of the sessions. Thursday morning at 9.
Thursday morning at 9 o'clock. He is the man to talk to about fees. Should that be scary for developers, or what's your opinion? Well, you know, there are fees to use Windows Media as well. Like what? They came out with a licensing structure a few months ago. Structure, media.
just like MPEG-4 does. That was widely publicized? Yeah. Okay. OK. Probably not as widely as the MPEG-4. What's going on? But it's an evolving standard. This is MPEG-4. Like you said, it's been out for us for about a year. If you look back in the history of any codec or any file format, we're in 1.0. We'll soon be in 2.0, then 3.0, then 4.0. Any other questions about encoding in general? Can we please also have the slide machine?
Yes, sir. I was curious. We're using right now the squeeze for the reason of quality at this point, and also for batching. But using, let's say, the QuickTime Pro, is there any kind of batching functionality that I'm unaware of? DAVID J. QuickTime Pro? Yeah. DAVID J. Not-- Or through-- Through Final Cut, you can get patching. Is that, with Sorensen, it's basically one clip, and you can go to, like, several different hit rates. Is there a way to do multiple source files to multiple outputs?
To multiple outputs? Basically, yeah. Let's say you're using Final Cut Pro in there. Well, Apple kind of changed the thing after I handed in this presentation by launching Final Cut 4 with Compressor. So Final Cut 4 now comes with Apple's compression program. And I wasn't able to do the demo using that in this presentation. I didn't really have the hands-on with it that I wanted to be able to talk about it. But it is possible with it. It is possible with Compressor, definitely, to do batches, multiple clips, multiple settings. It's like using Squeezer Cleaner. You have the ability to queue up sequences, raw clips that been marked and output multiple versions. So out of something like Final Cut Pro, you save everything you want to do as a unique sequence. You do those all, and then you go to batch export them. And you can batch export them to not really multiple settings, but you could do your 300 run, and then you can go in and do another batch export of your 56k run, that kind of deal. So Final Cut will enable you to do batch export using the QuickTime exporter without having cleaner squeeze. But definitely, if you're in the market to update, Final Cut 4 is-- they did really-- CHRIS YING: It's a good choice. DAN COFFEY: Yeah, really good job with that. One of the limitations of the MP4 format was the annotation support. Yes. What tools are there for the limited annotation support that is there?
for the limited annotation support. Honestly, don't know of any tools. The way I've done it is I've tried putting in annotations before I export something in QuickTime Player and see if it makes it through. So things like name can go through. Things like that. Honestly, I've only done it through trial and error. There actually is a trick you can use with QuickTime. Are you familiar with ref movies? So you can put annotations in a ref movie, and inside the ref movie called to it.mp4, and if you don't, and then the ref movie can actually use its annotations that show up when you play the mp4 movie.
Yeah. Anytime I've had to do something with serious annotations where clients demanded the copyright in the artist's website and all that, I've just taken the mpeg4 file that I put out and then do, oh, my screen's not up anywhere. I do export in a movie to QuickTime movie, and you pass through the video and audio so you've got no transcoding, and you can put in the annotations you want to do.
Not after the fact, not really, no. It's more function of the spec than it is any limitation of QuickTime. Yes, sir? This might seem a bit ignorant. I don't know. Is there any way of making real-time encoding for the audio, just to AAC? Or is this a question for core audio? CHRIS BROADFOOT: Things like QuickTime Broadcaster, so real-time as in live streaming. I was actually thinking about the use of game audio for real-time encoding and decoding of all these streams, like input.
I mean, taking a real-time screen, encoding it, setting our network, and then decoding it again. DAVID MALAN: Well, that's what iChat now does, is real-time encoding in-- Is it using MPEG-4? DAVID MALAN: I believe it is using MPEG-4. But that's probably a good question for Frank Casanova. DAVID MALAN: That's a good question for Frank Casanova. No, we just got it yesterday, so we've been playing with it in our off time.
Ask us tomorrow. DAVID MALAN: Yeah, ask us tomorrow. But no, I do believe that it's MPEG-4. It seems to me that Apple would want to use MPEG-4 in that because everything else they're pushing is MPEG-4. It could very well be H.263, though, which-- It is H.263. It is H.263. No, thank you. There we go. Which is basically MPEG-4. As MPEG-4, yeah. MPEG-4 short header. OK. Sure. But yeah, something like QuickTime Broadcaster, you can do live encoding of AAC audio streams.
OK. If there are no more questions, I'd like to thank you. Oh, I'm sorry. Please, if you have a question, please just hit a mic. Don't be shy. Well, don't hit it. There's a lot of talk about the scalability of the I've seen have either been 3D, like this morning, or streaming, or downloads. How about stuff that's at least DVD resolution or higher?
either with Apple's coding or as you're saying in the future there may be better encoders? Yes, well there's something that was mentioned in the MPEG-4 session earlier this morning as well as at the QuickTime keynote, the AVC, Advanced Video Coding. And so that is kind of what is pegged towards the higher bitrate DVD quality files. And so Apple, of course, doesn't comment on future plans, but if they were to implement that, that's definitely the way to go. So other than that, yeah, you can do MPEG-4 files at monstrous bit rates, and you get very good quality. And we're even doing some experimenting with some set-top box playback and that kind of thing with it. So it's completely viable. It looks very nice, and it'll look probably even better in the future as it goes forward. I think the thing that we're liking about it so far is the file sizes involved, even for a really heavy-duty, high-quality, are still very attractive. They're not huge. And also, Apple's implementation of MPEG-4 and AAC doesn't tax the processor very well. It's optimized to run especially on a G4. So playing back a 640x480 MPEG-4 file with AAC audio plays back a lot smoother than a Sorensen 3 Cue Design Music Pro file did, because especially Cue Design is a processor hog. So doing the big things kind of pays off in MPEG-4 because you can throw the bitch you need to at it to get the video quality, and then you get benefits because of the low processor tax.
So yeah, it's definitely something, if that's where you're looking to go, throw lots of bits at MPEG-4 and see what you come up with. We've actually done a lot of demos at Macworld and NAB where the basic comparison we show is an MPEG-2 file encoded at, say, x megabits, four or five megabits. We actually do an MPEG-4 version of the same file of the same dimensions at half the data rate. And you're hard pressed to tell the difference between the two files.
Yes, sir. One more general question. Can we expect to see DRM on video like it is in the icon stores? I think-- Yeah, so-- Don't announce future products, obviously, but this Thursday we have a session, the feedback forum 011 with Frank Casanova is our director, the person asked about that. DRM's an interesting situation. I mean, DRM, there's a lot of different opinions right now about DRM. The person to really ask that question of is Frank Casanova in the feedback forum. We're more the how guys. We're the guys that actually have to take the files and then code. We're not much on the product marketing side.
If I knew, I couldn't tell you. And I just don't know, because they don't tell me anything. If he knew, he wouldn't tell us. Yes, sir? the Sorensen MPEG-4 compressor compared to Apple's compressor? Compared to Apple's compressor? I think that at times the Sorensen MPEG-4 compressor can produce better quality videos. I think at times the Apple compressor can produce better quality videos. Honestly, it's a matter of what the source is. It's a matter of what you do to it. And so I've had similar results with the two. I know that you can do side-by-side tests and come with this one does better and this one does better. But honestly, at the end of the day, you're getting pretty close to the same results. You know, Sorensen has done, you know, a little more extra work on their codec because it came out later than Apple's. So, you know, they have a little more under the hood. But, yeah, they're both very good, very viable MPEG-4 video codecs. Thank you.
And are there any more questions? If there's not, I would definitely like to thank you guys for your time today. It's been very good. In wrapping up, I think there was a couple of, a little bit of information Dennis has. - Yeah, there are a couple more sessions on MPEG-4. You guys all had this scheduled, but coming up tomorrow, there are some specific sessions on 3GPP in the afternoon at two, and then there's one at five. And then again, the real story about MPEG-4, visual licensing. This is the Larry Horn. Larry Horn is the president, director of a group called MPEG LA, which is the MPEG. VP of licensing. VP, thank you. He is the, sort of the, at the forefront of MPEG license, MPEG 4 licensing. So I'm sure he'd love to hear your feedback as well.
Yes. And again, please come to the 3.30 Thursday session, the QuickTime Feedback Forum. This is the place where you've got Frank Casanova, who heads up the product marketing team, but you also have some real key engineers on the QuickTime team. And this is an incredible opportunity for you guys to tell us things and for us to hear, you know, tell you things, what we can tell you, about what we think about different things. So please come. Let us know what you think. One more question. Please.
You said there was this new codec. Is it AVC? ADAM POWELL: Advanced Video Coding. It's part of the MPEG-4 specification. Is there going to be anything going on where we'll hear more about that? Probably 3:30 on Thursday. Thursday at 3:30, ask Frank Casanova. Would be the best. He's playing guitar. He's answering all the MPEG-4 questions. Watch his eyebrows. Watch his eyebrows.
Yes? I've been to the Centec 4 licensing session all day, scheduled for that room across the hall. You might want to think about moving it somewhere. Maybe we can move it to a bigger room. Yeah. That's going to be a good session. I'm going to be there. And it'll be in a bigger room. OK. Well, thank you. Thank you very much for your time. Thank you.