Compressor 2 and 2-Pass VBR?
- Thread starter ullanta
- Start date
ullanta
Veteran
Sorry. The problems are common and often discussed...
Compressor 2 is slower than Compressor 1, and has some peculiarities in 2-pass VBR encoding that Apple has acknowledged but not yet fixed.
Basically, in 2-pass vbr encoding, it's common for
1) The mpeg file to be significantly larger than the average bit-rate would indicate
2) The mpeg file to have far more blockiness than the same file encode by Compressor 1
3) Occasional strange artifacts (I'll try ti edit a link in later)
4) The encoding process to take an extremely long time
Switching to 1-pass VBR solves all these problems, but doesn't of course give the benefit of a 2-pass encoding.
There have been lots of suggested fixes, and workarounds for some parts of the problem (and some interactions with previous installations of Compressor 1), but nothing has yet worked for at least a significant number of people.
So, have you successfully used 2-pass VBR encoding in Compressor 2? How does it come out vis-a-vis one-pass or Compressor 1; and how much slower is it? 1.5 times slower would be the edge of reasonable; 4 times slower would be aproblem...
Compressor 2 is slower than Compressor 1, and has some peculiarities in 2-pass VBR encoding that Apple has acknowledged but not yet fixed.
Basically, in 2-pass vbr encoding, it's common for
1) The mpeg file to be significantly larger than the average bit-rate would indicate
2) The mpeg file to have far more blockiness than the same file encode by Compressor 1
3) Occasional strange artifacts (I'll try ti edit a link in later)
4) The encoding process to take an extremely long time
Switching to 1-pass VBR solves all these problems, but doesn't of course give the benefit of a 2-pass encoding.
There have been lots of suggested fixes, and workarounds for some parts of the problem (and some interactions with previous installations of Compressor 1), but nothing has yet worked for at least a significant number of people.
So, have you successfully used 2-pass VBR encoding in Compressor 2? How does it come out vis-a-vis one-pass or Compressor 1; and how much slower is it? 1.5 times slower would be the edge of reasonable; 4 times slower would be aproblem...
Blaine
Mr. Hollywood
ullanta said:
Okay. The only thing I used the 2-pass VBR for was a 27 minute short and yes it took a LONG, LONG, LONG time. After that I went to single pass just for time savings. I am going to try a 2-pass on the feature we working on, though. I'll just start it and go on vacation for a few days, maybe it'll be done when I get back.:shocked:
I've never used Compressor 1. I got the FCP HD 4.5 Production Suite before upgrading to FCP Studio so all I've ever known is Compressor 2. (That wasn't Compressor 1 on FCP HD 4.5 was it?)
At the time I did the 2-pass, I wasn't that concerned with size because room wasn't a problem, so I didn't pay that much attention, however, I'll be paying close atttention now. Thanks.:thumbsup:
Well, I had tried Compressor 1 for a good while to come out with a watchable video, but never made it, so I stopped using it and made a move over to Sorenson Squeeze, instead. During my experience with Sorenson Squeeze, I'd talk real bad things about Compressor.
Just a few weeks ago, I was asked, again, about Compressor by a friend who just bought and installed FCP Studio. I went over to his place to show him how to export a .mov file to compress with Sorenson Squeeze. After that short lecture, I decided to show him how bad compress was. However, Compressor 2 gave us a good quality video! So I told my buddy, *well, I guess they've done a good upgrade.* However, I later figured out that it wasn't Compressor's problem, but mine. The first time I used Compressor 1, I didn't really have proper knowledge on video compressing.
Anyways, there are a few things, if not more, to consider and tell Compressor to perform. Please correct me if I'm wrong or there's more.
1. Keyframe: this is basically how compression's done. Compressor compresses what's between two keyframes. For example, if you put 24, it will compress based on the movements in between every 24 frames. The smaller number you give, the better quality you will get, but the larger the size of file you will get. It's recommended to start with 10, and go from there.
2. Method of calculation: Either 1 or 2 pass, and either CBR or VBR. This is to tell the compressor how many times to calculate before actually compressing. There's no doubt 2 pass gives you better result. If there's more movement than static image on your video, it is highly recommended to perform 2 pass so that it carefully looks through your footage. CBR is recommended if there's less movement or no-movement at all. I, personally, recommend not to use it if there's any movement at all, either camera's or objects'. VBR, Variable BitRate, compresses based on how heavy the movement is at any given moment of your video.
3. Maximum size of data per second: Even though Compressor is supposed to does all the calculation for you, you should be the one actually calculates to make the best possible video. For example, if you're compressing for DVD, you should calculate so that it comes out to be less than the amount of data your DVD accepts. Also, there's a certain number that you shouldn't exceed, but I can't think of the exact number right now. I think it's somewhere around 700,000bps. Don't quote on me and look up on either FCP or Compressor manual. If you exceed that number, it won't play right. It's also one of the causes which makes DVD Studio Pro fail to burn.
These three things are the basic elements that you need to be aware of to make compression happen. To further make your video look the best (and who doesn't wnat to?), you should use Compression Markers on FCP during editing. By putting markers at the beggining and the end of where you see any heavy movement, you tell compressor to compress less so that there won't be any artifacts. You can do so by typing *<compression>* in where you make notes about makers by hitting keyboard *M* one more time after hitting *M* to put a marker. There are a couple of rules that I don't really remember, but it's like there should be at least two frames in between two compression markers, and so on. Also, if you bring out Compressor in FCP, Compressor automatically puts compression makers, which is keyframes, on every cut there is.
Please bare with my english since it's not my first language. I'm sure I forgot to mention something, so reply if there's any other questions. This should give you good results. FCP itself does a wonderful compressing job, too. Happy comression!
Just a few weeks ago, I was asked, again, about Compressor by a friend who just bought and installed FCP Studio. I went over to his place to show him how to export a .mov file to compress with Sorenson Squeeze. After that short lecture, I decided to show him how bad compress was. However, Compressor 2 gave us a good quality video! So I told my buddy, *well, I guess they've done a good upgrade.* However, I later figured out that it wasn't Compressor's problem, but mine. The first time I used Compressor 1, I didn't really have proper knowledge on video compressing.
Anyways, there are a few things, if not more, to consider and tell Compressor to perform. Please correct me if I'm wrong or there's more.
1. Keyframe: this is basically how compression's done. Compressor compresses what's between two keyframes. For example, if you put 24, it will compress based on the movements in between every 24 frames. The smaller number you give, the better quality you will get, but the larger the size of file you will get. It's recommended to start with 10, and go from there.
2. Method of calculation: Either 1 or 2 pass, and either CBR or VBR. This is to tell the compressor how many times to calculate before actually compressing. There's no doubt 2 pass gives you better result. If there's more movement than static image on your video, it is highly recommended to perform 2 pass so that it carefully looks through your footage. CBR is recommended if there's less movement or no-movement at all. I, personally, recommend not to use it if there's any movement at all, either camera's or objects'. VBR, Variable BitRate, compresses based on how heavy the movement is at any given moment of your video.
3. Maximum size of data per second: Even though Compressor is supposed to does all the calculation for you, you should be the one actually calculates to make the best possible video. For example, if you're compressing for DVD, you should calculate so that it comes out to be less than the amount of data your DVD accepts. Also, there's a certain number that you shouldn't exceed, but I can't think of the exact number right now. I think it's somewhere around 700,000bps. Don't quote on me and look up on either FCP or Compressor manual. If you exceed that number, it won't play right. It's also one of the causes which makes DVD Studio Pro fail to burn.
These three things are the basic elements that you need to be aware of to make compression happen. To further make your video look the best (and who doesn't wnat to?), you should use Compression Markers on FCP during editing. By putting markers at the beggining and the end of where you see any heavy movement, you tell compressor to compress less so that there won't be any artifacts. You can do so by typing *<compression>* in where you make notes about makers by hitting keyboard *M* one more time after hitting *M* to put a marker. There are a couple of rules that I don't really remember, but it's like there should be at least two frames in between two compression markers, and so on. Also, if you bring out Compressor in FCP, Compressor automatically puts compression makers, which is keyframes, on every cut there is.
Please bare with my english since it's not my first language. I'm sure I forgot to mention something, so reply if there's any other questions. This should give you good results. FCP itself does a wonderful compressing job, too. Happy comression!
ullanta
Veteran
This is all true, and good advice... the problem is that Compressor 2 in many cases doesn't obey, especially in regards to point 3... doesn't match your specified data rates, etc.
Your friend had a clean install of Compressor 2 the first time, eh? No previous install of Compressor 1? Most of Apple's attempted fixes have had to do with leftover prefs and such from Compressor 1... so these issues may not affect people who never installed 1...
Your friend had a clean install of Compressor 2 the first time, eh? No previous install of Compressor 1? Most of Apple's attempted fixes have had to do with leftover prefs and such from Compressor 1... so these issues may not affect people who never installed 1...
<<the problem is that Compressor 2 in many cases doesn't obey, especially in regards to point 3... doesn't match your specified data rates, etc.>>
Regarding point 3, after you choose any of the presets on Copressor 2, you can tweak little options in *inspector* window. I thought we could set the maximum Bit Rate there, too. I have Compressor 1, so I have to check it later. I apologize if I'm wrong.
<<Your friend had a clean install of Compressor 2 the first time, eh? No previous install of Compressor 1? Most of Apple's attempted fixes have had to do with leftover prefs and such from Compressor 1... so these issues may not affect people who never installed 1...>>
Yes, he did. He bought his G5 and FCP Studio at the same time, so he didn't have chance to install Compressor 1, I assume.
Regarding point 3, after you choose any of the presets on Copressor 2, you can tweak little options in *inspector* window. I thought we could set the maximum Bit Rate there, too. I have Compressor 1, so I have to check it later. I apologize if I'm wrong.
<<Your friend had a clean install of Compressor 2 the first time, eh? No previous install of Compressor 1? Most of Apple's attempted fixes have had to do with leftover prefs and such from Compressor 1... so these issues may not affect people who never installed 1...>>
Yes, he did. He bought his G5 and FCP Studio at the same time, so he didn't have chance to install Compressor 1, I assume.
ullanta
Veteran
janghos said:
You're correct. What I'm saying is, it doesn't work. You specify an average bit rate that should result in a 2GB file, and get a 3GB file, etc. Works fine for one-pass, just, sometimes, not for 2-pass. Plus, 2-pass often has glitches and artifacts, and sometimes takes an outrageous amount of time (e.g., 10x as long as 1-pass). I understand well what Compressor should do, and what it did for me in version 1. Some people, including me, have problems with compressor 2. Someone has just confirmed for me that he's seen the same problems on a clean install of Compressor 2 on a machine that never had Compressor 1...
This subject is more than worthy to further discuss, since the final outcome is what we deliver.
First of all, depending on the type of codec used and type of the source video, the data rate you get will always vary. Could you specify what codec you used? If I assume we're talking about MPEG-2, and if your picture has extremely grainy or high-motion video, it could and will produce substantially higher data rate than what you asked for. This is just a fact, no matter which version of compressor we're talking about.
Second, 1-pass vs. 2-pass.. As simple as it gets, it's just matter of whether to calculate the complexity of motion (redenduncy). 1-pass compresses as the video is being read, which means *no time to calculate.* It is not that 2-pass is supposed to take twice as much time as 1-pass takes. Again, depending on how complex your video is, how long 2-pass take to perform varies, but will take more than 1-pass.
I'm not sure why you're getting worse result with 2-pass than 1-pass. I assume it is possible for 2-pass to produce poorer result when extremely static video is given, which I don't think is your case.
Okay, let's say that Compressor has some bug until we figure something out. Nevertheless, if you want to perform the best compression, you should perform *preprocessing* like I explained in earlier post. By putting compression markers throughout your video after much careful examination of motion complexity, you will likely get the best result possible.
Still, 2-pass should not give you poorer result when used under proper settings.
I'm planning on running some test tomorrow on this issue. I will get back with the result. I think it will be helpful for you to compress another video and see how it comes out. So long.
First of all, depending on the type of codec used and type of the source video, the data rate you get will always vary. Could you specify what codec you used? If I assume we're talking about MPEG-2, and if your picture has extremely grainy or high-motion video, it could and will produce substantially higher data rate than what you asked for. This is just a fact, no matter which version of compressor we're talking about.
Second, 1-pass vs. 2-pass.. As simple as it gets, it's just matter of whether to calculate the complexity of motion (redenduncy). 1-pass compresses as the video is being read, which means *no time to calculate.* It is not that 2-pass is supposed to take twice as much time as 1-pass takes. Again, depending on how complex your video is, how long 2-pass take to perform varies, but will take more than 1-pass.
I'm not sure why you're getting worse result with 2-pass than 1-pass. I assume it is possible for 2-pass to produce poorer result when extremely static video is given, which I don't think is your case.
Okay, let's say that Compressor has some bug until we figure something out. Nevertheless, if you want to perform the best compression, you should perform *preprocessing* like I explained in earlier post. By putting compression markers throughout your video after much careful examination of motion complexity, you will likely get the best result possible.
Still, 2-pass should not give you poorer result when used under proper settings.
I'm planning on running some test tomorrow on this issue. I will get back with the result. I think it will be helpful for you to compress another video and see how it comes out. So long.