TMPGEnc 2Pass VBR Confusion

[kenmo]

I've resized both within VirtualDub and in Tmpgenc. I find the time to encode is similar.

[Anonymoususer]

I found 2 guides that give a pretty good explanation, I'm just gathering imformation right now, I'm gonna start trying them out over the weekend. DJ, I already have the manual....but thanks for the offer though. Pretty soon I'm gonna know CCE inside, out like TMPGEnc...also one of the guide sets 1 as the multipass option, I've been reading the 2 Pass for CCE is pretty good, BUT I'm gonna try 3 pass, just to see how much better it's gonna look, I'm gonna start experimenting over the weekend. I will posts results as they happen.

Thanks to all the people for the very very useful, and informative information. Keep up the good work Moderators.

[DJRumpy]

CCE behaves a little differently than TMPGenc. If a VAF file does not exist, setting the pass value to '1' will technically get you a two pass mpeg. CCE does not count the first information gathering pass used to create the VAF file. If you set it to '2', and this file doesn't exist, then you will actually have three passes performed on your AVI/AVS file.

[Anonymoususer]

could you explain a little bit more about the VAF file please..? and how to find out if you have it or not..?

[RabidDog]

Isnt some of the explanation of what is happening re tmpgenc 2 pass versus cq contained in the logfile. This contains lots of info and does refer to CQ even on the two pass mode. If I understood what it says i would explain it. my explanation of 2pass vbr vs cq would be that with 2 pass you set a final file size and tmpgenc will distribute the bandwidth fairly given that constraint whereas with cq it simply tries to use the minimum bandwidth to keep up a given quality but you cannot be certain of the final file size
or am I wrong ??

[Umen Pich]

Isnt some of the explanation of what is happening re tmpgenc 2 pass versus cq contained in the logfile. This contains lots of info and does refer to CQ even on the two pass mode. If I understood what it says i would explain it. my explanation of 2pass vbr vs cq would be that with 2 pass you set a final file size and tmpgenc will distribute the bandwidth fairly given that constraint whereas with cq it simply tries to use the minimum bandwidth to keep up a given quality but you cannot be certain of the final file size
or am I wrong ??

Exactly !
In 2-pass mode you can always be sure about the final file size.
Even if the movie as a whole requires more bandwith then the average, the final average will be that.
The scenes with more bandwith requirements will get more, and that with less will get less but the average will be what you specify.

[puertorican138]

2pass Variable Bitrate(VBR): This means that your bitrate has an minimum, maximum, and average bitrate. On this one you can set the average to a fixed setting and have it either raise to the maximum you set during high motion scenes or fall to the minimum you set when there is little motion. This setting is good for figuring out the size of the file that is being outputted and to have it reach higher bitrates yet still maintaining a lower file size. under the settings of this setting is where
you would put in everything you need to. There is also a setting, "Enable padding When falling below the minimum bitrate", if checked. forces the rate of the movie not to fall under the minimum bitrate you entered. Also this setting will go over the whole movie first then at 50% it will start to encode it, and this takes time...Allot of time! Why would anyone do this you ask. Well I do this because I know approx. what the file size will be and I feel that I will get the most out of my movie. This is mainly for MPEG-2.

Hope That Helps!!! :P

[puertorican138]

Constant Quality (CQ) is sorta like Constant Bitrate (CBR) in that it uses one pass encoding. It uses variable bitrate and I believe it can give you the highest quality with a very unpredictable filesize.

Hope That Helps!!! :P

[puertorican138]

I hope there's not a war between Constant Quality (CQ) and 2-Pass Variale Bitrate (2 Pass-VBR) again.

[Umen Pich]

I hope there's not a war between Constant Quality (CQ) and 2-Pass Variale Bitrate (2 Pass-VBR) again.

I hope so

if there is a war it is only to recruit newbies for the one army or the other

but no one from the old soldiers will step back

[Anonymoususer]

I was asking as to what is a VAF file mentioned above by DjRumpy. I'm gonna encode using CCE for the first time and I wanna get it right cuz I have a very slow PC and I don't want to waste like a day and not get the good results. After reading about 2 Pass VBR and Constant Bitrate, I encoded using 2 Pass VBR for TMPGEnc, and all I have to say is WOW !....It looks very incredible, considering I was encoding Fatal Fury The Motion Picture which is a ANIME AND had a lot of high action scenes. Just amazing. Very few blocks, but someone mentioned about that CCE's 2 Pass is even better then TMPGEnc and FASTER, I wanna see it for myself.

Started a lot of reading on CCE, will post more as I learn more.

VAF VAF VAF...???

[Anonymoususer]

Ok I found a very good guide for converting to Mpeg2's. on this website on Doom9

http://www.doom9.org/index.html?/mpg/cce-advanced.htm

It talks about AVISynth and how I paste the lines onto notepad ans change them. Im kinda geting confused as to what I'm supossed to change the path name/file, could someone put this in simple terms.

It;s the second step of the guide.

using AVISynth.

[paulb]

You can cut your 2-pass VBR encoding time in half by enabling the Cache setting under Environment. After the first pass, all data is pulled from cache instead of analysing again.

It looks like it takes as long, but once you hit 50%, it runs through the rest of the file in 20-30 minuts. At the highest quality settings, encoding that should take 7-8 hours is taking 3.5.

[kenmo]

I'm gonna be trying this setting in Tmpgenc (environmental->cache).

I was using CBR and now I'm a big VBR convert...I get better quality and a smaller file size

Gotta love it ! I can not imagine how Canopus Pro Coder or CCE could do better.

[Anonymoususer]

" I can not imagine how Canopus Pro Coder or CCE could do better."

Ok I don't know about Canopus BUT I do know that CCE is a very expensive software and you said that You couldn't imagine how those 2 could do better, even though I haven't tried out CCE YET..which I will today, CCE is almost as flexible as TMPGEnc AND if you've used the 2 Pass VBR option in TMPGEnc you will be surprised that you can do a 3 Pass VBR or maybe even a 4, If Im not wrong usually Big company's use 4 or above Pass to "refine" their movies, so I'm pretty sure CCE is 10 times better then TMPGEnc, not that TMPGEnc is bad, but obviously their is no contest. am I right or wrong ?

[DJRumpy]

DivXerouS, sorry for the delay, I've been swamped in projects this week. The VAF file is the file used to store the first pass data when encoding an AVI/AVS in CCE. That first pass is used to gather information about the MPEG. The second and subsequent passes refine the bitrate allocaiton, based off of the information in the VAF file. It is only created when you select a multipass option.

In regards to which encoder is better, most will agree that CCE is superior in both speed, and quality for MPEG-2. I used to believe that TMPGenc was, however, better at MPEG-1 encoding (VCD) in regards to quality. A recent VCD encoding I did for a friend has changed all of that. The quality has to be one of the best I've seen for VCD in a long time. I used the suggested MATRIX settings from per Custom Technology (Effect restricted vretical filter [16], and Quantizer characteristics (Flat part priority) set to [32]. Block scan order set to Zigzag and last but not least, the matrix was set to Ultra Low Bitrate), with my settings optimized for a progressive scan source, the encoded result looked excellent. This was on version 2.66 of CCE.

It's suggest you use 'Ultra Low Bitrate' for VCD (anything below 2Mb/sec bitrate), 'Very Low Bitrate' for 2-4 Mb/s is a good setting for SVCD, 'Standard' for most DVD's, and MPEG Standard for high motion (complex) DVD's. The other setting for 'Smooth' is for computer art/anime type encodings, or basically video with a lot of sharp edges.

Let us know how your tests go...

[Anonymoususer]

" It is only created when you select a multipass option."

so what your saying is that you CAN'T actually create the VAF file, rather CCE creates it for ONLY if you use the Multipass option, which I will, but in one of your posts you wrote

" If a VAF file does not exist, setting the pass value to '1' will technically get you a two pass mpeg. CCE does not count the first information gathering pass used to create the VAF file. If you set it to '2', and this file doesn't exist, then you will actually have three passes performed on your AVI/AVS file. "

could you elaborate a little bit on that. ?

P.S how do I use the quote option. ?

[Anonymoususer]

Considering you can create XSVCD's, or Non Standard Mpeg2's or whatever you wanna call it, could you create the same thing using CCE ?

[Anonymoususer]

Ok After reading the entire manual for CCE, I found some pretty ineresting things, did you know that their is a formula for converting the RGB to YcBcR or the Luminence level ? just amazing. Anyway my question is that I've set everything and everywhere but I'm having problem when I START to encode. I have installed Nimo Codecs Package, rebooted my machine and when I start CCE gives me the following error

" Couldn't Find appropriate video codec for DIV3"
" vinfo_open()failed:***could not open [Pathname of my file] + the VAF extension."

Now my question is how do I create a .VAF File...as DJ Rump explained above what it is, but how would I create it, I guess so CCE could read it and gather the required information about the AVI and start procesing...or get manipulated into thinking their is a .VAF file present...any ideas..looking forward to some answers... [Troubling]

[DJRumpy]

Your still missing a codec. From the looks of it, it's the DivX 3 Low Motion codec. It's not installed as part of the nimo pak. Search google for it.

Don't worry about the VAF file. It willt take care of itself.

To use the Quote option. Click the 'Quote' button above the text editor. You will get an opening tag for a quote. Copy/paste the text of your quote AFTER that tag. Click the Quote button again. It should place a closing quote tag after your pasted text.

In regards to your xSVCD question, it doesn't matter what settings you use. CCE just encodes given the settings you give it. If your source AVI is 480x480 and you set the bitrate to 5Mb/sec, then you've just created what could be called an xSVCD. CCE doesn't care what it is. It just encodes.

[Anonymoususer]

DJRumpy your a genius man, a genius, you got the answer for everything man...how long have you been encoding ? Also I downloaded the latest DivX5.03Pro Bundle for the DIV3 problem, hopefully it will work. Gonna give my PC a restart and post results as SOON as I start.

[Anonymoususer]

Ok it still says it can't find the approprite DIV3 Video codec...Im gonna try searching around the forums...

[adam]

Unistall the nimo codec pack its the most unstable package of software I have ever seen. Reinstall divx 5 and your problems will probably go away. If not, then go ahead and find the specific divx 3 codec and install it.

DJRumpy: Going back to what you said about what would happen if you used a low avg, thus giving the encoder a bad example of sorts for later bitrate distribution... It doesn't work like that. When the encoder makes the initial CBR encode it is only gathering information about the source, not the result of the encode. It looks at motion changes, complexity of images, patterns etc... that will be present regardless of what bitrate you encode at. It doesn't matter if your CBR encode is at 500kbits or 5Mbits, the information gathered will be the same. To put it roughly, the encoder basically makes a line graph prioritizing the given scenes. Regardless of how much bitrate you ultimately give it to work with, it will just attempt to distribute the bitrate where needed most, even if all scenes turn out as crap regardless. The easiest way to see this principle in action is to look in the advanced settings in CCE. Do a CBR pass at 500kbits and generate a vaf file. Go into the advanced settings and look at the bitrate curve. Now raise the avg to whatever you want and watch what happens. The entire line will shift but everything will remain in proportion to each other. The given bitrate doesn't affect how it is distributed, it just affects how much is actually distributed AFTER it is decided where bitrate is needed most.

As to a suggested benefit of CQ in this regard, its all relative. If using a low avg interfered with the bitrate allocation of the first pass of x-pass vbr, than it would interfere equally with the CQ encode.

A low avg does have a detrimental effect on VBR encoding but this happens regardless of what mode you encode in. With a low avg, you have a much smaller interval from which the encoder can gather bitrate. If your avg is 500kbits the encoder is not going to be able to go much lower than 300kbits on any given scene, so your not going to be able to go much higher than that either. So for the most part, you are almost encoding in CBR. CCE does let you tweak how aggressively it prioritizes scenes, though, so you could get some extreme results, ie: low motion scenes are unwatchable but high motion scenes look somewhat decent or vice versa.

[banjazzer]

As to a suggested benefit of CQ in this regard, its all relative. If using a low avg interfered with the bitrate allocation of the first pass of x-pass vbr, than it would interfere equally with the CQ encode.

You don't set an average bitrate with CQ mode. To get a bad result with CQ you would need to choose a low quality setting. Using (for example) a quality setting between 75 and 85 you are unlikely to get a bad result. However, you have no idea of the bitrate used until the actual encode. The argument is not that x-pass vbr is not capable of the best allocation of bits, but rather that it is not responsible for the total bits allocated. That figure is chosen by the user.

[Anonymoususer]

Ok I uninstalled Nimo Codecs Package, and I think you might be right about it being very unstable, because I had been noticing that all AVI's I download always play jerky, not in sync, etc you I usually don't watch any movies on PC just install codecs so the Encoder can process the file, I removed Nimo Codecs, Reinstalled DivXPro5.03, it STILL gives me the stupid DIV3 Error...still searching for the right codec...


--------------------------------------------------------------------------------------
I uninstalled DivXPro5/03 and Installed 3.11Alpha & It worked BUT now Im faced with a new problem I'm getting kinda pissed off..well as soon as it starts to begin this is the error I get from CCE

"cce_encoding_init()failed: unknown error"
"vinfo_open()failed***Could not open the file [Pathname ofmyfile.vaf]"
"The system cannot find the specified file"
" Windows error code: 0x2 2"

[Rips out hair]

[adam]

Banjazzer. With either mode you are free to raise or lower the bitrate, either directly or indirectly, according to the desired output size. As far as the final output size, in other words total bitrate used, its completely relative because it totally depends on your settings and more importantly on physical limitations. A CDR holds the same amount of data regardless of what mode you encoded in, ultimately we are all shooting for the same target sizes. As far as the allocation of those bits, it depends on the algorithms used by the encoder in that mode. Without saying which one is superior, there are obvious differences between the two but in comparing any encoding modes you have to assume they are using the same amount of total bits otherwise there is nothing to compare. If you choose a low bitrate in x-pass vbr there are going to be detrimental effects inherant in low bitrate mpeg encoding. You would get the same detrimental effects in CQ if you chose comparable settings. The lesson is to just use realistic settings regardless of what mode you use. If you use both modes effectively then the avg bitrate, and final size, should be the same because the goal is to just fill the desired number of disks without going over.

DivXerous do you have a vaf file already in the directory that CCE is outputting to? Maybe when it messed up those previous times it output a vaf file. If you are doing a new encode then you need to make a new vaf file. I don't see how you could get this error otherwise because CCE doesnt need a vaf file until at least the 2nd pass.

[Anonymoususer]

The VAF file is the file used to store the first pass data when encoding an AVI/AVS in CCE. That first pass is used to gather information about the MPEG. The second and subsequent passes refine the bitrate allocaiton, based off of the information in the VAF file. It is only created when you select a multipass option.

No I don't have a VAF file in the directory, I thought it created the VAF file itself, as mentioned above by DJRumpy, is their a disabling mechanism, or do I have to create a VAF file in a special way or something, and is their a manipulation procedure into creating it ? This is my first time using CCE so I'm kind of a newbie BUT Iam familiar with all of the settings of CCE....but I'm having major problems starting up CCE to encode...it's very frustrating...

__________________________________________________ ___________

Tried changing the path for the VAF output file, still no luck...

__________________________________________________ ___________

After reading some guides, I have learned that to framserve which I haven't or

If you have CCE 2.50, CCE 2.64.01.10 or CCE 2.66 you can use Avisynth instead which is a bit more complicated, but also much faster.

I loaded up the AVI directly into CCE without going through all the AVISynth procedures...could that be causing the problem.

Anyone ?

[banjazzer]

there are obvious differences between the two but in comparing any encoding modes you have to assume they are using the same amount of total bits otherwise there is nothing to compare.

I don't particularly wish to re-open this old argument, any more than I suppose you do. In practice, used sensibly, both encoding modes will produce good quality mpeg2. The approach is different with CQ, however, because you can decide to encode to a quality setting, and output to the number of discs determined by the final size of the mpeg. This might mean an extra disc than anticipated, and most likely some wasted disc space. The way we are attempting to compare them is however different, because ultimately we are not considering comparing both methods using equivalent bitrates. The average bitrate is the factor that separates the two modes.

This is all largely hypothetical, though, because I almost exclusively use CCE now because of the speed, and I agree it is satisfying to get a high quality encode, speedily, and at a predicted size. But for someone using TMPG, who might just be unhappy about the time taken for 2-pass vbr, and who might just get the average bitrate wrong because of a desire to fit a movie to 2 discs, then CQ mode has got a lot to recommend it. And TMPG's CQ mode is in no way inferior, quality wise, to its 2-pass vbr mode.

[banjazzer]

I loaded up the AVI directly into CCE without going through all the AVISynth procedures...could that be causing the problem.

That is unlikely, but why not try loading a different file just to make sure? What version of CCE are you using? What is your CPU, and how exactly are you loading your file into CCE?

[DJRumpy]

Adam, thanks. That's exactly the part I was having difficulty with. I wasn't seperating the method from the result. Is the averaging for bit allocation done only on a bits per/second basis, or cdoes the encoder spread the bits over something else (say a GOP for example). Worded another way, does the encoder have 2500 kb (example) to allocate to a 1 second scene, varying bit's, depending on frame needs, or does it spread this over a larger scale?

In regards to the CQ vs Multipass VBR, VBR would be superior, given suitable settings equivalent to CQ, simply because VBR would have advanced knowledge of scene changes, where CQ mode has to use prediction. If/when CQ guesses wrong, the scene quality degrades. This is a given, although the drop in quality may be unnoticable, the drop is there.

It sounds like this has turned more to a discussion as to whether or not CQ is a good compromise for TMPGenc users, wanting smaller size, while retaining 'good' quality. In answer to that, I'd have to say yes. Although I too use CCE exclusively, and can't answer for the quality a CQ mode mpeg gives in TMPGenc, I'm sure it's probably comparable to the CQ mode in CCE, which I have used before, with good results (uugh..talk about a long run-on sentence...).

DivXerouS, the error your seeing is usually due to previous encoding attempts, which in your case were failing. Even a failed attempt creates a VAF file, and possibly other output, depending on your encoding settings.

In your AVI's directory, you should have only the AVI, and/or your AVS script. Nothing else should be required. If you are using and AVS script, drop the script onto Windows Media Player, and ensure your video plays properly. It could be a simple script typo.

Also, in the Tools section to the left, look for a program called 'GSpot'. It will tell you exactly what codecs an AVI uses, and whether or not you have all of the necessary codecs installed to properly decode/encode your file. The particular codec you were looking for is not installed as part of the DivX 5 package. I had to install mine seperately as well. GSpot will confirm if your all set there. Also, remember that since you uninstalled Nemo, you may have also uninstalled your MP3 audio codecs. Many DivX, and XviD AVI's use MP3 audio. Be sure to install the LAME codec for MP3 support.