Could VBR ever be better than CBR??

[boathouse]

I'm using CCE to encode my home movies to SVCD format and can't decide if I should use variable or constant bitrate?

I thought that the only advantage of VBR was that the encoded files would end up smaller, but on the downside encoding would take a lot longer than CBR because of the multiple passes required.

I'm confused because I've read some comments/posts that suggest VBR gives better results?

If my VBR encoding has a maximum bitrate setting of 2520, and my CBR setting is also 2520, I can't see how VBR can be better, since with CBR you're giving it the maximum bitrate all the time.

Please can someone clarify this for me, which is the best approach?

[LeeBear]

VBR can give you better results then CBR because the extra passes allows the encoder to have better scene prediction so scenes with motion can look better.

-LeeBear

[boathouse]

I don't dispute what you're saying, I don't know the answer to my question.

However I thought the multiple passes in VBR enabled the encoder to analyse the video source, so that it can then allocate bits in the most effecient manner, saving bits from frames that don't need them.

If CBR is running at max bitrate (say 2520) all the time, that's wastefull but surely will produce as good a result as VBR with a maximum of 2520.

..or maybe I'm missing something else?

boathouse you are correct, if you do a VBR and CBR encode and the CBR bitrate is the same as the max of the VBR, THE CBR ENCODE WILL ALWAYS GIVE BETTER RESULTS.

CHEERS

[sjm817]

You didn't say what length video you are trying to fit on what size medium. I'm also encoding my home movies to SVCD, but using TMPGEnc. I am putting 1 Hr of video on each 80 min CD. CBR in this case would not produce good results due to the low ~ 1600 bitrate for that size. With 2 pass VBR, I can get 1 Hr of very good quality video on the disk. For what I am doing, VBR is by far the best method.

[boathouse]

Yes I agree, if the size of the file matters then VBR is best.

I don't mind getting 35-40 mins per cd, so a constant bitrate of 2520 will be fine.

I want to be sure of getting the best quality so that's why I was keen on CBR. Also it won't be long before rewritable dvd becomes affordable so that will make storage easier

Have you ever considered using the CVD standard for your home movies? Basically they are like SVCD but 352x576/480 with the same max bitrate of 2520. I understand they are compatible with dvd and so will be easier to write to dvd in the future.

[mickboss]

Boathouse:

CBR at 2520 and VBR at maximum 2520 will be almost the same!

THE SECRET is to use VBR AVERAGE 2520 and maximum like 3000/3500!!!

CBR 2520 and VBR AVERAGE 2520 WILL produce files with almost the same size. BUT VBR will alocate 3000/3500 in scenes with fast motion and less 50/300/500 in scenes with slow motion. THE AVERAGE will give 2520!!! IN CBR it will waste bitrate in slow motion scenes and will limit fast motion to 2520.

More VBR passes will improve the prediction and bitrate distribution (more than 5 passes will not improve) 8)

[boathouse]

Sounds like a good idea, but if the video exceeds 2520 at any point doesn't that mean it will be outside the svcd/cvd standard.

I don't suppose it matters, since one day I hope to burn my movies to dvd anyway.

So if I encode at 352x576(cvd) and the bitrate exceeds 2520, will I still be able to create dvd's with these files without having to re-encode?

[sjm817]

The CVD looks interesting, but I'm almost done with the project. I've made 18 SVCD's and have a couple left to go. I'm quite happy with the way they turned out. Quality is very good, plays well on the home DVD player.

Since the audio from the 8mm camcorder home movie is not that great to start, I encode using 96 K for audio. This leaves 2620 for video. For 1 hr video, I use 1680/2620/300 Ave/Max/Min. If I play the MPEG2 file with Power DVD, I can see the bitrate as it plays. It is usually 1300 - 1800 or so, but it often shoots up to 2000 - 2500 when there a lot of motion.

My new camera is mini DV. I would like to start recording the new movies to DVD. Just waiting for the prices to come down a bit and the standards to shake out some more.

[adam]

boathouse you are correct, if you do a VBR and CBR encode and the CBR bitrate is the same as the max of the VBR, THE CBR ENCODE WILL ALWAYS GIVE BETTER RESULTS.

CHEERS

This is not true, as others have already stated in this post. It is true that with a VBR max set to the same level as the CBR level than at any given point the VBR encoded file will never have more bitrate than the CBR encoded file. However, there is more to quality than just the TOTAL amount of bits used in encoding.

With a decent encoder and an reasonably high avg the bitrate allocation will be good enough that the extra bitrate in the CBR encode won't translate to higher quality. Basically what I'm saying is that the scenes that actually need 2.5mbits will still have it in both encodes, and in scenes that need less the CBR will just be wasting bits. As far as bitrate and bitrate allocation are concerned the two encoding modes are about equal at these settings.

BUT even at the same bitrate multipass VBR has its advantages. LeeBear is exactly right. Additional passes can aid the encoder in motion search precision, yielding better results during fast motion, or otherwise difficult types of motion, ex: very long pans.

The truth is that with VBR max set to your CBR level you probably won't see any difference in quality at all, but the multipass VBR encode has the potential to be higher quality, even though in any given scene it never has more bitrate than the CBR encoded file.

I stand by what I wrote originally

D_Head wrote:
boathouse you are correct, if you do a VBR and CBR encode and the CBR bitrate is the same as the max of the VBR, THE CBR ENCODE WILL ALWAYS GIVE BETTER RESULTS.

Adam, I usually respect what you have to say, but sometimes I wonder whether you understand the fundamentals of compression, in this case mpeg.

An Example
If you encode a sample at 2520 CBR.

Now you encode the same sample using VBR using a max of 2520
,now depending on what you set your average/min bitrate to, the size of the output file will change.
e.g. smaller average smaller output file


The lower the average is set the worse the quality will be in comparision to the CBR encode.(fact)



Now all that aside, if I was doing a 480x576 SVCD encode and wanted an average bitrate of 2520 and was using VBR x-pass, i would use something like this........

min 200 -av 2520 max 3300.........

now I would agree you should get a better result than the 2520 CBR encode.

Cheers

[joes]

I would say the quoted phrase of D_Head is absolutely true, just simple logic, one can't do better than best, no matter how many passes.

When U don't have to press more than 35 min of SVCD onto one 74 min CD-R, do CBR 2520.

I don't know what's wrong with my copy of TMPGEnc (both 12a and the last version), it never produce the same quality (sometimes near) in VBR mode as CBR do (when max br in VBR = CBR rate), someone feel like to discuss the issue?

[SatStorm]

With TMPGenc 2.56 plus, I have those results (strictly unofficional and based my resent tests).

A CBR 2520 is quality equal with a CQ 2520max - 1200min @95%. The CQ file is smaller

A CBR 2520 is qualilty equal with a 2pass VBR 1200min - 2450aver-2520max. The VBR file is (of course...) smaller in size

A CBR 2520 has an overall lower quality than 2Pass VBR 1200min - 2520average-2520maximum . The VBR file tents to be same filesize the CBR (sometime bigger, sometime smaller)

A CBR 2520 has much Less Quality than a 2 Pass VBR with 1200min - 2520 average - 3520max. This is X for CD based media, but it is OK for 2 DVD.

With new TMPGenc versions, tests show me different results than those I post with PAL. New versions needs further tests and new optimizations from the user, expecially with the wideness of the min - max values.

A CBR 2520 is quality equal with a CQ 2520max - 1200min @95%. The CQ file is smaller

May appear to look the same but 95% is not 100%

A CBR 2520 is qualilty equal with a 2pass VBR 1200min - 2450aver-2520max. The VBR file is (of course...) smaller in size

May also appear to look the same(most of the time) as the CBR but 2450 as the average is only a little less than 2520 .

A CBR 2520 has an overall lower quality than 2Pass VBR 1200min - 2520average-2520maximum . The VBR file tents to be same filesize the CBR (sometime bigger, sometime smaller)

No WAY!!(bullshit). You gain nothing(quality wise) having 2520 as AV & MAX

A CBR 2520 has much Less Quality than a 2 Pass VBR with 1200min - 2520 average - 3520max. .

This last statement I agree with.


Cheers

[SatStorm]

@D_Head
Post whatever you like, the way you want it but please don't use that kind of phrases: "No WAY!!(bullshit)"
I done extremely tests for days to post those results and I have no problem someone say something different. A test from a third gonna show him who is right.
But I do have a problem when someone name my conclusions "bullshit". You have no right to do this.

Also, you don't seem to disagree with me on the other parts of my post, so, what's the point for quating me?

----------------------------------------------------------------------------------

A generall suggestion to forum members with many Nicknames: When you change nickname, also try to change the way you express yourself.

[LeeBear]

D_Head: You are the one that doesn't seem to understand. If we were encoding still images sure 2520 bits will always give you a better encode then a 2420 bits, but movies are not still image they move so motion search precision makes a big difference in picture quality. If you don't believe me encode something with TMPG using CBR of 2520 setting the motion search precision to "lowest quality (very fast)" then encode the same clip using "high quality (slow)". Using your argument both should look the same because they are both using a CBR of 2520... but they don't the one with "high quality" motion search precision looks better.

Now VBR will make the motion search even better so that's how you can get a better picture using VBR over CBR.

D_Head i suggest you shut up and stop talking about things you don't understand.

-LeeBear

[adam]

The lower the average is set the worse the quality will be in comparision to the CBR encode.(fact)

No this is not a fact and this it is this assumption that makes your original statement invalid. By lowering your average you DO NOT necessarily lower quality compared to CBR, that is the whole point of VBR encoding! If the bitrate is allocated well enough (decent encoder + settings), and you still give the encoder enough bitrate to work with (high enough average bitrate) then you do not lose any quality encoding with an avg lower than your CBR setting as long as your max is still at or above that CBR level. I'll say it again, with a decent encoder multipass VBR min 0, avg 2000kbits, max 2520kbits should be equal or better than CBR 2520.

A CBR encode may be higher quality than the equivalent VBR encode at a lower average. Naturally more bitrate never hurts, but the point is that it doesnt necessarily help either if the extra bitrate is just wasted. And when you bring added motion search precision into the picture with multipass vbr then the advantages of those extra bits in CBR become even less important. CBR may yield better quality than VBR but it certainly wont do so in all cases as you suggest, and in fact multipass VBR will probably outperform CBR in most cases, though not by much. Still, if you only need to fit say, 40 mins per cd in SVCD format you should just use CBR. Any benefits of VBR in this instance will be marginal and not warrant the extra encoding time.

D_Head creating aliases to post in your own defense is very immature and it doesnt exactly give any more credit to your posts, especially when both of your aliases are flat out wrong in their assumptions. At least be a little more inconspicuous about it next time. I hope you also realize that the forum admin can see the ip of every poster.

Satstorm & leebear, dudes this is a forum for discussion, if you dont like other people having a difference of opinion then dont post!! I would hardly consider bullshit offensive

Adam i do not like liars at any time, and as a moderator you should know better, especially when you have to resort to lies to backup an argument.


Quote from ADAM

D

_Head creating aliases to post in your own defense is very immature and it doesnt exactly give any more credit to your posts, especially when both of your aliases are flat out wrong in their assumptions. At least be a little more inconspicuous about it next time. I hope you also realize that the forum admin can see the ip of every poster.

What exactly is my other alias?

I have never used any other name on this forum except D_head. It would be nice if another Administrator would confirm this especially for this post!.(baldrick, Sefy etc etc)

An apology would be nice ADAM

Cheers

Matt


Ps some of you guys take things way to personal, chill, have a beer or somthing 8)

[adam]

Satstorm & leebear, dudes this is a forum for discussion, if you dont like other people having a difference of opinion then dont post!! I would hardly consider bullshit offensive

Calling someone's statements bullshit is most certainly offensive. This being a forum for discussion you should either learn some manners or stop posting.

Adam i do not like liars at any time, and as a moderator you should know better, especially when you have to resort to lies to backup an argument.

After having read this I sincerely hope that D_Head and joes are in fact two different people, otherwise that is just sad. Are you trying to be ironic?

Ps some of you guys take things way to personal, chill, have a beer or somthing 8)

Myself and others corrected your mistake at which point you went off the deep end, needlessly insulting anyone who disagreed with you. Who needs to chill? If it makes any difference to you I am drinking a beer right now.

I think boathouse got the answer he needed and I'm sure he can separate the fact from fiction in this thread. Move along folks, nothing to see here.

Joes agreed with me ADAM.

Check The IPS ADAM.

Cheers
Matt

[shardison]

1. CBR at 2520kbs wastes bits on easy-to-encode stuff.
2. CBR at 2520kbs doesn't have enough bits for hard-to-encode stuff.
3. Size of file is larger than it needs to be but maximum quality for the bitrate constraint.

4. VBR constrained to amaximum 2520kbs (with x average kbs) saves bits on the easy-to-encode stuff.
5. VBR saves those bits for more time--- not for the hard-to-encode stuff... that's constrained to 2520kbs anyway and it's still not enough.
6. Size of the file is smaller than CBR, but unless the encoder knows exactly how to save and allocate those extra found bits, quality will suffer.

Just because it sounds good that VBR saves bits from the easy stuff, doesn't meant that it can do so perfectly. VBR can't give better results. It can only allow you to fit more video time on a disc.

VBR can only give better results than CBR if you have a high bitrate overhead available. If the SVCD's spec could handle 6000kbs, then VBR would win easily--- as long as your goal was 2530kbs average.

[adam]

You are not listening. No one said VBR can achieve higher results than CBR, only that it can achieve the same results even with an avg bitrate lower than that of the CBR encoded file.

What HAS been stated is that MULTIPASS VBR can achieve better results than CBR even when the VBR max does not exceed the CBR level. There is more to quality than just the bits used in any given scene. Even if the CBR clip has as much or more bitrate at any given point in the encode than the multpass vbr encode the Multipass VBR encode CAN still achieve higher quality because the motion search precision will be increased.

I strongly suggest that people do what LeeBear suggests and encode the same clip in CBR at 2520 with both highest and lowest motion search precision and note that there is a noticable difference in quality even though the bitrate is identical.

And if you doubt that additional passes can increase motion search precision than do one CBR encode at 2520 and another in 2-pass VBR with a min, max, and avg of 2520 (essentially CBR) and again you should see a difference in quality even though the bitrate is the same. Of course in this test it helps to have a clip with lots of motion.

The added quality we are talking about is due to the extra passes, NOT the variable bit encoding.

[Nelson37]

One thing to note is that specifying a particular max and avg in an encoder does not necessarily mean that your file will have those exact max and avg values, or even very close. CBR 2520 will give you a file of avg bitrate 2520. VBR with avg 2000 and max 2500 could have real avg bitrate of 1800 and max of 2200.

CCE, for instance, seems to keep closer to the average. If you want REAL max values around 2500, specify 2800 to 3000 in settings.

This does not make visual comparison more difficult or invalid, it DOES make creating equivalent files more difficult.

[adam]

Well personally I only see this happen in TMPGenc. Pretty much every other commercial encoder I have tested seems to respect min, max, and avg settings fairly well.

[shardison]

2400kbs is not a high enough bit rate to make a block-free copy of every movie out there....CBR or VBR.

If VBR multipass is just a way to increase motion search precision, then why not just increase the motion search precision setting in your CBR encodes?

Save up your pennies and buy a DVD burner and all these problems will go away.

[adam]

2400kbs is not a high enough bit rate to make a block-free copy of every movie out there....CBR or VBR.

What does that have to do with this thread?

If VBR multipass is just a way to increase motion search precision, then why not just increase the motion search precision setting in your CBR encodes?

Well obviously this is not the sole purpose of multipass vbr, and using a high motion search precision will narrow the gap between multipass vbr and cbr, but there is simply no replacement for actually scanning the movie multiple times before encoding it. The encoder is doing alot more than just allocating bitrate and compensating for motion. The more information it has about the source the better and there is no way to get that information without using multipass vbr. I'm not saying that at the same bitrate multipass vbr will be dramatically better quality, or even better quality at all. Obviously this depends on the source and your settings. What I'm saying is that CBR will not necessarily be better than VBR at the same bitrate, which others have claimed. If all there was to encoding was the amount of bitrate used then it would not matter which encoder you used....think about it.

Save up your pennies and buy a DVD burner and all these problems will go away.

What problems?

dear ADAM and your misguided faithful followers.

ADAM WROTE

You are not listening. No one said VBR can achieve higher results than CBR, only that it can achieve the same results even with an avg bitrate lower than that of the CBR encoded file.

I think it is you not listening? maybe?

Remember this was the original Question.

If my VBR encoding has a maximum bitrate setting of 2520, and my CBR setting is also 2520, I can't see how VBR can be better, since with CBR you're giving it the maximum bitrate all the time.

like I said at the start

With the obove settings the CBR will give the better encode `quality wise`.
The VBR encode will approach and even appear match the quality of the CBR encode depending on the settings for the VBR encode. But setting the Max and average of a VBR encode to the same value defeats the purpose of VBR.

The only way the quality of the VBR encode could exceed the quality of the CBR encode is if the max of the VBR is higher than the CBR bitrate.(and the average of the VBR is high enough)

I still stand by my coment to satstorm

Satstorm wrote

A CBR 2520 has an overall lower quality than 2Pass VBR 1200min - 2520average-2520maximum . The VBR file tents to be same filesize the CBR (sometime bigger, sometime smaller)

No WAY!!(bxxxxxxt). You gain nothing(quality wise) having 2520 as AV & MAX






Cheers

oh 1 more comment ADAM.(Call it immature but I cant resist)

When a moderator misuses his position on a forum and uses lies in his argument this is really really LOW.
What does the forum admin say......funny you wont tell us all ADAM

For Example
Quote:-ADAM

D_Head creating aliases to post in your own defense is very immature and it doesnt exactly give any more credit to your posts, especially when both of your aliases are flat out wrong in their assumptions. At least be a little more inconspicuous about it next time. I hope you also realize that the forum admin can see the ip of every poster.

cheers
Matt

[SatStorm]

Look Matt (D_Head), let me explain you some things:

First of all, you using some phrases and you write the same way some well known older users of the forum. Also, the way you quote it is identical those same other users. So, or you are the same person, or you simply adapt the style of certain forum members, which all are the same user.
Complicated isn't it? 8)

Basicly, there is a party in this forum, against CVD, TMPGEnc and multipass VBR. They all say the same thing with the same order and the same phrases, that make me wonder: Are all of them twins or something?
Because I know how to use that search fuction, I was able to lock something. Then, we a little help of some people around, I found exactly who were they. Why I did this? Because the easy way was to live the forum and move to other ones. Well, I don't like the easy way

About the bullshit thing: You have to realise (as I did) that this is an international forum. Also, english isn't the native language for all of as. Now, what is the problem: A phrase or a word from one english dialect might be offensive for those who use other dialect. For example, since last moth, I used to use a lot the phrase "ain't" ("This ain't right" for example). Then, someone post and ask gently to stop using this phrase so much, because he find it extreme. So, I did a little digging just to find out that for Brits the use of "ain't" is nasty slang while for US ones isn't nasty So, I stoped use this phrase because some users in this forum may feel unconfortable reading it!
2 months ago, someone reply me in one of my posts: "Man What stuff are you smoking - must be good". Well, maybe that was a joke for him (it wasn't he never asked sorry), but it wasn't for me.

Overall, this is a technical forum and very official I have to say. So, let's keep it correct. We don't need phrases and words like that when we talk each other. We can use them when we want to discribe something (for example: It tried the new nero mpeg2 plug in, and the final file looked like bullshit) but we can't use it when we talk to others (for example: "Dick Head, you talk bullshit").

I don't like that many users here find me trouble maker or easy offensive but thank God much more people find me usefull!

Now back to our discussion:

I wrote " CBR 2520 has an overall lower quality than 2Pass VBR 1200min - 2520average-2520maximum . The VBR file tents to be same filesize the CBR (sometime bigger, sometime smaller)". I also wrote that "With TMPGenc 2.56 plus, I have those results (strictly unofficional and based my resent tests)".
You see, technically, in a ideal world with perfect software you are right. But this TMPGenc version isn't perfect and I had those results. I determine that I think.

About followers etc: There are no followers here. They are users love to post there conclusions and talk with others for the results. The only judge is the one who interest to test all the opinions.
So, I post something, you test and you judge compared to something else. You post you opinion and your results and the cicle ends here. If you do it honestly and correct, you have help all those who read your post.
That's it, nothing more nothing less.

So that's it Matt, have a nice day!

Hi Satstorm

Thank you for a sincere reply, although I dont agree with you on a few minor points, no probs.

One thing I can garantee you is D_head (me MaTT) uses no other aliases and I am sure the ADMIN can confirm this.

I am also not with a certian `Party`on this forum against VBR CVD or anything else for that matter. My opinions are 100% mine.

I also think you should be able to dissagree with someone without upsetting someone & no one person is right all the time. (I will try not to use bullshit again)

This is a great forum and I have learnt a lot from many of your posts, even though I dont agree with you all the time.

back to business

I wrote " CBR 2520 has an overall lower quality than 2Pass VBR 1200min - 2520average-2520maximum . The VBR file tents to be same filesize the CBR (sometime bigger, sometime smaller)". I also wrote that "With TMPGenc 2.56 plus, I have those results (strictly unofficional and based my resent tests)".
You see, technically, in a ideal world with perfect software you are right. But this TMPGenc version isn't perfect and I had those results. I determine that I thin

I have to ask you satstorm, why would somone use VBR with 2520av & Max, why not just use CBR.....2520?