CBR vs CQ

[BV2001]

is there a big different between these mode ?? CBR or CQ using TMPGEnc LOW RESOLUTION DVD template ( 352x240 )
thanks

[johns0]

Cq will give you better quality video but with unpredictable file size,so one 90 minute movie might be 600mb and another could be 1500mb where as cbr is straight bitrate distribution so 1150 kbps video with 224kps audio will always give you 10mb per minute.

[MpegEncoder]

CQ is not available for VCD. VCD is CBR.

If you're SVCD, CQ is VBR which means that it will always be smaller than CBR.

[johns0]

You can use cq for 352x240 1/4 dvd if you want,this isnt about vcd anyways.

[MpegEncoder]

Sorry, when you mentioned 352x240 1150k video and 224k audio at CBR, that's pretty much VCD.

The point that I was trying to make it that CQ is VBR with a maximum bitrate of whatever you set it to. CBR at that same rate will look at least as good as CQ but will take up more space. CQ is simply a single pass VBR technique.

[spectroelectro]

CQ is nothing more than single pass VBR. So CQ saves a lot of space compared to CBR. If the file ends up too large for a DVD, download ReJig (freeware) or DVDShrink and squish the MPEG-2 file till it fits a DVD-R.

[MpegEncoder]

CQ is nothing more than single pass VBR. So CQ saves a lot of space compared to CBR. If the file ends up too large for a DVD, download ReJig (freeware) or DVDShrink and squish the MPEG-2 file till it fits a DVD-R.

It's still better to get it right the first time (for maximum quality).

[migf1]

Also, although I don't remember where I did, I read that TMPGEnc is optimized to work best with CQ.

[MpegEncoder]

Also, although I don't remember where I did, I read that TMPGEnc is optimized to work best with CQ.

You probably read it on this forum where someone said that they heard that TMPGEnc is optimized for CQ.

CQ works great if you want one pass VBR. Two-pass will always give the best quality/space. That's quality per unit space. The extra pass (in two-pass) gives the encoder the extra information that it needs to do the absolute best with the bits available. Using the bits where they are really needed and saving some in place that can do without. It also makes it possible to DEFINE the final file size. I keep hearing folks in the forum say that two-pass give better file size PREDICTIONS, but it's not a prediction, it's a definition. You tell it the average bitrate you want and that's what it gives you.

[SatStorm]

I am one of those who objervate that after version 2.56, TMPGenc focus on optimising CQ mode over 2 Pass VBR mode.

In theory 2 Pass VBR is better CQ (also known as 1 Pass VBR). But TMPGenc doesn't follow the theory, unfortunatelly.
If you set up to TMPGenc to encode 2 pass VBR with 1000min, 3000 average and 5000 maximum, and compare something you encoded from the same source with CQ @ 1000min - 5000 maximum @ 65% (the default setting), the CQ file is far - far better. And if you filter your source correct before the encoding (I don't talk here for the built in filters...) then the CQ file with also be far smaller in filesize...

I suggest other encoders if you wish to work with 2 Pass Mode. Imho, both Mainconcept 1.4.1 and CCE Basic, is better on this. For those who use to work with TMPGenc, Mainconcept is a better choice not because of the picture quality (CCE is better...), but because is it looks and feels far more user friendly for them.
The best root of course, is to learn avisynth and use CCE.

[MpegEncoder]

I am one of those who objervate that after version 2.56, TMPGenc focus on optimising CQ mode over 2 Pass VBR mode.

In theory 2 Pass VBR is better CQ (also known as 1 Pass VBR). But TMPGenc doesn't follow the theory, unfortunatelly.
If you set up to TMPGenc to encode 2 pass VBR with 1000min, 3000 average and 5000 maximum, and compare something you encoded from the same source with CQ @ 1000min - 5000 maximum @ 65% (the default setting), the CQ file is far - far better. And if you filter your source correct before the encoding (I don't talk here for the built in filters...) then the CQ file with also be far smaller in filesize...

Well, I love anecdotal storys as much as the next guy, but this isn't a terribly scientific way to approach the issue.

I ran a little test of my own. A 5 minute movie clip encoded with the same settings that you used (I don't know what "Motion Search Precision" you used, I just used "Motion Search Estimate (fast)").

Here are my results:
2 Pass VBR: 129,095KB
CQ 65%: 150,893KB
CBR @ 3000Kbps: 129,142KB

There was NO noticeable difference in the quality between the 2 pass VBR and the CQ (The CBR looked pretty crappy due to lack of high end bitrate).

I'll go out on a limb here and make this statement: There is just NO WAY that you will get a smaller VBR mpg file using CQ (CQ is VBR) that looks FAR FAR BETTER (that's what you said), than using 2 pass VBR.

It might look just as good, and it will definitely take less time to encode.

CQ is not magic.

[SatStorm]

I don't approach scientific anything. I objervate.
If i wanted scientifical approaching, I wouldn't be here. I would be at doom9's forum.

Now, read my words:

IF you filter your source and you encode a 2 HOUR project to both CQ and 2 Pass VBR with TMPGEnc Plus, then the CQ file will be BETTER if you follow my suggested example.
Also, a 5 min test with an UNFILTERED source, is a joke for TMPGenc.

CQ is not magic. 2 Pass is better.
But not for TMPGEnc. With TMPGenc, all later versions simply gave up on 2 Pass VBR. Hori Son gave up on it. Simply like that. I can't blame him...

[SatStorm]

From the other hand, you are right. Since I didn't show you how to test it yourself in the correct way, I telling bullshit.

So, I just decided that my next article in VCDHelp won't be how to capture PAL60, but how to set up and test yourself the TMPGenc Encoder and the modes it offers.

Meanwhile, repait the tests with a long time length source (1 hour for example). A filtered source. This is neccessary for CQ, but not so neccessary for 2 Pass.

CQ mode on TMPGenc, rise the bitrate over the minimum value, up to the % you set it up, when the encoder ask for it. So, if the picture is really noisy, with CQ the bitrate gonna rise a lot just to keep up with no issues.

Another problem you have with TMPGenc and the 2 Pass VBR mode it offers, is that you have to rise a lot the minimum value of an encoding, so to eliminate the noise in the static background scenes. With CQ, you don't have this issue.

For now, I shall repait what I post the last months: If you wish to use 2 Pass VBR for your projects, do it with other encoders, just not with TMPGenc. Mainconcept and CCE is FAR better TMPGenc on those modes.

What I mean "FAR"? Probably a 5% difference. But if you turn picky on quality terms over the years, like I did, this fuckin 5% is a huge difference.