2-pass CBR in CCE Basic better than CBR in TMPGEnc?

[Tom101]

Hello,

I want to encode 60 minutes of video at 8.5 MBits/s, CBR. Will the 2-pass CBR in CCE Basic give better results than CBR in TMPGEnc at this bitrate? The video is interlaced. I realise there have been many discussions on the relative merits of these two encoders, but I would like some advice that relates to my specific situation.

Thanks,

Tom.

[bugster]

I don't think it is possible to correctly answer this question.
Firts off, I assume you mean CCE 2 pass VBR in CCE, 2-pass CBR does not make any sense.

Secondly, Quality is very subjective and these two encoders abilities are pretty close, so what you think looks better may look the porrer to another person.

Thirdly, you don't mention your source material. If it is VHS or some other noisy analog source, you could probably achieve better results with TmpGenc and some of its built in filters than with with CCE alone. Using external filters (vdub, avisynth) and frameserving to CCE may be better than TmpGenc.

Why not simply take a 5 minute clip from your video, encode it both ways and see which you prefer!

[Tom101]

Thanks for your reply. CCE Basic does have 2-pass CBR, in addition to 1-pass CBR. It analyses on the first pass and encodes on the second, but at a constant bitrate. In the window you can actually see it complete the first pass and then perform the second.

I need to encode high quality DV. I thought the question was worth asking because SatStorm once wrote that TMPGEnc's CBR is "clearly better" than CCE Basic's CBR.

I cannot compare the two encoders myself because I do not have the equipment to so, I wish I did!

Thanks again,

Tom.

[flaninacupboard]

What equipment? all you need is the machine you're typing on and the two bits of software! You can get the trial of TMPGenc for nothing, and i believe CCE basic is only $40 or so. go check it out!
As for which is better, technically CCE should be because of the two passes, but you should try yourself. personally (assuming the MPEG stream is for DVD) i would encode 3pass VBR 100min 8500average and 9800max for the best quality.

[Gazorgan]

I've done 100's of short clips on a PII 333 Mhz. Granted most were 45-300 seconds, but it was adequate for the task. However, if you running an AMD K6, your hosed.

[DVWannaB]

Hey experts help me out here. Why would you want to or ever need to do a multi-pass CBR Would not a CBR at whatever bitrate on slow or slowest mode generate the very same output as a multi-pass After all, with a fixed bitrate, there is a limit on how good your output can ever attain. Sounds like a waste of time to me. Am I correct or just way off on this

[bugster]

I have to admit myself as to being confused on the benefits of 2-pass CBR. It seems almost like a contradiction in terms, hence my original comment.

Surely, with CBR, you have fixed amount of data per GOP, so no amount of passes can change that. I suppose there might be some merit in varying the bit rate distribution within a GOP, but this should still be possible with a single pass. Anyone got a decent explanation?

[flaninacupboard]

Well, using multipass means different levels of compression can be applied to different parts of the frame (and GOP) resulting in (hopefully!) a nicer image. if you remember all the "CQ is better than CBR!!!!" rows, this is basically the same tihng.

[DJRumpy]

Multipass CBR does have benefits over single pass CBR. The additional passes allows the encoder to intelligently handle scene changes and fades. With this information, it can place additional I-Frames in the stream to prevent the bitrate from being eaten up to the point where picture quality deteriorates. During a scene fade for instance, the encoder will see little, or no motion, until the fade reaches the point that it causes the entire frame to look 'new', requiring the entire frame to be encoded until the next I-Frame is encountered. By positioning I-Frames in these transitions, it can better handle the bit allocation.

The same is true for a scene change. If the encoder does not detect the scene change, then every frame after will appear as 'new' motion, requiring bitrate to encode. Since the encoder does a first pass, it see's the increase in bitrate needed, and can properly assume that a scene change probably took place. It would then insert an I-Frame to allow proper detection of motion under the new scene.

Bugster, CBR is fixed at X number of bits per second, however, you will see variance of the bitrate within that second. This would lead me to believe that the encoder could take advantage of this, assuming it has foreknowledge of each GOP. It could better allocate bitrate at the frame or macroblock level, while keeping the bits per second constant.

[bugster]

@flaninacupboard, CQ is a single pass VBR encoding, we are discussing CBR so your point is irrelevant here.

@DJRumpy, your explanation does make sense, esp w.r.t scene changes/fades etc, thanks for that.

[flaninacupboard]

I know what CQ is
My point was this,
CQ was being touted as marvellous because as there was less restriction on bitrates better image quality is achieved, as opposed to X-pass VBR situations where bitrate is the limiting factor. Similarly, a one pass CBR encode, bitrate is the limiting factor, wheras a multi pass CBR can use information both ahead and behind of the current frame to best reproduce the original, i.e. exactly what DJrumpy said; especially the point on variable GOP sizes, as this is a clever way to get better picture quality from what is still a CBR MPEG as far as each second is concerned.

[SatStorm]

IMHO, CCE and TMPGenc both gonna look identical in your eyes with this bitrate you prepare to use (8.5 MBits/s)!

2Pass CBR is better CBR. Detects better scene changes, add better the intra frames and allows better pixel alocation per frame/field.

With TMPGenc you don't have 2 Pass CBR. You can emulate this mode if you choose 2 Pass VBR and set min/average/max values all the same. Motion Estimate Search Mode can be used for faster encoding with no quality loss in this situation (IMHO)

Why IMHO gonna look identical?

CCE SP (standard I don't know...) is better for 1/1 D1 Framesize than TMPGenc.
TMPGenc handless better Interlace sources.

With this bitrate (8.5 MBits/s) you can use for sure 10DC for both encoders.

So, CCE gonna look smoother, TMPGenc gonna look sharper.
Overall, both gonna look identical!

Of course, you realise that you gonna see the difference only on a 40"+ Plasma screen (not LCD or CRT) with a Hi-End DVD standalone. In a typical 32" TV, I don't believe it is possible to see a difference with this bitrate (8.5 MBits/s)

That's my opinion. Others may have a different!