CBR vs 2-pass VBR

[edDV]

edDV,

That is EXACTLY the "frosting on the cake" COMMENT I needed, and pretty much what I was thinking. Along those same lines, I archived my video to DV tape as well, and I looking for the BEST dvd VERSION of that tape archive. I had originally selected lpcm but have been unsure as to the video bitrate. Is your DV backup 8-8.5 mbps vbr or cbr?

So really, the end goal is
1. DV tape archive
2. DVD backup of that DV tape
3. DVD distribution copy

1. I log my original camcorder DV tapes and save raw (unprocessed) copies of the clips to DV tape maintaining raw PCM audio. Usually only 50-75% of the camcorder original video survives the logging process. I also save other audio sources and graphics files to DVDR in their original formats.

2. The final timeline of Premiere or Vegas gets exported to DV tape. I also back that up with a MPeg2 DVDR 7.5-8.0 Mb/s with LPCM audio (sometimes 3-4 tracks). If I think the the project is likely to be re-edited, I save the hard drive. This maintains the timeline and tmp files.

I once had a multi-8 track audio dub (off a mixer) that I saved in PCM synced to a 6Mb/s reference video track. The 2 channel mix was saved with the DV video.

3. DVD distribution copies are usually done 8.0 Mb/s CBR with 2ch AC-3 tracks for up to one hour. If more minutes are needed I use VBR or 2+ DVD volumes.

This is for original work not TV captures.

[slacker]

edDV,

Since you didn't specify, I'm guessing your DVD backups are variable bitrate encoded? (Unlike your distribution dvds.) If so, what do you use for min and max values?

[edDV]

edDV,

Since you didn't specify, I'm guessing your DVD backups are variable bitrate encoded? (Unlike your distribution dvds.) If so, what do you use for min and max values?

Backups are CBR. I'm trying for close to max quality without using a second DV tape per hour.

It could be argued that upping the bitrate to the max 10Mb/s might get some additional quality but DVD players may have difficulty.

[slacker]

Yes, I don't know how far you can take the video bitrate "thing". My Panasonic es30vs dvd recorder consistently outputs 9558 mbps video (variable) and 256 kbps audio. If I use THAT as my template, I would calculate backwardly...

9.800 mbps total (conservatively - rather than 10.080)
1.536 mbps audio (lpcm)
8.264 mbps video (max capable)

Does that seem rather high? Or such a minor difference from 8.000 mbps that it's simply not worth worrying about?

Or maybe MPEG Video Wizard and VirtualDub are giving me inaccurate figures for video bitrate? It is calculated on a variable bitrate encoded file, so sometimes I'm skeptical of the statistics.

[edDV]

Sounds right and the difference 8.0 to 8.2 would be small.

[slacker]

edDV,

Let me backup a minute! So your DVD backup is actually the mpeg2 file and lpcm file rendered from the DV tape and copied to a blank dvd for storage, i.e. not a dvd authoring but rather a data backup. The dvd authoring is only for distribution purposes? THAT I LIKE!

[edDV]

Not exactly, for serious projects I backup the timeline to DV tape.

Distribution copies are DVD Mpeg2 and AC-3 audio. I keep a distribution master ISO for copies.

DVD backups are usually DVD Meg2 with 2 ch PCM audio, sometimes I add 3-4 PCM audio tracks. These are usually authored as DVD but sometimes I just save the MPeg2 as UDF data. It could be done either way.

[Nelson37]

One thing usually overlooked in the VBR vs CBR debate are the benefits of the multi-pass encoding that is usually available with VBR but not, in many programs, with CBR. Using maximum bitrates eliminates the bitrate allocation issues, but the multi-pass can yield other benefits, such as improved motion prediction.

Note that this is not specifically a "benefit" of VBR, but a bonus derived from the multi-pass method, which is almost always available with VBR, but often not possible with CBR.

To summarize Vhelp's post (we need somebody full-time to do this!) I think the point was that many times faulty playback is attributed to bitrates which are too high when the real problem is a defective audio file. This may be related to not-completely-standard AC3 encoding, which may only become an issue when high bitrates are being used. Therefore the testing scenario, and the conclusions reached, are not valid.

[SatStorm]

I already said that Nelson37!

In Theory, a multipass CBR does a better pixel allocation per frame, even on higher bitrates.
IMO it is a waste of time.

But seems like they thought it was a typo....

[slacker]

Nelson37,

So you believe that if my target video bitrate is 8.2 mbps, for example, that 2-pass vbr would still yield a better result than 1-pass cbr, even at those high bitrates? These would be my two options using Vegas' MainConcept encoder.

SatStorm,

I still don't understand what you believe is a waste of time? What is the IT that you are referring to? Sorry.

[Nelson37]

Yes, there is an advantage in 2-pass encoding that is completely seperate from the bitrate allocation.

Sat, IMO, no I don't think you did. Your statement on multipass referred only to CBR, and only to "pixel allocation", by which I think you meant bitrate allocation. While I agree with this, there is more to the point.

To correct my earlier statement, hi bitrates do not eliminate the allocation issue, but they do significantly minimize it.

Multi-pass is better. Even at high bitrates, there may be minimal improvement in bitrate allocation. In addition, there can be a significant improvement in motion prediction which results from the subsequent passes. This is not relevant to CBR or VBR, only in that most encoders simply do not do multi-pass CBR. There were other factors besides just motion prediction, this was heavily debated long ago and I do not recall all the issues in detail.

Whether or not this difference is significant, or is worth the extra time, is a judgement call and is really not debatable. My main point is that there is something besides just bitrate allocation that must be considered in making this decision.

[TBoneit]

You may also find that TMPGEnc Plus & Xpress both have a CQ setting that can save space. CQ = Constant Quality. It ends up having a variable file size because the encoder uses less bitrate for low motion scenes than high speed motion, But keeps the quality the same. it is faster than VBR in my testing and seems to yield smaller file sizes than CBR.

[slacker]

I'm stuck with Vegas for now. (I'm not complaining!)

[edDV]

Nelson37,

So you believe that if my target video bitrate is 8.2 mbps, for example, that 2-pass vbr would still yield a better result than 1-pass cbr, even at those high bitrates? These would be my two options using Vegas' MainConcept encoder.

SatStorm,

I still don't understand what you believe is a waste of time? What is the IT that you are referring to? Sorry.

In my experience, 60min CBR @ 8Mb/s gets good results. For the next level of debate, you might want to ask this over in the DMN Vegas forums.

http://www.blue7media.com/vegas/

[bugster]

Referring to Satstorms comment about 2-pass CBR. Some encoders, well at least CCE, support 2-pass CBR. The thing is CBR is constant bitrate per GOP, not per frame. So, utilising 2-passes, even for CBR, in theory allows for better allocation of bits per frame within each GOP as well as giving the benefit of better motion compensation. Does it work, dunno, try it and see.

[rhegedus]

Read quick...

I read in a thread this week (can't find it now!) that if your bitrate is high enough (8 mbps or higher) you don't need to use vbr and that you would achieve better results using cbr for typical home videos. Any truth to this?

Yes - the closer the bitrate to the maximum allowed, the less gain there is to be had from VBR.

How to set optimal bitrate for MPEG

[SatStorm]

Thank bugster!
(I need to practice my english again....)