How to use the Robshot-Bach method in DVDRebuilder + CCE

[Fran-K]

I'm really a fan of CCE encoding using the RB method. For me, it's the best 2-pass encoding you can get.
When I start to use DVDRebuilder, I couldn't stay without it.

For those who want to try it, let's do it:

1. On DVDRB, de-select "One Click Mode". Select "CCE mode".
2. Click on "Prepare"
3. After "Prepare" is finished, go to Windows Explorer and make a copy of the file REBUILDER.ECL.
4. Edit this copy with a text editor.
5. Change all instances of "vid_out=1" to "vid_out=0". This will disable the generation of the video files.
6. Change all instances of "vmode=1(or 2)" to "vmode=0". This will enable the One-pass encoding on CCE.
7. Save this copy of the ECL file, open in CCE and Encode it. After a few hours you'll get the 1st step VAF files.
8. Exit CCE, back to DVDRB then click on "ENCODE". After this you'll get the 2nd step M2V files.
9. Click on "Rebuild". You'll get a VIDEO_TS folder with all the files needed to burn it.

Of course, You can edit other lines on the file (opv_q_fator,...) but it seens make no difference on the encoding quality.

[guns1inger]

What does this do to improve quality ? All I can see this doing is a full analysis pass first, the video encode pass second, rather than going segment by segment. I can't see how this would change the outcome.

Can you give me a little more detail as to why this is so good ?

[Gurm]

I'm curious too. Is doing the entire analysis pass all at once somehow better?

- Gurm

[guns1inger]

It's still going to do the analysis on a segment by segment basis, so if segment boundaries are an issue (I haven't seen anything to suggest that they are), this wont change anything.

[Chip718]

I tried to use DVDRB and CCE to backup Mystic River, but after I 'Prepared' the movie in DVDRB and clicked on encode,CCE kept giving me Quality Setting windows every few seconds, so I just canceled the program. Does anyone know why that is? I configured DVDrb right and I also had the "One Click"setting selected. TIA

[Gurm]

gunslinger:

That's how I thought CCE behaved.

In that case, I see NO reason to do this, unless you LIKE making more work for yourself.

- Gurm

[archaeo]

Chip718 wrote:

CCE kept giving me Quality Setting windows every few seconds

In RB under <options> <setup>, check the 'Run Encoder(s) Minimized' box. It will keep the main CCE window from coming up.


-a

[guns1inger]

Gurm

Exactly. I'm begining to think April 1 has come late to someone this year.

[XL3/\/7]

More work delivers better quality.DVD Rebuilder simplifies things a lot and delivers good quality. I don't use it, since I prefer the good old way but it's better than many other programs that may not come for free either.

[Fran-K]

Hi everyone:

The purpose of this topic is to help users those are familiar with CCE, the RB method, and think that it gives better results than a straight 2-pass VBR.

There's tons of topics discussing MPEG encoding tools and this one is not intended to be another one.

I do batch encoding using CCE and BatchCCEWS using the suplied "Robshot-Bach" template, and then make this guide to reproduce, inside DVDRB, the same process done by BatchCCEWS:

1. Do a One Pass VBR enconding. Without filesize restrictions, the VAF files keeps the best possible distribution of the bitrate in each segment.
2. Do a 2-pass Multipass VBR encoding. CCE then simply adjust to the desired filesizes using the VAFs done in step 1.

The step 1 is done outside DVDRB, using the edited copy of the ECL file. The step 2 is the standard ECL file done inside DVDRB, but the VAF files made on step 1 force CCE jump right to the 2nd pass.

I'm not a CCE expert, but for me the RB method is the best of both worlds.

[Gurm]

Again, I'm failing to see the point.

We do the first pass outside RB, but with exactly the same parameters and settings... so why bother doing it outside RB?

- Gurm

[adam]

Fran-K, the way you described it, that's not the Roba method. I suppose you are using the Roba setting in DoCCE4u to accomplish your first pass. If you are a fan of the Roba method then this is a viable process. As written it just sounds like you are doing manually, what DVDRebuilder is doing automatically. That's why there's all the confusion.

Roba uses quality based encoding for the first pass. The whole method is really nothing more than an algorithm to calculate the correct bias to use for the first pass to, hopefully, achieve a better quantization level on the first pass then what the encoder would otherwise reach doing its own first pass at the default bias. The method could potentially reach the optimum quantization level on the first pass, but if you are willing to do 2 or 3 passes in regular VBR mode its not like the Roba encode will be any higher quality. Basically, it can reach the same end result in less passes, but there's no guarantee. In my opinion, the only use for Roba for DVD backups is to try to come close to your target size on the first pass. If so, your done encoding and the quality should be about as good as a 2 or 3 pass encode. If not, then you have to do the second encode to adjust the filesize to your target size, and you've negated the benefit of Roba (less passes.) I'd just assume set my encoder to 3 passes and go to sleep.

The real use for Roba is when you have lots of clips and no idea how many disks you want to spread them out on. You can do quality based encodes only attempting to reach a desired Q, not caring how much bitrate you use. Basically, you'd be using just enough to get relative quality on each clip, as opposed to just using the same avg for all clips and wasting bitrate on some while not using enough on the really complex ones.

In summary, Roba is different then regular multipass VBR, but if you are doing 3 or 4 passes anyway then it really makes no difference at all.

[Gurm]

Ahh yes.

The confusion here is that what Fran-K has described for DVD-RB is NOT Roba method at all.



We were right, he's just doing too much work. LOL.

- Gurm

[Fran-K]

Adam:
Thanx for your reply.

All my knowledge about the RoBa method came from inspecting the ECL files created by BatchCCEWS (0.9.0.8) for the suplied template named "Robshot-Bach". My version of DoCCE4u don't have a RoBa template.

When you create a job on BatchCCEWS using the RoBa template, two CCE encodings will be done.

. The first is a OPV encoding, but generating only the VAF file.
. The second is a Multipass encoding with Passes=2 (you can set to more on the "Encoding quality options"). CCE finds the corresponding VAf file, so it jumps to the second pass.

I've heard a lot of the RoBa method, but at that time, I couldn't find any guide to make it. When I've found the template on BatchCCEWS, I've assumed that's OK.

In summary, Roba is different then regular multipass VBR, but if you are doing 3 or 4 passes anyway then it really makes no difference at all.

I agree. But my point is to find the best 2-pass encoding with fixed filesize. Sometimes I have to encode on a PIII 750, and a 3+pass is out of limits.[/b]

Without worring about if the RoBa template on BatchCCEWS is right or not, the trick of making the first pass of encoding using the OPV mode sounds logical to me.

On OPV mode, CCE has only one pass to achieve the best quality. The VAF should have the most acurate settings for quantization and bitrate. This make the things easier for the next passes.

For me, the RoBa method give me best results than a 2-pass MPV, especialy on high compression DVD jobs (> 45%).

[adam]

I agree. But my point is to find the best 2-pass encoding with fixed filesize. Sometimes I have to encode on a PIII 750, and a 3+pass is out of limits.

That makes perfect sense, and ROBA is a good solution for this.

On OPV mode, CCE has only one pass to achieve the best quality. The VAF should have the most acurate settings for quantization and bitrate. This make the things easier for the next passes.

Not true. If your bias and other settings are the same then OPV mode and the first pass of your x-pass will produce the same result. Simply doing a separate quality based encode to create the vaf, as opposed to letting CCE do it, is NOT Roba and it doesn't accomplish anything. The idea is to try to calculate a good bias according to how much bitrate you think you will need. If you aren't calculating your bias according to the Roba algorithms then you aren't doing Roba.

If you want to keep things simple, I suggest you play around with different versions of BatchCCE4u and maybe even go all the way back to DoCCE4u. Some versions have a roba option built in and they will calculate the bias for you.

I'm sorry to say that your current method isn't accomplishing anything.

[Gurm]

I agree that ROBA may be wonderful.

But what you described, in your original post, way up at the top of this thread... DOESN'T DO ROBA.

- Gurm

[Fran-K]

From the BatchCCEWS Help documentation:

Automates running the Robshot-Bach (RoBa) method of encoding as described on the Doom9 website.

So, the calculations done by this SW should be a good reference.

If you aren't calculating your bias according to the Roba algorithms then you aren't doing Roba.

Agreed. Now I've notice that when I change the avg bitrate in BatchCCEWS, the bias setting adjusts accordly. And the bias IS NOT THE SAME generated by DVDRB on the REBUILDER.ECL file for a MPV encoding!

This don't affect the first part of RoBa, because it don't use the bias value, just the quantization level. BUT affects the second part of RoBa.

This way, to really do RoBa on a DVDRB project, I'll have to load all the AVS files on BatchCCEWS with the calculated bitrates in DVDRB and let BatcCCEWS do the encoding. Or change the bias settings on the REBUILDER.ECL file using the BatchCCEWS calculations. It's a lot of hand work! While DVDRB don't offer this option, that's the way.

Well, back to the lab, Igor!

Thanks everyone for the tips.

[Gurm]

Good luck. If you find a semi-automated way to do this, let us know.