Clairification on a Rebuilder Setting

[ron spencer]

Hey all, the settings guide says:

"VBR_passes: Sets the total number of passes the CCE encoder makes. NOTE: This setting is not equal to CCE's "passes" setting because CCE does not count the initial .vaf creation pass."

Since these two are not equal, does this mean that if I want CCE to do 3 passes I need to set DVD Rebuilder to 4 passes? Just wanted to be sure on my slower PIII 866 before staring out.

[Joergen]

Writing the .vaf is one pass, thus 3 passes is .vaf pass and 2 extra passes in a new type of window of CCE. But 3 = 3

[ron spencer]

Thanks....so when people say they get the best result at 3 passes that is setting DVD Rebuilder's CCE options to 3 then? I was just confused by the english.

[jdobbs]

I personally normally set the passes in DVD-RB to 2. Sometimes on a long movie I will set it to 3. If I just want to be nuts I go to 4. The other settings are for those who have lots of time on their hands.

[wmansir]

Originally posted by ron spencer
Thanks....so when people say they get the best result at 3 passes that is setting DVD Rebuilder's CCE options to 3 then? I was just confused by the english.

When most people say 3-passes or 9-passes, they are using CCE's setting. Meaning 3 +1 or 9 + 1 for the .vaf creation pass. But Jdobbs refuses to bend to CCE's will and not count the .vaf creation pass. ;)

[Joergen]

But the .vaf pass is a full pass with almost exactly the same type of calculations as the rest of the passes. Perhaps CCE writing a .vaf separately is merely to show vital information about the first pass and not starting the passes-window with one pass already included. Its merely an engineering choice.

A CBR pass is a pass aswell, Pass 1/1, not 0/0.

[jdobbs]

Originally posted by Joergen
But the .vaf pass is a full pass with almost exactly the same type of calculations as the rest of the passes. Perhaps CCE writing a .vaf separately is merely to show vital information about the first pass and not starting the passes-window with one pass already included. Its merely an engineering choice.

A CBR pass is a pass aswell, Pass 1/1, not 0/0. It's not even "almost" -- it is a pass with the exact same calculations. I am sticking to my guns.... people should be arguing with Custom Technology, not me. As for the other passes... as much as I love CCE, they are virtually useless and is (I think) a sales gimmick. More isn't always better. How many other MPEG-2 encoders have you ever seen with more than 2 passes? It's mathmatics. Once you have seen the data one time through -- you should have all the information necessary to be able to calculate the best bit distribution across the entire stream. What possibly could have changed in the third pass? Remember you are running over the same source data every time.

[onesoul]

How many passes should be needed with interlaced sources, the one I did (60%) didn't look very good (somewhat blocky) with 2 passes (dvd-rb nomenclature) ~4200 bps and dynamic assigned cell bitrates, (with dvdshrink/recode it actually was better except the complicated logo of HBO fading in where cce did much better). Also another question the bias setting isn't used by 2.67 is it? Any suggestion for quality precision (or even bias) with interlaced format?

And another one more question, what is the quantization matrice being applied if any important?

Thanks

[quantum]

Originally posted by wmansir
When most people say 3-passes or 9-passes, they are using CCE's setting. Meaning 3 +1 or 9 + 1 for the .vaf creation pass. But Jdobbs refuses to bend to CCE's will and not count the .vaf creation pass. ;) I never used that numbering. I'm with jdobbs on this one. I always considered two passes as vaf+1 pass. Otherwise how many passes is a cbr? Zero passes? That's real quick..

[jdobbs]

Originally posted by onesoul
How many passes should be needed with interlaced sources, the one I did (60%) didn't look very good (somewhat blocky) with 2 passes (dvd-rb nomenclature) ~4200 bps and dynamic assigned cell bitrates, (with dvdshrink/recode it actually was better except the complicated logo of HBO fading in where cce did much better). Also another question the bias setting isn't used by 2.67 is it? Any suggestion for quality precision (or even bias) with interlaced format?

And another one more question, what is the quantization matrice being applied if any important?

Thanks That's surprising. You can usually get a good picture at 4200 Kbps. Is it a high action movie? I'd say the biggest challenges I've seen are homemade videos done by a non-professional-- where the camera movement is constant. This might be one where an extra pass is warranted (sometimes a couple percentage points of improvement is worth the time).

That reminds me. I haven't implemented a way to go back and do an additional pass, have I (for SP versions)?

vbr_bias: Hmmm. The newest version of the SP software that I have is 2.66 and it was still there then. It isn't in 2.67 Basic but I assumed that is because it only does 2 pass... I'll have to go to the CCE forum and do a little research.

I use the default quantizer on everything, but I'm pretty confident you can do better on some kinds of material (like animation) with different matrices. Maybe I should add that to the CCE options box.

[onesoul]

Originally posted by jdobbs
That's surprising. You can usually get a good picture at 4200 Kbps. Is it a high action movie? I'd say the biggest challenges I've seen are homemade videos done by a non-professional-- where the camera movement is constant. This might be one where an extra pass is warranted (sometimes a couple percentage points of improvement is worth the time).

That reminds me. I haven't implemented a way to go back and do an additional pass, have I (for SP versions)?

vbr_bias: Hmmm. The newest version of the SP software that I have is 2.66 and it was still there then. It isn't in 2.67 Basic but I assumed that is because it only does 2 pass... I'll have to go to the CCE forum and do a little research.

I use the default quantizer on everything, but I'm pretty confident you can do better on some kinds of material (like animation) with different matrices. Maybe I should add that to the CCE options box. The DVD is Robin Williams Live on Broadway (1 hour and 40 minutes lenght) (PAL) (funny guy) and he moves like hell while talking from begin to end, different cameras are always pursuing him. I am using CCE SP 2.67.00.23 trial, I searched through the settings and couldn't find Bias (or is it me?). I normally use the standard matrice, I will try using mb1 interlaced dv matrice, but if if you had any suggestions for this and other sources I would definitely use them.
Going back and doing doing additional pass, I think it is doing so, (encoding took 160 minutes 2 passes with 2100 mhz (180FSB) athlon) how do I make sure?
(Do you think I use parenthesis a lot?)

[jdobbs]

@onesoul

The bias is set in the multipass "Advanced" screen in 2.50 and the "Bit Allocation" button in 2.66 -- the only newer 2.67 version I have is Basic (which doesn't support it).

You know, you're right. If you go back in again with Phase II and choose "No" to the "Resume" question -- it will do the number of encodes you have selected in addition to the ones that were already done. Hey, I'm smarter than I thought! :eek:

But that also brings about a possible bug. If you have any kind of failure the first time through and a corrupted .VAF exists, it may cause failure on the second run...

I like (parentheses) also.

[onesoul]

@jdobbs

Why don't you try the latest the cce sp (I just went to cinemacraft page and by coincidence it was released today a new version 2.67.00.27 wich along bug fixes includes an enhancement "Added blanking function. (standalone version)") at http://www.cinemacraft.com/jpn/download.html

[onesoul]

I tried extreme settings all around (even found bias to change :) which is 30 by default to cce), I just can't get a high quality output with interlaced sources at 4200 kbs, previously I said it appeared somewhat blocking but is more like mosquitos everywhere.
Many people say procoder is the best with interlaced material, I didn't try it yet, what do you think? If it is so good could it be integrated at some point with DVD-RB?

edit: trying rejig now

[onesoul]

Hi again, (rejig didn't behave too well with interlaced, worse that cce). Just to confirm as I haven't been able to, to the moment it isn't possible to change quantization matrices?

[dvdRENEGADE]

Originally posted by onesoul
Hi again, (rejig didn't behave too well with interlaced, worse that cce). Just to confirm as I haven't been able to, to the moment it isn't possible to change quantization matrices?

Just curious...what was the original's avg. bitrate of the Robin Williams dvd?

[onesoul]

http://pwp.netcabo.pt/onesoul/bitrate.jpg

(Why can't I attach an image?)
EDIT: Thanks wmansir.

[wmansir]

Originally posted by onesoul


(Why can't I attach an image?)

In-line images are a per-forum setting, which is off by default. It looks like when they created this sub-forum it wasn't turned on. I'll request it be done.

EDIT: Done.