TMPGEnc 2.56 Plus 2-passes

[Arnnie]

I am officially convinced that x-pass encoding yields a greater quality than that of CQ VBR encoding, with the help of Adam's justification. 8)

OK, so lets say for the purposes of this post that x-pass encoding is the way to go. I recently stumbled upon TMPGEnc's Plus-2.56.39.143 version which includes two implementation of 2-pass encoding (2-pass old-type and 2-pass).

This may not be news to encoders within this advanced forum. But after further testing of the two mechanisms I have discovered the new 2-pass method provides a much higher quality than that of the old one. This was verified with a bitrate viewer, displaying a much more efficent allocation of bits in high/low action sequences.

So I consider the best method for encoding DV to SVCD or in my case CVD with TMPGEnc (don't want to start a TMPGEnc vs ??? war) is with 2.56 plus's new 2-pass encoding mechanism.

I would be interested in other people experiences with both VBR encoding, whether it be CQ or x-pass and SVCD or CVD.

I would like to acknowledge Satstorm and vhelp for the informative posts of CVD encoding. I will never go back to SVCD.

Arnnie

[energy80s]

Arnnie, I am testing the 2.56 plus version (in MPEG2 mode) and have found that the blacks are coming out a bit grey looking. Also noticed "interference lines" streaking across the picture during credits on various DVD conversions. Have you noticed this. It doesn't seem to happen with 2.54.

[Arnnie]

have found that the blacks are coming out a bit grey looking

energy80s, I have not noticed this, granted most of my tests were conducted on Gladiators opening scene, which is quite grey in the first place. However with all my tests, none of them indicate the symptoms you describe. I will endeavour to examine your problem further.

Has anybody got issues with this method?

Arnnie

[SatStorm]

With TMPGenc you have problems in dark scenes if you don't give enough bitrate. With CVD I think with a 1200min you get excellent results.

[bbb]

BACKDROP:
I had a problem when I purchased a new dvd player (Pioneer 440) a while ago-it would stutter at normal SVCD bitrates and would exhibit green random dots/white dragon tail blocks and other strange artifacts when using lower bitrates.

After encoding a video segment 30-40 times using various different bitrates and encoding methods (CBR, CQ, 2-pass VBR, without buffer, with buffer, with various audio bitrates, auto mux by Tmpgenc, manual mux by BBMPEG), I found that my dvd player had a below averge upper bitrate limit and that it also had a lower bitrate limit. I eventually found the perfect range of bitrates to encode my SVCDs/SXVCDs.

CBR, CQ, 2-pass VBR:
As a side consequence, I was able to compare CQ to CBR to 2 pass VBR.
On my CDRW I would have 6-10 video clips of the same sequence encoded with slightly different settings and bitrates. Using a single blind visual quality examination and taking notes (to look for the random blocks and wierd artifacts) I inadvertently came to the conclusion that 2 pass VBR produced the best quality video (nearing CBR) in an efficient file size (only slightly larger than my CQ encodings). I had underwent many CDRW burns and using the same upper and lower limits (padded) for CQ and 2-pass VBR. The CQ encodings would exhibit more pixel and color artifacting than the 2-pass VBR encodings; only at a CQ of 90+ would the CQ quality approach that of a 2-pass VBR encoding. However at a CQ of 90, the file size would exceed that of a 2-pass VBR encoding.

Despite the hype that CQ provides the best quality video, my extensive tests and copious notes have shown that 2-pass VBR provides superior picture quality. But this doesn't mean that CQ doesn't have a use, it encodes faster than 2-pass VBR and yields a smaller bitrate than CBR. CQ is a compromise for decent quality at a decent speed.

GRADES:
CBR: A (encoding speed), A (quality), C- (file size)
CQ: B (encoding speed), B- (quality), B (file size)
2-pass VBR: D (encoding speed), A- (quality), A (file size)

Which method to use depends on your needs. If you have a DVD burner and don't want to spend too much time encoding, use CBR. If you make high quality SVCDs and don't like to switch discs very 35 minutes and have a fast CPU, use 2-pass VBR. If you make decent quality SVCDs and don't like to switch discs often but also don't want to spend too much time encoding or you have a slower CPU, use CQ.

[pauncz]

Hi,
Could you post your most successful 2pass VBR settings??

I too am trying to figure out the best setings for SVCD that would play on most home DVD players.

Also, do you have any recommendations for media? CD-R vs CD-RW?

Thanks
Tom

[W_Eagle]

Arnnie,

Are those 2 modes of VBR built in on TMPGenc 2.56 Plus,
ot is it a kind of settings?

Take care

[Arnnie]

Are those 2 modes of VBR built in on TMPGenc 2.56 Plus

Yes W_Eagle, these two implementations of 2-pass VBR encoding are built into the program.

- 2-pass (old type) -> supported in all Free copies of TMPGEnc
- 2-pass -> TMPGEnc Plus 2.56, which uses a higher quality algorithm compared to that of the old type (according to my test and analysis.

What settings do I use?

I encode all my DV to CVD format 352x576 (pal)
Audio -> 128-224kbps @ 48Khz (CVD standard, not valid for SVCD)
max video -> 2520kbps (always)
min video -> 600 -> 1200kbps (depending on duration)
Avg -> (depends on duration) 1400Kbps gives great results, I use 90min CDs to get that extra 80MB. I would use 99min but DVD player compatibility will be reduced.

A good guide is to just load the SVCD templates provided by TMPGEnc, unlock the parameters and adjust CVD and VBR settings accordingly. EASY!!!

I don't undestand why fellow encoders insist on VBR CQ method, in terms of quality anyway. If time is not a factor i.e overnight encoding, then x-pass encoding is for you.

Arnnie

[SatStorm]

A generall suggestion:
Don't load the DVD template of TMPGenc. Load the SVCD one and adjust the picture output resolution to 352 X 576/480 and set the bitrate you wish.
A good start is 1200minimum, 2300 average, 2520 maximum. You can test at lower bitrates, most of the times an average of 1900 is enough!

The CVD audio is 44100 as SVCD and not 48000. BUT, as I wrote in my Faq, most of the standalones after 2000, are also capable to support 48000 audio on CDs. Not all, most! Try first if your player support this and then procceed!

[W_Eagle]

Arnnie and SatStorm

Thank you very much for your detailed reply.
I'm going to get TMPGEnc 2.56 Plus later on this evening.
Meanwhile, how does one choose (in the menus or templates)
the old or new 2 pass VBR?

Take care,

[banjazzer]

In the setting for 2-pass VBR you can choose 2-pass or 2-pass(old type). It defaults to 2-pass.

[SatStorm]

Don't use 2 pass old type...

[W_Eagle]

Hello pals,

Thank you very much again for all your
detailed replies, assitance ang good will.

As I'm fairly new to this mater of video authoring,
I'd like to add one more question. Yes, I know that
only by my own experiments I'll get the answers
that are right for me. Nevertheless, I appreciate your
opinion, based on your experience.

Did anyone compare quality of encoding SVCD/CVD from AVI,
between TMPGEnc 2.56 Plus abd TMPGEnc 12, 2-passes VBR?

BTW, this website and forum, with their participants, are blessed.

Bless you

[Los11]

I Know I might be late and i did a similar test also. I saw that CBR was the best way to go for me. Also i change the GOP settings as well to improve the picture quality in the fast motion scenes. I use 3-I pictures 3-P pictures and 0-B pictures with 12 frames per GOP. with this setting i have created top quality SVCDs with the long wait. I used to do the 2-pass VBR and then i used it with the improved GOP settings also. What i got was choppy play. So, to me, CBR is the best way to go. Try out the GOP settings provided. It will mostlikely help your picture quality improve.

[W_Eagle]

Thanks pal.

Take care