TMPGenc Vs. CCE ?

[q1aqza]

I agree with XED comment in principle, however, at risk of the wrath of the moderators, who the hell on this site has actually paid for CCE??

[Oracle]

Hi Satstorm

I am kind of interested where you get your figures from

For example

Anyway, CCE is better, let say about 05 - 10% TMPGenc in picture blockness.

or

There is a 5% difference and it is identical the High Quality mode

I´ll have to give motion search estimate a go, as I usually use normal or High quality.

[Oracle]

Q1qaza -good point, I would say no-one on this board if everyone was hosest has paid for CCE. Guess everyone is using the trial version

[banjazzer]

My tuppence worth. I did my own tests a while back (TMPG 2.53 & CCE2.5) and I concluded that TMPG was better quality for my camcorder DV encoding using CQ settinglong movie the speed advantage of CCE is fantastic.

hehe. Looks like the reports of the horse's death were exaggerated. Unless it was not a horse, but a phoenix. 8)

[Thorn]

Unless it was not a horse, but a phoenix.

Don't try to resurrect your pets by building a pyre, though... it doesn't work but it sure does stink...

SatStorm: What kinds of settings for TMPGenc do you think are important for using 2-pass VBR (using the latest version of TMPGenc plus)?

[Nelson37]

It's an old debate, but I will inject a few points.

1. As the two encoders use different algorithms to acheive their goal, it is difficult to produce equivalent files to compare. If you specify Min - Max - Avg as 500-2500-2000, Tmpgenc will produce a file with Avg slightly over 2000 and Max slightly over 2500, while CCE will produce a file with Avg slightly under 2000 and Max well under 2500. Increasing CCE Max value to 2800 - 3000 will produce a file with values very close to that of TMPGenc.

2. CCE is designed to do the first pass (VAF) at values dramatically higher than target bitrates, say 1000-8000-4000, then reset values to 500-2500-2000 for the subsequent 2-3 passes. This reduces the "smoothing" effect many have noticed, and again, produces a more equivalent file for comparison to TMPGenc.

With multiple versions, different source types, and different targets to achieve, comparing the two on equal footing is difficult. I find CCE faster, and better for CVD, SVCD, while TMPGenc has more tools, seems better on VCD (which I don't use much), and seems to handle interlaced source better.

[banjazzer]

Don't try to resurrect your pets by building a pyre, though... it doesn't work but it sure does stink...

I don't keep pets. I work/have worked with TMPG CQ, TMPG 2-pass VBR and CCE - all giving (to my eye) superb results.

However, I love it when the zealots get defensive.

Just goes to show, not all the Taliban were in Afghanistan.

[Kdiddy]

This is most resilient horse nature has ever made....(as I swiftly kick the horse in the nuts).

[banjazzer]

I suspect the jury is still out on the outcome of the autopsy!

[SatStorm]

there are demos of CCE if you want to test that proggy...
They simply add a stamp in the corner. Ain't bad for tests...

Oracle, I don't understand this expresion you are using, my english is far away from perfect!
You use the normal mode of TMPGenc? You are the first man I know doing this! Motion estimate search is far better! That' s why tmpgenc suggest it when you first load a SVCD template

About 2 pass vbr... I am going -x- all the time. 1000min-2000average-3000max gives me excellent picture. Some times I have to rise 1200-2500-3800 and to very - very difficult VHS source, I go even more: 3500 average... I have identical VHS/SVHS that way on my CD (by the way, let me tell you the bad news... My SCSI CD-R just broke down... Shit, I have to buy a new one...)

For standard CVD/SVCD legal bitrates I have excellent results only with 16:9 stuff (DVD or some DVB transmissions). Let say 1200 - 1880 - 2520
There is a tip for TMPGenc, well known years: Never set the lower value less 1200 for SVCD. For CVD this "rule" is ~1000kb/s

[banjazzer]

There is a tip for TMPGenc, well known years: Never set the lower value less 1200 for SVCD. For CVD this "rule" is ~1000kb/s

Thankyou. Useful information which is worth knowing.

Motion estimate search is far better!

I must admit I have never used this, as I assumed the mode which took the longest would be better! You say *Motion search precision* set to *motion estimate search* is better than set to *normal*. Where does the *High quality* setting fit in to this?

[SatStorm]

It is really difficult to me to expain it in English.
Motion estimate search mode is a different way to encode. It is like a shorcut method. Most of the times works fair enough. For sure it is better normal mode and at least is equal the high quality mode (most of the times I find it better, almost as good the Highest quality mode). It is also really fast!

Oracle I just translated exactly your reply. Seems like you don't like what I use to post or you doubt it ? Well you are not the first one for sure! If you have better conclusions to post, then do it. I'll be glad reading them. Maybe I will expend my knowledge!

[BJ_M]

i could see those min. values as pretty well on the mark as i found for dvd don't go below 2000 and if very high contrast or space background i accually increase this

[Thorn]

Ok... using sufficient bitrate isn't exactly a secret but it really is forgotten all to often (sometimes from prolific individuals who utter things like "dude, I got 600 hours on a cd using realmedia!"). With DVD that's less of an issue than with SVCD, since you can comfortably fit several hours at over 4Kbps so people get less stingy. I personally have tried to encode at around 5k-6kbps so it sounds like I should be OK with TMPGenc if my source is ok.

[Oracle]

@ satstorm

I was just trying wanting to to know how you arive at your statistics, thats all.

Interesting what you have to say about Motion search estimation, will do a little more testing on this.

Why set the minimum to 1000/1200 why not 0 or 300?

[SatStorm]

Hi oracle,
It has to do with the minimum bitrate requared by TMPGenc to support a mpeg 2 still shot if I remember correct. Also, helps shapes for a reason.
For SVCD this value is about 1200min, for CVD is about 1000kb/s in theory, but in praxis 1200 min still produce better picture IMO.
For full CCIR - 601 I think the minimum is about 1550kb/s
The same rule exist with the DVB European Transmissions.

This tip don't effect DVD standalone players with good mpeg decoders. It is for the others, the cheap ones. Some of them don't even play mpegs with a minimum value less 300-600kbs and that's why is that "enouble padding" fuction to TMPGenc.
Basicly, this tip helps a lot multipass encodings.