Log in

View Full Version : TMPEG vs CCE SP


poopity poop
20th October 2002, 21:03
I would like to share with everyone my findings of a recent test I did basically comparing CCE and TMPEG. I encoded SVCD's of Kenshin Episode 82(not that it matters) 3 times.

1st time CCE SP:
CCE with progressive frames, zigzag , auto DC etc, the stuff in the getting the best of of CCE guide.
bitrate = 350-4000 ave 1820.
1-pass (which means multipass, makes the vaf, then encodes, basically 2 full passes)

2nd time CCE SP:
same as above except:
3-pass (which means 4 total passes)

3rd time TMPEG:
2-pass VBR(VBR) with same bitrate settings as above
Highest motion search precision
everything esle defualt.

To sum everything up: I encoded the same thing with the same bitrates, using the same average bitrate, just did 2 passes with CCE total, 4 passes with CCE total, and a normal highest quality 2-pass with TMPEG



Results:

CCE produced the same file varing by only 10K (one was 341,860K, the other was 341,870K). They were pixel for pixel acurate with each other. In other words, they were the same damn file, no matter how many passes you did. Yes I looked at high motion scenes, low motion scenes, flash of light and the frames sourounding them, even fading in from black(where it looks real crappy no matter what you do :) ). They were pixel for pixel accurate, no blocking during high motion or anything, very clean, very nice encode.

TMPEG produced a VERY high quality encode that(of course) was not pixel for pixel accurate with the CCE encode(being a different program, duh!). But there were no blocking, and no difference or concentrations of any more or less ringing artifacts present in any of the frames I looked at, then the CCE encode. The ringing was either more or less present in some scenes that CCE, but on the flip side, CCE has more or less ringing in some scenes than TMPEG..just being different encoders.

Conclusion:
CCE is faster, which is nice, but doesn't change ANYTHING from more than 2 total passes(1 pass setting in the settings dialog).

TMPEG produced a file that was EXACTLY same the same quality as CCE even on 2 total passes in CCE.

oh wait...CCE doens't maximize the dual processor loading very well, so TMPEG processes almost as fast at CCE. Making TMPEG( at least for those of us with dual processors) a leagl way to produce a high quality encode for FREE! Yes.. no one ACTUAL bought CCE SP, its all hush hush here abotu warez, but we all know...

Anyway...TMPEG passed my inspection for being just as good as CCE. And CCE disappointed me by producing the same file file pixel for pixel, and byte for byte(within resonable tolerance or error). For all those who think CCE i the mother of all encoding programs: try TMPEG on highest motion search precision. nuff said.

WarpEnterprises
20th October 2002, 23:06
Only for comleteness: what versions of CCE and TMPG did you use?

shinakuma8
21st October 2002, 00:53
my question is, how do you know they are pixel for pixel? did you use some kind of diff program to compare the bits?

poopity poop
21st October 2002, 02:47
CCE SP 2.5
and
TMPEG 2.58.44.152

No I did not use a pixel comparison. I loaded them into virtualdub first, went to a bunch of different frames at 4x, and alt-tabed quickly between the two instances. Then I loaded them into TMPEG, went to noise reduction filter, and turned it off, then went to a coupel different frames at 6x. I jsut used the noise reduction to zoom in on it.

Trust me, pixel for pixel

shinakuma8
21st October 2002, 04:15
well, I find your test hardly conclusive.

first of all, you have 10k difference in two CCE tests. how do you account for that? you can't claim that the two encoding are exactly identical if the output of the tests are different.

maybe the frames you looked at are the same, or maybe the difference is so small that it's hard to observe with your eyes. all that proves is that with your anime source that you encoded, it made no difference if you used 2 pass or 4 pass.

since you are using anime source, could it be that because anime are not as complex as movies, the gain you get by using 4 pass with a clean source is minuscule. have you done encoding with non anime source? source filled with noise? you have to show more evidence before your findings can be credible.

I'm fairly new to the svcd encoding scene, so I'm not going to pretend that I know everything. And I just made the switch from TMPG to CCE, so I am in no way a CCE advocate. But I have to see more conclusive evidence before I believe that there's no difference between 4 pass and 2 pass.

poopity poop
21st October 2002, 09:11
I completely agree with you.

I soppose my main point was that most people think CCE SP is such a great encoder, when tmpeg on the highest settings produces the same results, quality wise.

WarpEnterprises
21st October 2002, 10:56
However, I would like to see some diff examples of CCE 2 vs 4 pass.

I nowhere saw this yet, only assumptions.

htc10825
22nd October 2002, 16:32
I've created following "long play" DVD-rip with CCD 2.5 1-p cq-vbr

1. DVD "Storm Riders"(147min, NTSC, Taiwan/Hongkong) to SVCD(AR 2.24:1)on ONE 80min CD-R, with 1 audio and 2 subs(CVD).
the qulity is good and fully acceptable.

2. DVD "The Lord of the Rings" (180min, PAL, German) to xVCD (AR 1.85:1)on ONE 80min CD-R, with 1 audio and burned subs.
The quality (with burned subs) is EXCELLET.

With multi pass VBR in CCE I have had only somewhat worse quality.
With TMPGenc I got very bad quality. For such ultra low bitrate.

I use the Matrix of "Andreas SVCD", the is important. With other matrix I got mostly shit.

poopity poop
24th October 2002, 08:07
Well I'm not sure putting LOTR on 1 CD-R is such a good idea, even at such a low resolution as an XVCD. But the defualt for tmpegt is normal motion search.. which yeilds awful quality, agreed, did you do it on Highest motion search?

I'm only saying that TMPEG is just as good as CCE only on Highest Motion search precision. I've tried it on normal.. it does actually suck. Anyway...I don't have any proof, because I did my test and deleted the mpg's. If you guys want proof, you can do two things:
1. Do the test yourself..going into it with an open mind
2. Trust my judgment as an encoder. If anyone knows anything about my encodes, they are primo...I know my %&*# <--- and yes, I bleeped myself :).

htc10825
24th October 2002, 16:50
I've compared the quality of my 1-CD Rip with the 3-CD rip(VCD standard) of a friend, who did it under DVDx 2.0: my rip is has the same or even better than that of 3-CD rip. The resolution of it is not as good as DVD, but way better than VHS. If you have the luck of the owner of a SONY 100Hz TV(>=29 inch), you'll see the loss of qulity due to transcoding is very limited.

If you want to put it on SVCD using regular bitrate you have to make 5 SVCDs - much fun while changing disks! But with CCE I can put it on 2 80min CD-Rs - with 2 Audio tracks german/engl., 2 subs(german/engl. switchable) and navigations menu for all the 40 chapters.

Of course I've not done it on highest motion search in TMPG. The speed of that is far from acceptable. But I'm sure that even with the highest motion search you cann't get the results of above in TMPG.
TMPG is for somewhat higher bitrate very good. So for all the friends of 1CD-rip: go for CCE 2.50(2.66)!