CCE vs DivX

[DVDHack]

I couldn't really find a good forum that crosses these topics but I use CCE so here I go.

I am curious if mpeg4 (DivX) is better than CCE in terms of quality universally or just at lower bitrates. If CCE is better at higher bitrates - where abouts is the bitrate breakeven point - assuming that it is accepted that mpeg4 is better than mpeg 2 at lower bitrates.

I have found that mpeg4 seems to appear worse than CCE mpeg2 (to my non-expert eyes) when I put a DVD onto a DVDR, it still seems to stack up at two DVDs an a DVDR but starts to fall off after that - but I have no real conclusive evidence and really don't know how to go about the test.

I'd like to see quality comparisons between CCE and DivX (or any mpeg4 tool) at various bitrates. Has anyone seen such a test?

[RB]

Recently there were tests of various MPEG-2 and MPEG-4 encoders in the well respected German c't magazine (http://www.heise.de/ct/english/) (article is not available online). This was not a head-to-head test between the two MPEG formats but MPEG2 and MPEG4 encoders were tested in separate articles in two separate issues. However, they used the same test videos (http://www.its.bldrdoc.gov/vqeg/) and approach to measure the difference in quality between original and encoded video in both cases. They used a special software called JNDmetrix (http://www.sarnoff.com/products_services/video_vision/jndmetrix/index.asp) by Sarnoff corporation. This software uses a "Human Visual System Model" that is less based on mathematical algorithms but pays more attention to what human eyes will quickly identify as artifacts. It calculates a "JND index" (Just Noticable Difference) where a lower value means better quality.

The conditions under which the encoders were tested were not equal of course as the following table will show but I picked a test were at least the video resolution was nearly identical.

Encoder | Video | Interlaced | Bitrate | JND index
| | | kbps |
----------------------+-------------------------+------------+---------+-----------
DivX 5.0.4 | NTSC 720x486, 29.97 fps | | 1300 | 5.86
----------------------+-------------------------+------------+---------+-----------
XviD (Koepi 05/04/03) | NTSC 720x486, 29.97 fps | | 1300 | 5.122
----------------------+-------------------------+------------+---------+-----------
CCE-SP 2.65 (??)* | NTSC 720x480, 29.97 fps | X | 4000 | 3.717
----------------------+-------------------------+------------+---------+-----------

*) They actually tested CCE-SP 2.65, I exchanged some email with the author of the article and he said that was the version they got from Custom Technology. The current CCE-SP on their web site however is 2.67 beta and before this there was CCE-SP 2.66

As you can see, the comparison is not quite fair because CCE-SP had to struggle with the original interlaced source while they deinterlaced it for the MPEG-4 codecs. But even if we assume that JND for CCE-SP had dropped to say, around 3 for deinterlaced material, it's IMHO clear that MPEG-4 should beat MPEG-2 at any given bitrate.

But of course in reality most people are still compressing 2 hours of film onto one CD with the MPEG-4 codecs and lucky owners of DVD burners can use much higher bitrates for MPEG-2 :) But that of course is not an argument. I think MPEG-4 and/or H.264/AVC is the standard of the future but it might still take some time because again the upcoming Blu-Ray technology is specified for MPEG-2.

I must also say that I can't provide any personal visual experience with MPEG-4, IIRC I looked at some MPEG-4 compressed material only once when I was playing with Vidomi's XviD, but that went away when I last rebuilt my machine. Honestly.

[Kika]

I have said enough about that stupid Article in German Boards, nothing more to say. It's a joke, not a test.

But you can't compair interlaced Video with progressive Video. And: if you are encoding interlaced Video, you must add 1/3 more Bitrate.

DivX IS better than MPEG2, no doubt, but on the other Hand: Using DVD-Burning for DVD-Ripping, you can use spezial Tools with nearly no quality loss. And that IS an Argument.

And talking about the Future: The MPEG4-Encoding maschine we use now are not the same we use in the future, right? ;)

[RB]

Well, I would have loved a detailed list of each encoder's settings rather than the "we used the highest quality settings" also, but then that would have easily added maybe 10 pages to that article. Have you ever played with MainConcept Encoder? It has dozens advanced settings.

Just send him eMail and rant :). I did that, pointing out that they missed that CCE-SP of course does offer manual I-Frame insertion and what caused that line-shift with CCE-SP and CCE-Basic (a classic case of how people get confused about "Upper Field First" :) ). I got a prompt reply and we exchanged some more eMail.

[Kika]

@RB

Oh, i "played" with all of those Encoders, that's because i know that the Test is not really correct. Do you know TMPGEnc in Detail? There are lots of settings and different Encoding-Modes (CQ, CQ_VBR and so on). What dit they use? The Author did not write anything about that.
Oh, and i bet, he didn't know the PAL-Bug of TMPGEnc.

And what's about ProCoder? Did they use Field-based or Frame-based interlaced encoding? That's the special Mode of this Encoder, but the Author wrote nothing about that.
That's because i can not name this as a Test.

[RB]

Like I said, I agree with you that they should have given more information on the settings used. But it's not the reviewers task to know and work around bugs. I know virtually nothing about TMPGEnc. But we are getting OT here.

[RB]

Bach, I've now exchanged some more email with the author of that article and he is sorry for making a simple typo. The version they tested was CCE-SP 2.66 in fact (and CCE-Basic 2.67 which was mislabeled as CCE-Basic 2.66 as well).

Kika, you may want to check out the DVD-SVCD-Forum Thread (http://www.dvd-svcd-forum.de/phorum/ikonboard.cgi?;act=ST;f=13;t=13024;st=20#postid116491) where I posted (with their permission) some emails from Holger Dambeck and Volker Zota, also regarding their "One-Click-Tools vs. CCE" test they performed in the Compressed Domain Transcoder Comparison in the same issue.

[DVDHack]

It would appear, in short, that there are no valid tests comparing mpeg 2 and 4.

I will do some testing of a source with various settings of bitrate. I don't realy think I'm capable of doing a true test but I should be able to do a "in my opinion" test. Does anyone know of a proper testing methodology that I could adhere to?

[Kika]

@RB
Ach, Du warst das... ;)
IF you like, we can discuss there. My english isn't good enough for technical discussions.

[RB]

Ja, ich war das :)

OK, let's continue discussion over there.

[mmatac]

@RB (the MOD)

if you want at the end to summarize the discussion on English doom9 forum, we appreciate it.:mad:

[RB]

What's that with the mad face? After all, this was an article in a German magazine and Kika already said that his English is not too good. Please think twice which smilie to use next time. I'll try to summarize later.

[symonjfox]

IMHO MPEG 4 is better than MPEG 2 just with some options.

Let's begin: MPEG 2 is horrible at low bitrates, but it's GREAT at High res using High bitrates. MPEG 2 was designed for DVB and DVD so there's a lot of bandwidth with them.

MPEG 4 can be fantastic just if you have some features enabled.
As referiment I won't use DIVX because it is few customizable.
XVID can be set for low bitrates/res and high quality stuff.

For low bitrates (1 CD) you must use H.263 matrix (that will soften the image), use B-Frames (2 or 3) and you could try QPel (that sharpens a little bit, but it's quite buggy). I'd enable VHQ=4 and other stuff (read guides in Xvid forum).

For high bitrates (2 or more CD) you must use MPEG matrix or (better) CUSTOMIZED MATRIX (download some examples in Xvid forums; I'd suggest HSV_BEST and PROFESSIONAL for very very better quality). You can use B-Frames if you want (none or 1) and nothing else (that will decrease quality). Maybe use VHQ=1 to get the top quality (will take longer to encode).

Quantization matrixes are the base of your encode. If you use H.263 matrix for 2 CD encode, you won't catch the best quality as you can) while if you use MPEG quant in 1 CD encode, you surely get a worse quality.

[mmatac]

-------------------------------------------------------
What's that with the mad face? After all, this was an article in a German magazine and Kika already said that his English is not too good. Please think twice which smilie to use next time. I'll try to summarize later.
-------------------------------------------------------

The mad face is related to the singular mod's behaviour. A mod has to discuss in the same forum he is the moderator.
It's simple also for you.
:devil:

I'm not interested to the fact that kika has problem with english. Also I. But I try.
You, think twice before write.
The smilie is used with the meaning of "furioso" that in your language is "verärgert".

Because the thread is interesting, it is a good thing for the moderatore to discuss in mod's forum and not in other forum.
It's simple also for you:)
Summarize.

[RB]

mmatac, take it slow. I'd agree with you if you were right. Reading the previous posts you'll realize (although it's probably not easy to get because we got OT) that the discussion in the German board is not about the CCE vs. DivX issue but about a MPEG-2 Encoder and One-Click-DVD-Backup tools comparison published in another issue of that magazine. So in fact there is nothing to summarize here (hell, even I got a bit confused what we are talking about here after all, sorry).