Our objective is to fight to each other. It is inevitable.
<-->

Can you, anyway, give me that source, then i could work over same sample that you and make a more realistic comparison.
Thank You.

How do you want me to give you a source ? I use a DVD ripped on my disk ! Do you have a 7 GB storage somewhere where I can post it ?

I also have an other source that I use for all my scripts comparison (but that I didn't use for this MCE<->Tmpgenc yet) but it's also a Vob that is 700 MB long. I can't send that to anyone.

When you ask for a sample to work with by yourself, you think about an avi. I'm sorry but when I see the results I obtain with a clean DVD source, I won't even try an avi.

Excuse me, I thought you could trim a vob.

Even 30 second in VOB take several dozens of MB. Try it : use ChopperXP, it is a really good vob trimmer.
The 700 MB vob I use is only 5min12s long !

Thanks, I'll give it a try.

Well, then how can I obtain a sample with tmpgenc with similar or equal size than one obtained with MCE. Should I use trim in the script? to get same number of frames for both tests?

Matrix reborn

In fact I use the command line "Sampler(length=25)"
(25 because I'm in PAL) at the end of a script. This generates a sample of 1 second per minute of the movie. In other word I have a 100 second sample for a 100 minute movie.

I encode this sample and I adapt the CQ of tmpgenc in order to obtain the same size I obtained with MCE (as I don't know how to adjust the file size with it I guess that I just have to change the AVG bitrate to achieve that).

Currently I'm using 450-6000 as bitrate range and 1800 as average in MCE. The equivalent CQ under TMPGENC was 68. Both encoded files were 26.5 MB in the end and that is how I can start to comapre the results.

And for the moment, MCE results are really too bad.

Maybe, but in this way:

First at all was "The Matrix",
Later, came "Reloaded",
and the last one "Revolutions".
Don't loose it yourself, coming soon...
<-->
"Matrix Reborn"

Just try CQ3, like I said before.

One of the forms is changing the avg btr. But MCE have many others, as you can see in the next screenshot (file names are showing settings). They works with VBR or with CQ encodings.

@Prodater64,

These are the main reasons why we don't use MCE.
(1) In order to work in CQ mode, you must set the MIN bitrate to zero. This renders the encoded mpeg useless on many standalones that don't support MIN bitrates below a pre-determined level.
(2) The resolution of the Q ( 1-31 ) is not granular enough for a correct prediction, again, rendering the accuracy incorrect almost every time.
The steps are to broad. Not like TMPEG, where you have decimal accuracy.
(3) For low bitrates, it's just not good enough of an encoder.

We consider all of these to be design flaws, and they have never addressed them, even after people have asked for the options. It seems MC is a very narrow minded company.

-kwag

Okay, it will be helpfull

Kwag, I know your reasons, I was reading all by this forum.
I'm just working with MCE, investigating it, and I think, I found out many tweaks that not was founded here before. I want to share those with everybody here, You, Dialhot, Inc, Jorel, etc. because you (they) have more knowledges than me and for this reason you (they) can make further tests that I can't.
I know too, that developers need to solve these design flaws, and I think that they are working on.

About your points:
1 - I think (for casual observations, not for testing it specifically) that min btr don't drop so down when you begin with DVD material.
The fact that some standalones don't support these min btrs would make same sense that the fact that many standalones don't like KVCD/KDVD.
(please, I hope that you know what I mean)
2 - I'll test it with my tweaks and Incredible's ping-pong prediction method, and later I'll post the results.
3 - Developers needs to solve that.

See you.

Okay,

Unfortunally I have to stop my tests on MCE. I didn't do them for my own use but just because I gave my words to prodater64. But I have way too much to do to make further tests. I even loose the result of my previous ones

What I can say is :

- whatever I did, MCE gives too much gibbs (mosquitos) effects arround edges. There is surely something to do but I didn't find.
- on plain surfaces, MCE is a little better that TMPGENC (less DCT blocks) but, to be honnest, the gain from this point is far below the loss due to the previous one : mce video are really hard to watch.
- according to time : MCE took 820 (13.83 min) seconds to do my sample, TMPEG took 16.5 in setting "hight" and 10 in setting "Motion Estimate" (but in this case, results from MCE are a little bit better)

I used every settings you suggested and targeted a 1.7 GB MPEG2 video (half a DVD). MCE won't be my future encoder

Thanks for your opinion and your work Phil