Low bitrate matrix. Your opinion needed

[markrb]

I was reading the CCE 2.64 user guide and according to it if your bitrate is below 4Mbps you should use the Low bitrate matrix and not the standard matrix.
What is your opinion on this?
Should we just use the standard matrix or change to the low bitrate matrix?
I plan to try some tests, but I would really like to know what others have seen with their own eyes.

edit: I just did a test encode with CCE's 2.62 low bitrate matrix and compared it with a normal matrix and I liked the normal matrix much better.

Mark

[chainsaw135]

I'm going to go ahead and do a test myself with the movie Bones and i'll post back with my findings..

Edit: ok i'm done with the test and after a careful look i noticed that the "low bitrate matrix" looked better then the standard.
here are the settings i used.

Encoder tab settings:

Image Quality Priority: 17
Anti Noise Filter: 2
Field Order: automatic
Linear quantizer scale: Checked
ZigZag Scanning Order: Checked
Progressive Frames: Checked and all the rest left as default..

Framserver tab settings:

Resize Method: BicubicResize
Bicubic b 0.00 c 0.60
Temporalsmoother: off
CBR mode
Averate bitrate was 2230

[m3taPT]

Just finished testing with a CCE 5pass (man that takes ages on my box).

Burned the 2 and compared them carefully. Unless my eyes are getting blurry due to aging ;) the normal is superior. And i could see that since the start until the end. No doubt about it.

[orbit-r]

hi
for me is the standard matrice better than the low-bitrate matrice(don`t know why since we`re making low bitrate encodings)

but i think the best is the andreas svcd matrice which i use all the time (with the correct settings figured out in bitrateviewer and rempeg[bitrate]and 5 pass, sometimes you can`t see a difference between the dvd and the svcd)

btw. im very happy about that dvd2svcd had included the matrix muxing feature (bbmpeg) since 106 build9

gruss
orbit-r

[markrb]

I just tried the SVCD matrix and I thought it made the image a little too dark. Otherwise it doesn't look bad, but I don't like the darkness.

Do you change the andreas svcd matrice in any way from standard before you encode?

Mark

[orbit-r]

no mark normally i don`t change the andreas svcd matrix
but a couple of weeks ago i made some tests with this:
intra:
8,16,19,22,26,27,99,99,
16,16,22,24,27,29,99,99,
19,22,26,27,29,34,99,99,
22,22,26,27,29,34,99,99,
22,26,27,29,32,35,99,99,
26,27,29,32,35,40,99,99,
26,27,29,34,38,46,99,99,
27,29,35,38,46,56,99,99

nonintra:
16,17,18,19,20,21,22,99,
17,18,19,20,21,22,23,99,
18,19,20,21,22,23,24,99,
19,20,21,22,23,24,26,99,
20,21,22,23,25,26,99,99,
21,22,23,24,26,99,99,99,
22,23,24,26,99,99,99,99,
23,24,25,99,99,99,99,99

the intra settings from the very low bitrate+the nonintra from andreas.... did`t look bad too (as far as i remeber)
maybe this makes the picture not so dark
time for me to test it again
:confused:
what do you think about the pic.quality of the standard versus andreas svcd excluded the darkness? (sharpness,blocks and so on)?

gruss
orbit-r

[orbit-r]

hi
got some news about the martix thing
1.when you use the older tsunami patcher and load for example the andreas svcd, you got the settings in the intrafield (there its right)
and in nonintra1 field.but thats wrong !!you have to put it into the non intra2 field!! and let the nonintra1 at default otherwise cce wiil take his standard matrices..
but dl the tsunami patcher028 i believe there its fixed..

2.bbmpeg do not mux the matrices (which dvd2svcd selects automatic)
into the final .mpg stream (i believed so). this feature of bbmpeg
is only for the bbmpeg encoder.you can check what matrices your .mpv
stream has with this programm:ReStream and then change it to what matrice you like... heres a link:http://mitglied.lycos.de/fitcd/ReStream_v080.rar

EDIT:
im not sure about what ReStream says about the .mpv stream on my tests it always showed standard matrice but its possible that if cce2.5 is patched correctly restream thinks that this is the default matrice... but definitly it writes the matrice which you like to the.mpv stream (the filez are little bit bigger after adding a new matrice which is not so if you do only bbmpeg matrice muxin)

i got this info from a german board where tsunami and ssh(the programmer of restream)discusses this thing...

gruss
orbit-r

[markrb]

I have only been using new matrix's with CCE 2.62 and not 2.5. I have never patched it. Maybe that is why the image is too dark. It might have something to do with VFAPI and not the matrix itself.

Mark

[orbit-r]

hi
don`know if this has something to do with the VFAPI
i always use 2.5 with avisynth ...
but seems very interesting that (i think most of the dvd2svcd users use 2.5 with avisynth) patching it with the older patcher will have no effects cause cce is taking always the standard matrix (and muxing with bbmpeg don`t have a effect too)

i did a movie encode last night(cce2.5 with oldpatcher patched matrix) today i found what i wrote above ,now i`ve added the matrix with restream into the .mpv file of dvd2svcd renamed it,and startet crashrecovery ...went fine..

gruss
orbit-r:D

[markrb]

Ok I ran the patcher and put in the Andreas SVCD Matrix with CCE 2.5 and I compared it directly against a video I encoded under normal and I couldn't tell the difference. I did not change the way BBmpeg muxes though. I let DVD2SVCD do it's normal thing. I am no expert on BBmpeg. What do I have to do to change the way it muxes?
I also didn't run anything, but the patcher. I did not run restream. Could you clarify what this does and should I try it and what program does it patch? I tried reading the included documents, but most are in german.

edit: I did some more tests with different matrix's and I couldn't tell any difference between any of them. Maybe I am blind?
I used the regular CCE 2.5, Very Low Bitrate and Andrea's SVCD.

Thanks,
Mark

[orbit-r]

hi
mark the thing is(which is new for me)that it seems cce 2.5 don`t write a matrix to the .mpv-steam (maybe its encoding with one,but don`t write it back, don`know) and there for you will see no difference

here`s why i believe this:I`ve checked the size of my original .mpv created by dvd2svcdandcce2.5 than loaded this .mpv into restream(with it its possible to really write a matrix into .mpv`s)
and put the andreas matrix into it.. the new .mpv file was a little
bigger than the original(this is the matrix)

a time ago there was a big thread about matrices ...there some guys done hex compares(riht word?) between cce2.5 patched and unpatched
they didn`t find a difference now with restream you can see easily if you added one

after some readings my opnion is bbmpeg and its feature to add marices is only for the build in encoder iwould change nothing in muxing

here is how restream works:load the .mpv simply in it
choose your intra and nonintra(out of the cce2.5 folder run when its finihed maybe delete the original and rename the new.mpv to the name dvd2svcd will take for crashrecovery
at the end you should have svcd`s with a real matrice on it..

after all im not an expert in matrice things(i don`t know which for zigzag or alternate scanning order..) but this way you can really do
a camparision between the different matrices on cce 2.5
btw.maybe a suggestion for dvd2svcd to add ReStream inro dvd2svcd
(which i think is possible) cause when this what i wite is right(hope so)we only have standard matrices on our svcddiscs..

maybe you guys do some tests yourself

im very interested what you think about that!
gruss
orbit-r

[markrb]

Unfortunately I have come across a problem for me.
I can't use the patch and still encode the way I like.
CCE always blows up. I uninstalled the patched CCE and
installed a fresh copy and it worked OK for me.
I would have liked to tried more, but I like what the
Sharpen filter does way too much to abandon it.

Mark

[orbit-r]

hi
mark do you still have the problem with the patcher since using avisnth107?
cause with avisynth105 i was not able to get cce2.5 working with the
sharpen command in the avs script but now with avisynth 107 everything
runs fine..(god bless dvd2svcd)

gruss
orbit-t:D

[chainsaw135]

@orbit-r It seems with that new avisynth.dll the problems with SHarpen crashing are fixed, were hoping this also will be fixing most of the other issues with cce crashing i guess testing for the next few weeks will tell given the users with cce crashes post that there problems are still there or have gone away.

[delly01]

Hello,

where did you get the new avisynth? I did not find it on videotools.net, and a search with google did not help, either :-(

Bye,

Detlev

[chainsaw135]

You can find it here http://forum.doom9.org/showthread.php?s=&threadid=22696&perpage=20&pagenumber=1