It can be a problem, but not exactly the Kernel :!:
It could be a conflict with your binary image, because it was compiled with kernel 2.4.x and not with 2.6.x, which is what you have.
Kernel 2.6 breaks a lot of compatibility with the 2.4.x branch.

-kwag

OUCH !! :banghead: :banghead: :banghead: :banghead:

I must search a solution :evil: :evil:

@sentinell:
I think Kwag is right, should be a kernel problem.
You are not already using an AMD64 :?:
Maybe it is a libc problem.

@vmesquita:
Great job 8O :!:
I've planned to implement something like this but hadn't the time to do so.
I wasn't able to follow the mencoder tests and your tool is a great start. Thank you.
My results are great so far :angel: :mrgreen:

I wish have one !! No I have a P4 1,4 Ghz :(
I'm sorry but I didn't think it. I Know people who changed his kernel from 2.4.x to 2.6.x and hadn't recompile all OS(KDE, Xfree..).
If only was the kernel, the solution is easy: VMesquita lend me the source code :roll: , I promise not show to the public. And I wil post my results and my 2.6.x release.

Thanks

P.S Excuse me my bad english :oops: [/url]

Hi VMesquita,

1) I'm using template DVD to KVCD. Is there a way to override the
height and width crop settings that are grayed out in MEncoderME??

2) I may have asked this before, but is there a way to simulate
overscan??

3) I'm crashing using 2 pass and 1 pass gives me excellent picture
quality. Will 2 pass make a more accurate file size prediction??

4) My file size for video is way under (351MB vs 740MB as correct).
The whole movie is there and looks very good, but how can I
set vrc_maxrate, vbitrate, vrc_minrate, vrc_buf_size, keyint,
for more accurate predictions.
I currently have for 352X240 mpeg-1, vrc_maxrate=2000, vbitrate=(
bitrate calculator), vrc_buf_size=327, keyint=24, and vrc_minrate=300.

Thanks :)

-BP

Happy birthday my friend!!!

Hope you enjoy alone :drink:
...or much better with friends :drunkard:

:lol:

Thanks!! :dunce: :D

Actually, I don't think there's a need for that since the method I am using is working very well. The problem with GTK2 is that for some reason the status line of mencoder is not being detected properly, so the log goes on forever (instead of the status line being rewritten every time). But I think this is a problem with wxGTK use of GTK2 , since it's "experimental".

@incredible
Looks like the problem is not related to my GUI, then. :? But in my tests this never happen, I always get a file near to the asked bitrate in both passes. Happy Birthday!

@sertinell
I don't think it's the kernel, because my Kurumin Linux has a package to update to 2.6 and I don't think they would put a precompiled package that makes everythingelse stop working. :roll: I can try to install the package since it keep both kernels and test.

No there isn't. In the previous release they could be set manually but I grayed ou because I thought nobody would want to tweak this. I'll free this control then.
Right now no, because I graied out the controls. :oops: Unless you're using MakeAVIS, of course.
In my experience, yes, specially in fast motion scenes. The accuracy is about the same to me.
Sometimes mencoder just can't quantisize less than he's doing. It is already not dropping near any information (except for the scenes that could use more bitrate but max won't allow), so it just can't get the file bigger. If you're using a 352x240 with widescreen matherial, this is very possible. :wink:

vmesquita,
When using makeAVIS template, it encodes mpeg2. It would be nice to have the option to encode either mpeg1 or mpeg2. Or better create a template with makeAVIS and mpeg1. I know it's very easy to make a new .conf template with one's settings. Just tell you if you want to add it to your next version.
A way to solve this may be adding noise, because we cannot raise maxrate when encoding KVCD.
Somehow, in my tests (maybe because encoding mpeg2) mencoder is blockier than TMPGEnc when encoding with makeAVIS and aiming KVCD. These are my first tests with mencoder for KVCD. I think we can try to improve this (maybe higher scplx_mask value,...)

@digitall.doc
Nice idea (mpeg1 MakeAVIS), I'll add that to the next release. It's very simple, anyway. :D

@ BP

I had the same problem

my first encoed was 352x240 and got only 299 mb of video for an 89 minute movie

Not glad its happining to you but i am glad i'm not alone because i thought i was doing something wrong :D

when i do avi to kdvd, i choose 720x576 but it says cropped size is 720x588
this is not standard dvd size and looks crap on stand alone player, i had 6 episodes of hustle at 640x368 8O the first four encoded to 720x576 but the last two don't (588 not 576)

any ideas?

bloody great gui by the way

@fozzieb
Are you sure all the source movies had the same width/height? Because I don't see how this could happen if both movies have the same width/height. Could you please confirm this?

it looks like the first four were 640x360 and the last two are 640x368

so it looks like that this is the problem, any ide how i can change the res quickly before i open in MencoderME?

cheers

Right now, cropped resolution is locked. I'll unlock it. :D
Actually, the problem is that your source has a distorted aspect ratio: it should be minimmum 1.7777 at the most (for 16:9 encoding), but it's 1.739!

This happended because the person who did the AVI decided to do macroblock optimization. To do that, width and height are a bit distorted to be divisible by 16, optimizing compression. But this also created a "bigger" than 16:9 A.R. movie. :roll: I can add a check for this. :wink:

Hi Vmesquita,

well ... do you use the build you do provide within the downloadable GUI mencoderME package?

I have the mencoderME.exe AND the mencoder.exe in the same folder, ... also that cgywin1.dll .... this folder is also settet in my systemVariables settings ... Ill do make an extra Folder somewhere else which will contain ONLY MencoderME.exe, cgywin1.dll, The Templates AND the mencoder.exe (your last generic build) ... so MencoderME GUI will look for mencoder.exe and will use ONLY that one in the same folder. :idea:

NOW IM GETTING INTO CRISIS!!!! :evil:

Even when using 2pass :arrow: never will get above 2000kbit avg eben if 3500kbit is set as bitrate :cry:

I tried all my builds of mencoder I have 8O Whats going on on my Machine :screwy:

Probably mencoder just "can't quantisize more". Happens sometimes.Check with BV, probably the average quantisizer will be very close to 2, it just won't be exactly 2 because sometimes there're bitrate peaks and mencoder needs to quantize more to respect max bitrate. :wink:

EDIT:Forgot to answer you last questioon. MencodeME calls mencoder without specifying path, so it will use the mencoder build in the same folder. I think that if it doesn'r find, then any mencoder.exe in the path will be used, but I not sure about that. :?

Interesting!!!

No matter if source is VOB or Makeavis (treid 4 diff. sources) --- it wont get above 2000kbit! (as said the weeks before I could get above) .

NOW I did an example and did set to Bitrate 400 and ..... it ends up in 1pass at 401!!! (Hell that 1pass is veeery exactly 8O )

AHA!

Id did ad at the end of the Matrix values of the config file a

:vb_qfactor=0.7:vi_qfactor=0.8

Which lowers the quantisation and at least it goes up to 3300 when set 5500 ;-)

Another test using this adding by setted 2500kbit! and 2489kbit resulted (1pass) .... now lets see how the bitrate peaks will behave :)

@Incredible

I set scplx_mask=0 in MakeAVIS.conf and my files end up being a couple MB's over, very accurate. I am doing TV caps, about 43 minutes per CD, 704 x 480 IVTc'd mpeg-2. I set vrc_maxrate=3000, desired average bitrate usually around 2400. A setting of 0.3 for masking is rather high but is helpful when doing low bitrate one CD movie encodes. Try something like scplx_mask=0.03 and see if your bitrate goes up.

@Vmesquita

A big thanks for a job well done. :D

Looks like there should be a "low bitrate" template and a "high bitrate" template! :D Nice discovery! 8)

Actually adjusting the spatial complexity masking had already been mentioned in another thread.

http://www.kvcd.net/forum/viewtopic.php?t=10391

I can remember using a value of as small as 0.1 back when I was using ffvfw and my filesize would decrease by as much as 25% 8O

Are MencoderMe results equal or better with internal Mencoder filters than with avisynth optimal scripts?
Does internal Mencoder filters reduces final filesize?
How mencoder file size prediction or calculation work while it is encoding?

@VMesquita & Dano

Well I dont know if still two templates are nedded, cause that "dicovery" I made already many weeks ago. It reduces also the "barrier" of B frame quantisation. And makes sense as I still can use Vbitrate and it results almost accurate on 1pass - as you see above.
Cause Dano, I dont want to get bigger filesizes generally, I want to kill that resulted avergae bitrate limit ;-)

@ Prodater

Yep, they do lower filesize as they do treat noise etc. like avs but we have to figure out what will be better avs or mencoder filters (I did post in here a screen where also postprocessing parameters are explained).

Prediction: I do set Vbitrate to the wanted avg bitrate and do a slicer() encoding ... in 95% of cases it matches veeery near at the first run, and then you do fine adjust by using another bitrate untill it matches. When using mencoder I do need max 3 pred. turns (ping-pong-ping)

GOP and Resize
 

Congratulations for the MEncoder-Me.

Now, very easy question:

¿What is the GOP for PAL KVCD or KSVCD? ¿15, 18, 24 or 25?

A curiosity:

¿How about of internal MEncoder resize? ¿Lanczos? ¿Bicubic?

Thanks.


-Maurus

Re: GOP and Resize
 

15 = PAL KDVD
18 = PAL KDVD

24 = NTSC KVCD
25 = PAL KVCD

as you can see the sws parameter, there the interpolation method is determined by a value i.e. 9 .... go into the mencoder readme and see whats about the values.

just compiled my own mencoder.exe under windows for athlon and i works great, on tests over 3 sources i get an extra 3-5 fps on my encodes .

@vmesquita

keep up the good work on this gui it rocks :D

ahtlon and mEncoder-Me 0.2
 

I can't use MEncoder-Me 0.2 in my Athlon 900 mhz? only over Pentium?

:(

Thanks

-Maurus

Re: ahtlon and mEncoder-Me 0.2
 

Yes you can... I have a AtlhonXP 2000+! What is the problem, exactly? 8O

I say it for this thread, where several people are compiling MEncoder for Athlon.

He is better (very fast) to use the version offered by Krassi or another version of MEncoder for Athlon in MEncoder-Me 0.2?

Thanks.


-Maurus.

Sincee you have an Atlhon 900, you need to use a generic build. Atlhon builds are actually for AtlhonXP that supports SSE. MencodeMe comes with a generic build :wink:

Ok, I will try this nigth MEncoder-ME 0.2 for make KVCD or KSVCD from a DVD.

Thanks.


-Maurus

Is that compile done frome the latest CVS? And ... can it handle makeavis generated fake-avi files?

If yes :arrow: What did you use MinGW or CygWin ? And how did you set it up to get makeavis working? (I changed the codecs.config to VMesquitas one before compiling but finally that resulted mencoder.exe will not encode makeavis sources.

Thanks for your reply
Inc.

PS: I heared, that the newest CVS gots a fix that when using :Scale= ... it wont change the colorspace to RGB anymore .... is that right?

@incredible

It was the latest cvs MPlayer-20040421.tar.bz2

I used MinGW to compile, I have not tried a makeavis avi yet, going to try tonight

I think the build included with MEncodeMe doesn't do that anymore, look at the logfile:

damn it, doesn't encode makeavis files

Check my last post here:
http://www.kvcd.net/forum/viewtopic....r=asc&start=32

i'll try that just now

Yes Dano, scplx_mask does "I kind" of filtering. In man_page it's advised to deblock when no other deblocking filter is used.. But if you use makeAVIS, that is, if you frameserve to mencoder through avisynth a already filtered image, I don't find the need to filter else with scplx_mask.
When encoding directly from unfiltered source (like a vob), even more if no filter is applied at all, I'd advise to use it at least at 0.2
Well, inc, you're right. By default, if I frames get a quantizer of 1, B frames will get (I*1.25)+1.25=about 2.50 quantizer... And in a GOP we have at least (when GOP=15) 10 B frames for each I frame... With your method you gave B frames a minimum quantizer about 1.50.
I'm still struggling with my microblocked encoding of StarWars II. Postprocessing helped a little, as helped filtering. Raising scplx_mask also helped. But they're still there, mainly in dark flat surfaces, but also all over :evil: . I thought it could help lowering quantizers...
A way to lower quantizers when vqmin=1 and mbqmin=1?. Try lowering lmin, for instance lmin=0.75... quantizers lower. Of course, lowering B quantizers also is of a help.

I did set lmin=0.1 .... by this I can go up to approx. 4000kbit avg! if I want ... and if I set lower to i.e. bitrate 3000kbit .... I get a 3009kbit avg resulted file in 1pass mode .... sexy! I think your way digi.doc a) makes more sense and b) does it well the same!

thanks! :D