Go SECAM :!:
Hey Hydeus: are you using both ffvfw an ffdshow or just ffvfw?
How about the speed and quality compared to tmpgenc without Notch Matrix?
Regards

Hi RDS.
Read edit above :(
I use only ffvfw.
And as for test with TMPwithout notch matrix, I'll do the test.

:idea: :idea: :idea: :idea: :idea: :idea: :idea: :idea: :idea: :idea:
:!: :!: BREAKING NEWS :!: :!:
You can't seem to be able to use custom matrixes...
But that's only with MPEG1!
Try changing to MPEG2 codec, use quantization type MPEG, check custom matrixes box, and voilá!
The custome matrixes options are now customizable.
So, now I only have 1 problem to solve: min and MAX bitrate...
Cheers

You can change max bitrate if you set two pass - 2nd pass option. But when I try to preform two pass encodeing my VDMod crash at all time :(

EDIT: As for encoding speed. On short sample avi (without any filters)
ffvfw: 41 sek
TMPG: 64 sek

As a matter of fact, earlier this afternoon (about 4 hours ago), I E-Mailed the author asking just that :mrgreen:

Edit: This is an ffmpeg thread. Let's continue on a ffvfw thread, which I'm just creating right now :!:

-kwag

So we both bombarded him with emails ;) I've inform him that doc page is down, and asked hov to change matrixes. I hope that he will not get mad at us ;)

For test: Quality equal.

Ok guys lets get back to the topic, shall we?

Hya guys,
Thanks for the lesson Hydeus. I think I'll give it a shot one of these days.
Now to the bad news: I can't match the quality I experienced a few weeks ago with mencoder :cry:
Tried different builds, tried different command lines even including IBBP command that Tuco told us, but no go.
I wonder if you guys have quality issues with mencoder...
Don't get me wrong because it's quite watchable but looks like DivX and it didn't when I first tried it :?
I think it's time you guys post your command lines here.
BTW here it's mine:Note the crop/scale/expand parameters: My movie is 720x576 but clipping black bands it gets 720x552.
FitCD says it should be cropped as 712x576 so I did using crop.
Using a wonderful explanation by Inc. on how to set FitCD or Moviestacker for manually resizing, FitCD finally said that I should resize to 688x400. So I did with scale. Note that I used 1 block overscan!
Since I need to get to 704x576 with black band added I used expand.
This gives me (at least to my eyes) a perfect AR and should help compressability...
Now this is all ok using mencoder parameters but I really think there should be an option to use avisynth.
The Mplayer team doesn't pay attention to this because Mplayer is basically for Linux and they don't have avisynth. And I feel so sorry for them...
Hey Kwag are you running some tests also? What about quality? Does it level with the one you saw on ffmpeg?
Cheers guys

Note: sorry for the double post by i think this belongs here

Hi rds_correia,

Haven't tried it yet :!:
Where's that link to the windows binary of mencoder :?:

-kwag

@ rds,

Can you post on the first page, a:

* Latest working mencoder script, and a
* BETA-Testing (or work-in-progress) script that you all are currently
working on during these threads ??

Yes, I'd like to copy/paste the lastest working script to test out mencoder,
but I'm not sure which script to use 8O

Just so that it makes it easy to go to one place for a working script, instead
of having to read through all the threads :wink:

BTW, and FWIW.. I thought the Spiderman movie was great!! Even if the
young spidy guy really did suck in his acking, the movie was great!! And,
it's a movie I can actually watch over and over. Like I said, FWIW.. :mrgreen:

@ kvc,

Are you refering to ???
* ftp://ftp.mplayerhq.hu/MPlayer/releases/win32-beta/old/

Thanks for understanding (considering)
-vhelp

Sorry guys,

Now I need to ask a dumb question..

Is there an already made thread here, for DUMMIES sticky somewheres ??
my mencoder is giving me headackes :roll:

BTW.. pasting this script inside my DOS box didn't quite work either. Must
be a better way. BAT file ??

What's your method for windows.

Thanks,
-vhelp

My test with mencodet stops when I could not pass through his ".. didn't find ..." infos :( For now I'm starting tests with only notch matrixes and bitrate limiter. I think that using any of filters for now is only suitable when we could get good quality withouth them.

@vhelp
Using BAT files is always best way to test any CLI aplication. I always do this way, while I can change single function withouth repasting and editing whole instruction in CMD window.

EDIT: I've made small test, and I can say that there is need to set quantizers values. Picture is screaming witch large blocks (so I guest they values are freely arranged form 2 to 31; ~13 was max in BR viewer) and I think this must be set sharp, for instance: min3 and max4 or 5. Small problem -> I don't know how to do that :roll: I think that this option is common fr windows and linux version, so maybe some "mencoder linux master" will help us with that.

Hi Hydeus,

Thanks for your response.

Ok, so if I use a .BAT file, and paste the (what I'm assujing to be) the latest
script file (above) into a .BAT file, and then Start the .BAT file, all should
work ??

Yes.. I know.. I'll have to try this, but it would be nice to know what the
answers are prior too, just in case there are other things to be aware of.

Well, I'll try this .BAT thing out soon enough - pfew!

-vhelp

Yes, this is the way. Only make sure to have proper paths of mencoder and input/output files.

Hello everyone!

I just got my hands on this new toy! :D I still can't comment the quality of the encoder, but I can agree that's very fast! The latest windows compile can be downloaded here: ftp://ftp.mplayerhq.hu/MPlayer/relea...32-1.0pre3.zip
I tried to compile the latest cvs in cygwin, but couldn't make it...

I got the command-line rds_correia suggested, with some modifications:

mencoder -sws 9 -vf pp=hb/vb/dr,hqdn3d,noise=lumaa50,scale=704:-3,expand=0:480 -ofps 23.976 -of mpeg -nosound -ovc lavc -lavcopts
vcodec=mpeg2video:intra_matrix=8,9,12,22,26,27,29, 34,9,10,14,26,27,29,34,37,12,14,18,27,29,34,37,38, 22,26,27,31,36,37,38,40,26,27,29,36,39,
38,40,48,27,29,34,37,38,40,48,58,29,34,37,38,40,48 ,58,69,34,37,38,40,48,58,69,79:inter_matrix=16,18, 20,22,24,26,28,30,18,20,22,24,26,28,30,32,
20,22,24,26,28,30,32,34,22,24,26,30,32,32,34,36,24 ,26,28,32,34,34,36,38,26,28,30,32,34,36,38,40,28,3 0,32,34,36,38,42,42,30,32,34,36,38,40,42,4
4:mbd=1:vqscale=2:vrc_minrate=300:vrc_maxrate=5000 :vrc_buf_size=3660:keyint=18:vlelim=-4:vcelim=7:lumi_mask=0.05:
dark_mask=0.01:aspect=4/3:vmax_b_frames=1 "D:\Videos\my movie.avi" -o encoded.mpg

Not let me explain what this stuff means. You can find complete explanations in the manpage that comes with the binary, but I thought that explain the main stuff would be useful.
-sws controls the resizer type. In this case we're using lanczos, bilinear and lanczos are also available.
-vf controls the filters being applicated. I added a post-processing filter and a noise one, and changed to 480 the resizer (NTSC)
vrc_buf_size-This value controls the VBV buffer size. The value I used produces a VBV 224.
-keyint controls the GOP size, I changed to 18.

EDIT: Changed the command-line, with some new interesting stuff:
hqdn3d is a hi-quality version of denoise3d, but incredibly, works faster! Here in my atlhon 2000+ encoding goes from 28 fps to 34 in this exchange.
mbd=1 : manpage says this method yelds better quality.
vqscale=2 : This is the Mencoder constant quality mode. Very interesting. You can use values from 2 to 31 to vary the quality. Quality is much better than 1 pass VBR. Float point numbers can be used.
vmax_b_frames: enable B-Frames (default disabled). This causes tons of errors to me... :?

edit2: vmax_b_frames:1 doesn't give errors, 2 or 3 will give tons of "PTS earlier than SCR". Also using 2 or 3 makes the GOP setting don't work anymore

hi,

I'm still having trouble executing anything w/ mencoder :(

I copied your scrip above, and saved in a text file, and named my source
and directory locations accordingly (using a frameserved .vdr as my source)

But, all I get is a dos window w/ what looks like an infinate looping of the
same text screen (the mencoder.bat) and flickers.

Filename: mencoder.bat
File contents: (copied your above script and pasted into mencoder.bat)
Execute: doubl-clicked mencoder.bat to run

8O

-vhelp

@vhlep
None frameservering (VD, VFAFI, makeAVIS, ASV) work with mencoder :( You must use pure video files (AVI, VOB, MPG, ....)

I did some quality compare againts CCE. Without avisynth filters, this encoder definatelly loses by far, picture is very blocky when encoding DivX files. I tried to tweak the internal filter but couldn't make it better.
So I decided to create a huffyuuv lossless file from my avs script, and encode in CCE and Mencoder. This time mencoder did much better. Of course, I was just testing, transcoding to huffyuv then encode with mencoder is not practical to me because I don't have too much HD space, and it makes the process much slower.
I know most people uses DVD sources. But I like my DivX. :D
Also post-processing is very aggressive, takes away all the noise! :( And the noise filter seems to be doing nothing.
I'll keep testing the internal filters. :!:

Hya vmesquita,
Good to see some more Portuguese speaking guys here at mencoder discussian :D . Some considerations about your tests:
You, mean you're using Lanczos with a -sws 9 and that both Bicubic and Bilinear are also available using - sws 2 or -sws 0, right?
Great stuff, uh? I'll try it with your pp filters mantaining my 576 vertical res.
Watch out v: you're ending up with a ~448 vbv buff when using the value you have on the command line you posted. Here's the equation for DVD where vbv buffer needs to be 224. Eq: 224*1024*8/1000=vrc_buf_size
For VCD use 40*1024*8/1000 and for SVCD use 112*1024*8/1000.
Don't trust what you see in bitrate viewer!
Great news. I'll try hqdn3d right away to see the results :wink:
Strange...This is Macroblock decision algorithm (high quality mode) and my manpage says that when using a value of 2 "MB mode which has the best rate distortion"...
About vqscale. Agree that it is a constant quality command ranging from 2 to 31 to vary quality, but I tried floating point numbers and it seams the encoder rounds the figure...
I haven't had any issues with it using vmax_b_frames=2. Can you explain why you say that using 2 or 3 resets the GOP?
Cheers

same here! :D :D :D
Yes.
Later I realized that those pp filters are very aggressive. Maybe they need to be tweaked, because they were taking away all the noise from my movie. But you can also use the decode codec post-processing filters by addind -pp (not -vf pp). Those looked much less agressive.
Thanks! I was trusting bitrate viewer. :)

Yes, but MB 1 has (=vhq)... I am not sure what is better.
I haven't checked that.
When I use 2 or 3, I get tons of
Error in stream: PTS earlier than SCR!
Error in stream: PTS to SCR delay 0 is too little!
when I use 3, GOP starts with 19 and subsequent GOPs are 16 (and I am using a GOP of 18)

Also: are you able to use huffyuv? I wanted to test the filters output by using huffyuv, so I could understand them better. Unfortunatelly, huffyuv and lossless jpeg don't work for me. I am using MJPEG with quantisizer 1 for this tests, but I don't think it's completelly lossless.