@Experts, Your comments are needed!

[birdy]

Ok guys, I have been reading and searching this forum and many guides and spent
time and did my homeworks :)

So I guess that now I know a little the basics! But really need your help and
comments, just to make sure that I'm on the right track and ofcours I guess I
still need to learn much much more.

That's why I decided to post a detailed doc on what I did, hoping that you guys
would help me out and tell me where I did wrong and what I could do for getting better result!

It could also be interesting, as there is a slight difference in what I do!
I did not rip any DVD but in this example I captured a movie from satellite tv
directly to Xvid! and the result is simply amazing!

So here is what I did:

My AVI source file was (Back to the future3) Captured @768X576 compressed directly to XVID with 1-pass CBR, 1780KBPS, H.263, Motion search5 and Quantizers
were: MinI2, MaxI4, MinP2, MaxP4. (all the other settings were codec defaults)

This gave me a file about 1.4 GIG (with mp3 audio 128kbps) That I frameserved with AVISYNTH to Vdubmod.

Here is the details of avs file <--- here is were I need advice (what do you think about the resize? filters? etc)
----------------------
#LOADED AND PLUGINS:
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\MPEG2Dec3.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\UnDot.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\UnFilter.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\Convolution3DYV12.dll")
AVISource("F:\Captures\Movies\Back To The Future 3.avi")
Crop(4,84,-0,-84)
UnDot()
Lumafilter(-2)
Convolution3d(preset="movieHQ")
UnFilter(-5,-5)

BilinearResize(576,288)

--------------------------


As the audio was 100MB MP3 I needed the movie to become 600MB
so I did a 2 pass xvid and choose 612352 for output size.

And here is the settings used in XVid for encoding<--- (again help and comments needed)

2pass first pass:
Motions search 6
H.263
Max I frame 250
Chroma motion enabled
bframes: 3, 100,200
No lumi masking, no Qpel, No GMC
And the rest was the same as iago Guide.


After I did the first pass I opend the stat file with Xvid Statreader
and went for the second pass.
In the second pass I just set i and p frame quantizers to:
2,6,2,12
the rest of settings were just the same as first pass.

I have attached a picture called Source and Encoded.
As there name say source is from my source avi and encoded is a snap shot from
my encoded file done at real (original size) done with bsplayer.This way you can
get an idea if the quality I got!

The movie really looks nice, but please tell me if I am on right track or not!

[birdy]

Encoded sample:

[birdy]

Sample 2:

[DJ Bobo]

1) Why not capture in a less lossy format? like MJPEG for example? for better quality!

2) Sound should be captured in WAV, then afterwards encoded in mp3 (but this is not rea~lly an issue if you're intending to keep it 1:1, I'm just being picky here :D)

2) You're using way too many filters :D
TemporalSmoother is completely sufficient and it helps keeping as much details as possible without sacrifying compressibility.

[NuclearFusi0n]

"In the second pass I just set i and p frame quantizers to:
2,6,2,12"

dont do that
leave it at the default 2 - 31 for both

[iago]

leave it (quantizers) at the default 2 - 31 for bothMay I learn why?

[birdy]

@ DJ Bobo ,
You are deffinitivly right. MJPEG would give me damn better result I guess. But the problem is that I am subscribed to Movie channel on satelliet and they don't wait for me :) 2 captures and with the time it takes me to encode My HDD is already out of space :( As I capture at least 3, 4 movies a day!

But for the sound as you said, I capture to wav and then encode to mp3.
For your filter tip, I am gonna do right now another encode using only TemporalSmoother and see what happens.
Btw: no undot needed?

@NuclearFusi0n,
Right on man, 2-31 you say, 2-31 it will be. gonna do that right away.
You know, I am not experienced with encoding but almost experienced with capturing and lower quntizers would give me better results when capturing. Thats why I used lower values for my encoding too!

[iago]

There's nothing wrong with capping quantizers. Kinda safety belt. I usually cap them 2-4/2-8.

Temporal Smoother has rarely been sufficient/pleasing for me so far. UnDot, UnFilter, and Convolution3D are all great filters when used properly and as to match the requirements of the source and the encode.

Btw, afaik all smoothers/denoisers (temporal and/or spatial) "help" compressibility and not "sacrifice" it anyway.

[birdy]

@ iago,
All my respects cheef. Your reply to my posting gives me the occasion to thank you, as I almost learned most of what I know from you. For encoding from your guid and much from your replies to others.

For the quantizer setting I was hoping that a more agresive setting would give nice result and wanted even to give it another test with higher quant tonight for experince! (something like 2-4, 2-(8) or (6)

But it seems that there are divided points of view about this matter.

[birdy]

There was something wierd i found out when using TemporalSmoother.

If I use this filter AVI synth gives me an error telling me that it works only with YUY2 Or RGB32!

My source avi file is captured with YUY2 but apparently its no longer YUY2!! (is this normal?)
So I added this line in my avs file just after avi source:
ConvertToYUY2() (if conveting is needed wouldn't it be better to convert to RGB32?) or YUY2 is ok?

Also when I get file properties in vdub is says that decompressor for me avi is DIVX 5. My avi is XVID compressed with fourCC set to Xvid and when played in bsplayer its decoded with nics decoder! But when frameserved vdub think its divx!!!

[Alxemi]

I´m not an expert, but i can make you a suggestion :)
What about using Ogg-Vorbis for audio streams instead MP3?
You could obtain better sound quality with lower bitrates, and free space for the video. :D

[birdy]

Thanx to all of you for you help and valuable advices.
Infact by lowering my filters I achived more sharper picture.

I have another question about encoding:

As mentioned after I do my first pass encoding, I open the file with
"xvid statsreader" and I do the second pass as external with the new
saved stat file done by statreader.

I saw another program called "Xvid Stats viewer" by Nitrogen.
here is my question:

What I understood from this app is that it gives you values to set for
the high and low distence from average.

So I used the Stat file I made with "Xvid statreader" and used the sugested
values from "Stat Viewer" !!!

Is this the right thing to do or should I use the original 1 pass stat file?

What do you think about using Stat viewer in all? Could you please sugest me
how to use it best?