Avisynth: Codec (YUV12 or YV12), Which is Which ?

[vhelp]

@ all, or Kwag..

Is it me, or is the latest AVIsynth v2.52 color washed ??

I'm talking for DVD rips.

As far as what's in my script, there are NO color conversions

@ Kwag.. what YV12 CODEC are you using for your clips to
encode when your source is DVD rips ??

When I open any "DVD rip" related scripts inside vdub, it reports that
I'm using: ATI YUV12 (thought its not saying YV12)
which is a bit confusing to me. Am I using a YV12 or a supper-set of..

Please help/advise, and let me know what CODEC you are using for
your DVD rips, and I'll D/L it and give it a shot, K pal ??

Thanks in advance
-vhelp

[audi2honda]

what do you mean codec? You don't need codecs to rip or am i missing something here.

[vhelp]

@ audi2honda..

Well, lets say:

Analigy...
* that you have an .AVS script, and
* you frameserve it into vdub, and
* you verify the codec, by checking the "information", and
* it tells you it's ATI-YUV12.

Here is what I'm seeing thus far..
* color washout.

The color washout is probably due to the CODEC above.. ie, it's limitation
or something else effecting it. Though, I've never experienced this w/
AVIsynth 1.0 BETA, USING THE SAME METHOD ABOVE. For the first time,
I am experiencing color washout for DVD rips, when I have the script
and feed it into vdub. Mind you, I've done this for a very long time, and
w/out one hint of issue.. till today.. after I installed AVIsynth v2.52

So, I'm wondering if it's because of this latest AVIsynth version, or is it
something else that got messed up (which I doubt)

Lets continue w/ the Analigy...
Ok, after having frameserved it into vdub (to verify it's contents etc)
I then proceed to frameserve it yet again, into TMPG.. adding even more
loss to encoding. All these conversions. Actually, this is the route that I
have ben using for ever, because I'm am one of those who suffer the
color bug, when you feed an .AVS script directly into TMPG. For some,
there are no soluitons to this. I've share the tip, that if you frameserve
the .AVS script into vdub first, THEN into TMPG, the color bug dissopears.
Today, I decided to try the latest AVIsynth version to frameserve into
TMPG, to see if I STILL suffered from this unknown color bug. Well, to my
surprise, I STILL DO. So, my only resort, is to frameserve my .AVS scripts
into vdub, and then frameserve into TMPG for a final encode.

.AVS --> VDUB --> TMPG

The above is no problem to me. I've done this for 2 years now, and w/out
any quality loss and w/out any slow-ups either.

But, getting back to the Analigy...
I was asking about the YV12 CODEC, because I thought that if I had a
different YV12 codec to ues, the color washout would be eliminated

Now, if I take my psuedo .AVI file and open in vdub, the is no color washout
and quality is excellent.

I'm sure I will figure/sort it all out, but I was hoping for some team
support here

Thanks.
-vhelp

[kwag]

I think I know what you mean vhelp
Don't laugh, but any .avs I open with Vdub, the file information is: Decompressor: XviD MPEG-4 Codec
Is XviD CODEC designed to handle MPEG-2 from VOBs too
I guess so, and that's how I've been processing all my movies

-kwag

[jorel]

Kwag,i see the ffvfw codec

vhelp,
did you install avisynth 252 with ffvfw(mpeg-4 codec)?

after install this option, i uninstall the ati and everything works fine

[vhelp]

@ Kwag..
Yes, I laughed haha.. I'm not alone after all, in the weird
of thing hehe.. I'm gladd I'm not the only "crazy" around here.

@ pal Jorel..
what-do-ya know.. I selected "ffvfw" during the v2.52 install
But, I'll leave my ATI were it is, cause believe it or not, I STILL use it
hehe.. in my ATI captures, cause they come out great when I select the
YV12 during capturing. What can I say.. I'm VERY expandable. I always
try everything.

Anyways.. I ended up adding a ConvertToYV12 and it seemed to
fix everything (when looking at it in vdub) I'll verifty the final quality
after a doe a few test encodes. Also, you might want to add it at the END
of your script, or if you are using the MD script, I put it just before the
function, like this

Quote:

ConvertToYV12
## Functions ###

The above is for DVD rip sources. If there is another way (w/out having
to resort to yet more conversions) I'm all for it. But, the above seems to
be ok for now. I'll get back to ya shortly on my semi-final analisys.

Now that I have all my VFAPI and what-nots resolved, I couldn't be all the
more happier. I'm soo happy hehe..

@ Kwag..
more sample clps for motivation

-vhelp

[kwag]

Quote:

Originally Posted by vhelp

@ Kwag..
more sample clps for motivation

Sure
Let me see what I can rip

-kwag

[vhelp]

.
.
I lost my Contact DVD. All I have on this now, is the WideScreen
version on VHS. I must find it. Just don't use A Perfect Storm, as
that must have ben murder on the final encode, and I have no plans on
backing that one up. I do perfer WideScreen movies, case I like to take
advantage of the AR and things.

some I have are.. I've got like, two whole racks full of DVD's hehe..
I love running encoding tests on them.. and the new MD script is deffinately
a challenge for me.

@ Kwag..
I've ben playing around w/ the code snip below to see if I can carry out
two more more functions in one line. It looks like it works, but I'm not sure.
(at the moment) I'm familiar w/ AVIsynth scripting/programming syntax,
but I'm learning fast. Mind you, I was just toying around w/ different
functoins (below ie, undot() )
Can you tell AVIsynth allows more functions in one ScriptClip()
function ??

  ScriptClip("nf = YDifferenceToNext()"+chr(13)+ "nf > 2.5 ? asharp( -(fmin((nf/30), 1)), 0 ) : \ 
  TemporalSoften(2,7,7,3,2) undot() ")

I'll try and throw some up myself, if I can - - if ya want hehe..

thanks.
-vhelp

[kwag]

You can just concatenate more functions with a "." dot. But the "undot()" filter you have there is not needed, because it's already applied at the top of the script as a static filter, which is applied to every frame

-kwag

[kwag]

Sample ready
Some of you guys remember the K19 samples back in December
Well, look again with the adaptive filter. K19 is 138 minutes long (2 hours, 18 minutes) movie, and this is how it would look on one CD-R at 528x480 with a CQ of 58.112
http://www.kvcd.net/k19-adaptive-sample.mpg

Next sample is K-Pax

-kwag

[kwag]

Here we go:
www.kvcd.net/kpax-adaptive-sample.mpg
And this one ROCKS
K-Pax is 2 hours long. Here it the sampler at CQ=62.618, 704x480 to fit on one CD-R

EDIT: Correction, the CQ on this sample was 57.13

-kwag

[ovg64]

Quote:

Originally Posted by kwag

Sample ready
Some of you guys remember the K19 samples back in December
Well, look again with the adaptive filter. K19 is 138 minutes long (2 hours, 18 minutes) movie, and this is how it would look on one CD-R at 528x480 with a CQ of 58.112
http://www.kvcd.net/k19-adaptive-sample.mpg

Next sample is K-Pax

-kwag

I sure like to know what you secret is for that much compression
Im having enough trouble getting over 60% CQ on a 1:40min. movie
maybe i should just try movies with less action.

[kwag]

K-Pax is a very low action movie. With the current adaptive script, the thing looks like the original DVD at 704x480
I'm encoding the complete movie right now.

-kwag

[audi2honda]

kwag I thought the current adaptive script did better on high action movies, because the motion blur in high action scenes left more bits for low action flat scenes?

[kwag]

Quote:

Originally Posted by audi2honda

kwag I thought the current adaptive script did better on high action movies, because the motion blur in high action scenes left more bits for low action flat scenes?

Yes, but it doesn't mean there's no blur on low action movies
It just means that the more action, the more blur. Even on slow camera pans, the blur treshold will be met, and blur will be applied.

-kwag

[audi2honda]

ha I didn't even notice it on slow camera pans. Guess that's why your script is so good. One last question.

Does your blur routine blur selective parts of a frame or the whole frame when it decides blur is needed.

For example a person on a bike going down the road. They are stationary on the camera, but the background is wizing by. Does your script blur everything but the person on the bike who isn't moving? or does the entire frame get the effect?

[kwag]

Quote:

Originally Posted by audi2honda

Does your blur routine blur selective parts of a frame or the whole frame when it decides blur is needed.

The blur is done on the complete frame. We already tried selective filtering ( thread here: http://www.kvcd.net/forum/viewtopic....er=asc&start=0 ) and it requires too much CPU power to accomplish. The screen has to be divided into quadrants, detect movement relative to next frame on each quadrant, and then stacked vertically/horizontally to reconstruct the new frame. We did it, and the result takes FOREVER . Check the thread.

-kwag

[vhelp]

@ Kwag..

I thought I had my DVD of k-pax.. ben searching all evening for this Movie.

Looks like eather I don't have it, or I lost it or something. Was hoping to
do a sample like yours. Oh well.

I'm digging around my collection now, and since I'm in the mood, I'm gonna
do a favoriate of mine, opening scene of Blue Streak. ever since I
saw a SVCD (PAL version) sample of this 2 years ago, I've had an inkling
to do it, but w/ all the techniques I learn over at vcdhelp, and here.

I'm also gonna take a crack at the opening scene of Deep Blue Sea
And, also Miss CONGENIALITY cause I love Sandra in that Movie.
wow.. NOW I know where I remember where I saw the guy in Red Planet's
from.. Agents Grace Hart's partner hehe.. Anyways..

By the way.. has ayone encoded the title above: Miss CONGENIALITY
and how, AR etc. ??
I have an Idea about the AR on this. It's in FullScreen, Like Batman, even
though the Movie says it's Widescreen I'll come back w/ a sample to
show you all how I think it SHOULD be

My new format will be to specify Chapters, so that others can follow or try out
THEIR hand at encoding such Movies.

Thanks Kwag, for the motivate..
-vhelp

[kwag]

Hey great vhelp
I think I have that DVD somewhere, "Miss Congeniality".
I'll see if I can find it

-kwag

[vhelp]

@ Kwag..

Cool

Since I'm NOW doing chapters (for easy following) I remember a few
that I'd like to re-do out of curiosity.

Note, when I do Chapters for VOBs, I make the following settings:
Using Smartripper v2.27
* Settings/FileSplitting and select (o) Every Chapter
* press [none] and just select my [x] Chapters that I want to do test
...encodes with.

EDIT:
Notes regarding above...
This will give me a VOB for every chapter I do, so that I can do my
individual encodings with.
Also, Sometimes I will just run all the VOBs through dvd2avi v1.76 when
I need to do all (if I've done those chapters that are w/in sequencial order)
otherwise, I just dvd2avi to each VOB.
Then, I note the Source Range (TMPG) and record the numbers inside the
final MPEG, so that later on, when I improve my encoding skills or want to
try out a new method/filter or what-have you, I can quickly do them again
w/out issues.

Below, are some chapters that I'm using, case anyone wants to follow or
give a try using their own methods.

Blue Streak - Chapters:
* 1,2,18,20,21

Miss Congeniality - Chapters:
* 1,6,8,10,11,20,21

Deep Blue Sea - Chapters:
* 1,2

-vhelp