Anyone encoded series from DVD like buffy, xena...?

DvdKhl · 30th August 2003, 15:33

I'm encoding Buffy but there is a wired thing with interlaceing

If I just load the vob file.
It looks interlaced
but when I put in it
--------------
assumefieldbased()
separatefields()
weave()
--------------
the movie is progressive
and then I noticed that the 1&2, 3&4 ... horizontal lines are the same

so I changed the script like this
--------------
assumefieldbased()
separatefields().selecteven()
--------------

can somebody tell me why they do this (doubling the height)?
is there a better way to make the movie progressive?

cweb · 31st August 2003, 10:50

I check what the type is, by using DVD2AVI.

Edit: Sometimes Field operation has to be set to "forced film"(NTSC only), and in other cases you need to deinterlace with decomb.

There are many guides on this on this site etc.

Try also searching the forums.

Si · 31st August 2003, 20:46

I don' t thinkAssumeFieldBased doesn't do what you might think it does

(At least I think it doesn't)

AFAIK it tells Avisynth internally that the clip is a series of fields with 1 field in each frame (a bottom field then a top field then a bottom field then a top field etc)

It doesn't tell Avisynth that the source is interlaced or contains fields

It makes Avisynth think that a 2 field/frame video has had a SeparateFields() done to it.

So if you do SeparateFields() again Avisynth might/does get confused.

regards
Simon
All the above based on theory/memory so a possible a large fall from a great height is waiting to happen

DvdKhl · 31st August 2003, 21:52

I don' t thinkAssumeFieldBased doesn't do what you might think it does (At least I think it doesn't)

he he no I don't know what it does but it solves my problem

here some pictures
------------------------------------------------------------
MPEG2Source("C:\Dokumente und Einstellungen\DvdKhl\Eigene Dateien\Encoding Stuff\1.d2v",cpu2="oooooo",idct=6,iPP=false)

-------------------------------------------------------------
MPEG2Source("C:\Dokumente und Einstellungen\DvdKhl\Eigene Dateien\Encoding Stuff\1.d2v",cpu2="oooooo",idct=6,iPP=false)
assumefieldbased()
separatefields()
weave()

-------------------------------------------------------------
MPEG2Source("C:\Dokumente und Einstellungen\DvdKhl\Eigene Dateien\Encoding Stuff\1.d2v",cpu2="oooooo",idct=6,iPP=false)
separatefields()
weave()

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

without assumefieldbased() the commands separatefields(),weave()
in the script do nothing but with it like in the 2. example it makes the frames it progressive. But if you zoom the (progressive) picture so you can see the pixels you can notice that the 1&2, 3&4 ... horizontal lines are the same

I just wanted to upload a little vob file but my webspace doesn't let me upload file bigger than 500kb :/.

-----------------------------------------------
There are many guides on this on this site etc.
Try also searching the forums.
-----------------------------------------------
I already did. But they always say to deinterlace, but I think I get the best result with my script.
I just want to know if there's a better method, and I want to know the reason why they "interlace" that way and doubling the height.

Si · 31st August 2003, 22:09

Quote:

he he no I don't know what it does but it solves my problem

is not equal to

Quote:

But if you zoom the (progressive) picture so you can see the pixels you can notice that the 1&2, 3&4 ... horizontal lines are the same

IT DOES NOT SOLVE YOUR PROBLEM

Use KernelDeInt() - http://neuron2.net/kerneldeint/kerneldeint.html - it will

Video isn't simple and is difficult to understand for beginners

regards
Simon

DvdKhl · 31st August 2003, 22:33

ok I'm not the pro in video enc especially not in deinterlacing
but I can't say I'm a beginner... ;D

ok I tried what you suggested here are the results
(and yes I choose the correct field order)
-------------------
(resized them by 2 to see it better)
KernelDeIn

-------------------
assumefieldbased()
separatefields()
weave()

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

DvdKhl · 31st August 2003, 22:43

another question
why does
assumefieldbased()
separatefields()
weave()

result a progressive picture?
or does it anyhow copy the lines?

Guest · 1st September 2003, 06:14

There is a bug in SeparateFields. The way it is supposed to work according to the documentation is:

"This filter has no effect if the clip is already field-based."

But by experiment you can see that after AssumeFieldBased(), it does a separate and returns undefined fields. I could not find a predictable pattern for what it returns. Some clips behaved like yours, some had changing dominance as the clip progressed, etc.

I've duplicated your observation, so you are not imagining things.

Bottom line: doing SeparateFields() on a field-based clip is not a valid operation.

Anyway, there are better (legal) ways to achieve field discarding/duplication. But field discarding is not the best way to deinterlace. KernelDeint() and others are much better.

DvdKhl · 1st September 2003, 11:03

But field discarding is not the best way to deinterlace....

thx thats what I just want to know, that if I use my script that I would lose a field.

I'm going to try some deinterlacers and hope I'll find some good parameters for it

30th August 2003, 15:33	#1 \| Link
DvdKhl Registered User Join Date: Dec 2002 Location: Germany Posts: 82	Anyone encoded series from DVD like buffy, xena...? I'm encoding Buffy but there is a wired thing with interlaceing If I just load the vob file. It looks interlaced but when I put in it -------------- assumefieldbased() separatefields() weave() -------------- the movie is progressive and then I noticed that the 1&2, 3&4 ... horizontal lines are the same so I changed the script like this -------------- assumefieldbased() separatefields().selecteven() -------------- can somebody tell me why they do this (doubling the height)? is there a better way to make the movie progressive? Last edited by DvdKhl; 30th August 2003 at 19:53.

31st August 2003, 10:50	#2 \| Link
cweb Registered User Join Date: Oct 2002 Location: The Pandorica Posts: 528	I check what the type is, by using DVD2AVI. Edit: Sometimes Field operation has to be set to "forced film"(NTSC only), and in other cases you need to deinterlace with decomb. There are many guides on this on this site etc. Try also searching the forums. Last edited by cweb; 31st August 2003 at 10:52.

31st August 2003, 20:46	#3 \| Link
Si Simply me Join Date: Aug 2002 Location: Lancashire, England Posts: 610	I don' t thinkAssumeFieldBased doesn't do what you might think it does (At least I think it doesn't) AFAIK it tells Avisynth internally that the clip is a series of fields with 1 field in each frame (a bottom field then a top field then a bottom field then a top field etc) It doesn't tell Avisynth that the source is interlaced or contains fields It makes Avisynth think that a 2 field/frame video has had a SeparateFields() done to it. So if you do SeparateFields() again Avisynth might/does get confused. regards Simon All the above based on theory/memory so a possible a large fall from a great height is waiting to happen __________________ http://www.geocities.com/siwalters_uk/fnews.html

31st August 2003, 21:52	#4 \| Link
DvdKhl Registered User Join Date: Dec 2002 Location: Germany Posts: 82	I don' t thinkAssumeFieldBased doesn't do what you might think it does (At least I think it doesn't) he he no I don't know what it does but it solves my problem here some pictures ------------------------------------------------------------ MPEG2Source("C:\Dokumente und Einstellungen\DvdKhl\Eigene Dateien\Encoding Stuff\1.d2v",cpu2="oooooo",idct=6,iPP=false) ------------------------------------------------------------- MPEG2Source("C:\Dokumente und Einstellungen\DvdKhl\Eigene Dateien\Encoding Stuff\1.d2v",cpu2="oooooo",idct=6,iPP=false) assumefieldbased() separatefields() weave() ------------------------------------------------------------- MPEG2Source("C:\Dokumente und Einstellungen\DvdKhl\Eigene Dateien\Encoding Stuff\1.d2v",cpu2="oooooo",idct=6,iPP=false) separatefields() weave() ------------------------------------------------------------- without assumefieldbased() the commands separatefields(),weave() in the script do nothing but with it like in the 2. example it makes the frames it progressive. But if you zoom the (progressive) picture so you can see the pixels you can notice that the 1&2, 3&4 ... horizontal lines are the same I just wanted to upload a little vob file but my webspace doesn't let me upload file bigger than 500kb :/. ----------------------------------------------- There are many guides on this on this site etc. Try also searching the forums. ----------------------------------------------- I already did. But they always say to deinterlace, but I think I get the best result with my script. I just want to know if there's a better method, and I want to know the reason why they "interlace" that way and doubling the height. Last edited by DvdKhl; 31st August 2003 at 22:04.

1st September 2003, 06:14	#8 \| Link
Guest Guest Join Date: Jan 2002 Posts: 21,901	There is a bug in SeparateFields. The way it is supposed to work according to the documentation is: "This filter has no effect if the clip is already field-based." But by experiment you can see that after AssumeFieldBased(), it does a separate and returns undefined fields. I could not find a predictable pattern for what it returns. Some clips behaved like yours, some had changing dominance as the clip progressed, etc. I've duplicated your observation, so you are not imagining things. Bottom line: doing SeparateFields() on a field-based clip is not a valid operation. Anyway, there are better (legal) ways to achieve field discarding/duplication. But field discarding is not the best way to deinterlace. KernelDeint() and others are much better. Last edited by Guest; 1st September 2003 at 06:20.

31st August 2003, 22:33	#6 \| Link
DvdKhl Registered User Join Date: Dec 2002 Location: Germany Posts: 82	ok I'm not the pro in video enc especially not in deinterlacing but I can't say I'm a beginner... ;D ok I tried what you suggested here are the results (and yes I choose the correct field order) ------------------- (resized them by 2 to see it better) KernelDeIn ------------------- assumefieldbased() separatefields() weave() -------------------

31st August 2003, 22:43	#7 \| Link
DvdKhl Registered User Join Date: Dec 2002 Location: Germany Posts: 82	another question why does assumefieldbased() separatefields() weave() result a progressive picture? or does it anyhow copy the lines?

1st September 2003, 11:03	#9 \| Link
DvdKhl Registered User Join Date: Dec 2002 Location: Germany Posts: 82	But field discarding is not the best way to deinterlace.... thx thats what I just want to know, that if I use my script that I would lose a field. I'm going to try some deinterlacers and hope I'll find some good parameters for it