avs script for NTSC Interlaced DV

[dreadycarpenter]

So I have this intervaced NTSC 720x480 dv footage, I'm trying to write an avisynth script to deinterlace and retain the 29.97fps and resolution,
so I can reencode with cce to be dvd-video compliant
I'm a bit new at writing scripts for avisynth, have tried the code posted on the dv to dvd guide

as:
DirectShowSource("D:\Videos\Projects\bday_vid\007.avi")
ReInterpolate411()
SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
Weave()
DoubleWeave.SelectOdd()

ERROR: script error, there is no funcion named "convolution3d"

so i load convolution3d
as:
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\Convolution3D.dll
DirectShowSource("D:\Videos\Projects\bday_vid\007.avi")
ReInterpolate411()
SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
Weave()
DoubleWeave.SelectOdd()

ERROR: script error. expected a,or)

here is a clip if anyone would like to help>
DV clip (http://personal.180com.net/dreadycarpenter/clip/dv_ntsc_interlaced.avi)
thanks

[Kika]

LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\Convolution3D.dll

Try this:

LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\Convolution3D.dll")

[dreadycarpenter]

so now :
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\Convolution3D.dll")
DirectShowSource("D:\Videos\Projects\bday_vid\007.avi")
ReInterpolate411()
SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
Weave()
DoubleWeave.SelectOdd()

and I get error:
plugin C:\Program Files\AviSynth 2.5\plugins\Convolution3D.dll is not an Avisynth 2.5 plugin.

im using Avisynth 2.55, build:sept 1 2004, the Convolution3D.dll is located in that dir..... :confused:

[henryho_hk]

1) Your script does not deinterlace the clip. Actually, you don't need to do it if you want to encode it to DVD.

2) Instead, you may want to pay attention to the field ordering: DV is bottom-field-first while DVD should be encoded top-field-first for best compatibility. (Or would CCE perform the conversion with a click of checkbox?)

3) Your Convolution3D.dll is probably an old version which is for Avisynth 2.0x. Take a look at the FAQ and you can find the right version.

4) Actually, I am always not sure if separatefields+filter+combine are performing correctly. The filter() thinks that the horizontal line are packed while they are actually one line apart in the real pictures.

5) Beware that your current DirectShow DV decoder may not be delivering the quality you desire.

[dreadycarpenter]

"So I have this intervaced NTSC 720x480 dv footage, I'm trying to write an avisynth script to deinterlace and retain the 29.97fps and resolution,
so I can reencode with cce to be dvd-video compliant
I'm a bit new at writing scripts for avisynth.

DV Clip (http://personal.180com.net/dreadycarpenter/clip/dv_ntsc_interlaced.avi)

any suggestions or sample scripts would be appreciated...
thanks.

[Peter1234]

I do not use cce so I am confused by what you want to do. DVD is interlaced. There is no need to de-interlace DV video to encode it to a DVD since it will just be re-interlaced on the DVD. lf you have a DV codec installed, all you need for a script in Avisynth is:

AVISource("source file address")
ConvertToYV12()

for example:

AVISource("C:\Documents and Settings\user\Desktop\DVtype2.avi")
ConvertToYV12()

For some types of filtering you may want to separate the fields and process the even and odd fields separately, but since you are new to this you should probably keep it simple to start with. Most DV codecs only work with type2 DV files so make sure you capture that way.

[bb]

I do not use cce so I am confused by what you want to do. DVD is interlaced. There is no need to de-interlace DV video to encode it to a DVD since it will just be re-interlaced on the DVD.[...]
I'm sorry, but this is plain wrong. Of course you can encode progressive DVD. Most commercial DVDs are progressive.

Another question is whether it makes sense to deinterlace interlaced DV in order to put it on a DVD. In general I wouldn't, because you can encode interlaced MPEG-2 and use that to author the DVD (thus creating an interlaced DVD), and you keep the motion fluidity and avoid deinterlacing artefacts. On a DVD you usually have enough bitrate available to encode interlaced video (for SVCDs you may argue, though).

In the rare case that anybody records progressive DV (as I do), there's of course no need to deinterlace, and you'd encode in progressive mode.

bb

[dreadycarpenter]

everytime I have encoded these DV clips to mpeg-2(using any encoder), they look like this:
http://personal.180com.net/dreadycarpenter/clip/dv1.jpg
if I use TmpgEnc3.0 to encode to mpeg-2 and deinterlace it looks good, I am very happy with the method of deinterlacing that is used with TmpgEnc, but I've experienced better results with CCE than TmpgEnc on other encodes(not dv), so I expect that I could get better quality with CCE if properly deinterlaced

I have used this script with CCE and works well, but there has got to be another script to do the same thing with better quality.


DirectShowSource("D:\projects\mak5thbday\mak 5th bday clip dv.avi")
ReInterpolate411()
bob
FDecimate(29.97)

thanks

[Kika]

everytime I have encoded these DV clips to mpeg-2(using any encoder), they look like this:

Looks good. It's interlaced video, therefore it HAS to look like this... ;)

[dreadycarpenter]

Looks good. It's interlaced video, therefore it HAS to look like this... ;)

are you kidding it looks horrible, none of my previous mpeg2 encodes have ever looked like this, except this particular batch of DV clips

this is what it looks like interlaced & deinterlaced (using TmpgEnc)

http://personal.180com.net/dreadycarpenter/clip/dv_2.jpg

a whole world of difference, It is my belief that this footage was captured or Interlaced different than most DV

thanks

[Kika]

As i said, the interlaced Version looks fine. :D
That's a normal effect on PC-Monitors. On a tv, you can't see these lines.
The deinterlaced version is horrible.
You have to read some guides about interlacing, interlacing is not a bug, it's a feature.

[dreadycarpenter]

so I got the script from the dv to dvd guide:
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\Convolution3D.dll")
DirectShowSource("D:\projects\mak5thbday\mak 5th bday clip dv.avi")
ReInterpolate411()
SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
Weave()
DoubleWeave.SelectOdd()
it's working (now that I got the right Convolution3d.dll) and it still looks the same

I really feel that it is Interlaced different than most dv

here is what I have found will deinterlaces properly:
bob(AviSynth filter)
or, using TmpgEnc3.0Xpress=Deinterlace,Deinterlace by bottom field, Interpolation adjusted

I'm trying to write an avisynth script to deinterlace and retain the 29.97fps and resolution,
so I can reencode with cce to be dvd-video compliant
I'm a bit new at writing scripts for avisynth.

I have used this script with CCE and works well,
DirectShowSource("D:\projects\mak5thbday\mak 5th bday clip dv.avi")
ReInterpolate411()
bob
FDecimate(29.97)

but there has got to be another script to do the same thing with better quality.

if anyone would like to check out my clip maybe it will help you better understand my problem
dv clip interlaced (http://personal.180com.net/dreadycarpenter/clip/dv_ntsc_interlaced.avi)

[dreadycarpenter]

As i said, the interlaced Version looks fine. :D
That's a normal effect on PC-Monitors. On a tv, you can't see these lines.
The deinterlaced version is horrible.
You have to read some guides about interlacing, interlacing is not a bug, it's a feature.

hmmm... well I encoded a clip without any deinterlacing, authored a burned a dvd, played it on my standalone dvp-642 and my neighbor's sony standalone dvd player, STILL the same thing, I deinterlaced,encoded, authored and burned, looks great on all!!

once again this is what im asking for
I'm trying to write an avisynth script to deinterlace and retain the 29.97fps
and resolution, so I can reencode with cce to be dvd-video compliant
thanks for all the posts :D

[Guest]

hmmm... well I encoded a clip without any deinterlacing, authored a burned a dvd, played it on my standalone dvp-642 and my neighbor's sony standalone dvd player, STILL the same thing Probably you resized it or the encoder resized it for you to be DVD compatible. Resizing interlaced content requires special handling.

[bb]

hmmm... well I encoded a clip without any deinterlacing, authored a burned a dvd, played it on my standalone dvp-642 and my neighbor's sony standalone dvd player, STILL the same thing, I deinterlaced,encoded, authored and burned, looks great on all!!
We have real video experts here, so believe me - you can trust us. Encode interlaced, and you'll get the best result. But you have to do it right!

Please post what exactly you did (AviSynth script, encoder settings, authoring, etc.), else I'm afraid we can't help you.

bb

[dreadycarpenter]

thanks to bb and neuron2 for the posts :D

Probably you resized it or the encoder resized it for you to be DVD compatible. Resizing interlaced content requires special handling.

first of all CCE "does not change the resolution", if the resolution is not dvd compliant then I cant author it....
second there is no setting within CCE to choose interlaced or progressive, I believe that it encodes interlaced/progressive depending on the source/script.
third, it is my belief that this is NOT INTERLACED AS MOST DV IS.

I've encoded using these different methods:

using TmpgEnc3.0Xpress-importing DV clip directly
all default settings
note it recognizes source as:
http://personal.180com.net/dreadycarpenter/clip/tmpeg1.jpg


and defualt deinterlacing method:
http://personal.180com.net/dreadycarpenter/clip/tmpeg2.jpg
then encode to mpeg-2vbr 2pass, 720x480 29.97fps,4:3 aspect ratio
then author burn and play on standalone dvd player
result:looks bad same as when I watch on my computer.


now using TmpgEnc3.0 with changed deinterlacing method:
http://personal.180com.net/dreadycarpenter/clip/tmpeg3.jpg
note have tried "deinterlace by bottom field" and no significant difference.
result: this is the best quality I have been able to achieve, looks great on dvd player and computer.

using CCEncoder 2.66-importing DV clip directly
all setting as on the doom9 dv to dvd guide
result: looks bad same as when I watch on my computer.

using CCE-importing avs script1
DirectShowSource("D:\projects\mak5thbday\mak 5th bday clip dv.avi")
ReInterpolate411()
bob
FDecimate(29.97)
result: looks pretty good, TmpgEnc with changed deinterlacing method still looks better


using CCE-importing avs script2
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\Convolution3D.dll")
DirectShowSource("D:\projects\mak5thbday\mak 5th bday clip dv.avi")
ReInterpolate411()
SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
Weave()
DoubleWeave.SelectOdd()
result: looks as good as default TmpgEnc deinterlacing method, which is bad

I'm willing to just use TmpEnc, but I know CCE encodes mpeg-2 with better quality, if I just had a avs script that would produce the same deinterlacing method that I used with TmpgEnc(changed deinterlacing method).

thanks again for all the replies :)

[Guest]

>I really feel that it is Interlaced different than most dv

Nope. It's perfectly ordinary BFF interlaced DV video.

I am thinking now that CCE is encoding your BFF material as TFF, and you haven't selected the right option to allow for that. See FAQ #10 in the CCE forum.

If done right, interlaced encoding will give you better results for playback on interlaced displays, i.e., standard TVs.

[Kika]

@dreadycarpenter
About your settings in TMPGenc. Don't use any kind of deinterlacing, and the encoder will do it's job perfectly. The deinterlacing methodes of TMPGEnc are outdated since... hum... 2002?
And as neuron2, bb (and me) already have postet: For interlaced DV source ecoded to standard DVD, deinterlacing is not neccesary.

[bb]

Well, I took the time and encoded the test clip above using CCE SP 2.66, authored an NTSC DVD and played it on my NTSC capable DVD player (PAL). The picture was nice, although I didn't try very hard :)

CCE settings:

1. Encode setting
- Input files: guess what ;)
- Timecode: everything set to 0
- Output files: Video file only
- Video setting:
-- Mode: MPEG-2 (ES, CBR)
-- Bitrate: Avg 6000 kbits/sec
-- None of the following checkboxes checked: 3:2 pulldown detection, Letter box hint, Panscan
-- Aspect ratio: 4:3

2. Advanced "Video":
- M = 3, N/M = 5 (default)
- GOP header 1
- SEQ header 1
- Checkboxes: only DVD Video compliant checked (Upper field first NOT checked!)
- Luminance level: 16 to 235
- Packet size: 2048

3. Advanced "Quality setting":
Nothing checked except the following:
- Intra block DC precision: 9 (you could use 10, though)
- Block scanning order: Alternate (which should be better for interlaced video)
- Progressive frame flag NOT checked!


This is just plain stupid direct encoding without any advanced filtering, and the picture quality was good. No interlace lines on TV, smooth motion (as smooth as NTSC can be on a PAL display ;) ).
MPEG-2 ES file (http://rapidshare.de/files/6754411/dv_ntsc_interlaced.mpv.html)

bb

[Kika]

Yes, in CCE 2.66 (the Version, i'm still using), these are the correct settings to handle DV-Source. In newer versions, they are different.
And TMPGEnc is also able to handle DV-Source correctly. Only the deinterlacing is the problem here. That's why i postet "Looks good. It's interlaced video, therefore it HAS to look like this... " earlier.

[bb]

@dreadycarpenter:

By the way, if you'd like to try the script from the "DV to DVD" guide you quoted above, delete the DoubleWeave.SelectOdd() line and try again. ;)

bb

[nebula8]

CCE settings:

1. Encode setting
- Input files: guess what ;)

Did you input the files with the avs script mentioned above?

:script: please... :D

I tried it with the avs script above (similar), and with the CCE settings listed in the DV to DVD guide. I burned it to a DVDRW and played it on my standalone, and get a strobe effect in motion areas.

My source is a TV cap using Cam passtrough, captured to DV (II) with WinDV.
This is my script
LoadPlugin("D:\Video\Filters\ReInterpolate\ReInterpolate411.dll")
LoadPlugin("D:\Video\Filters\Convolution3D\Convolution3D.dll")

avisource("d:\myth\cg01.avi", "d:\myth\cg02.avi")

empty = BlankClip()
AudioDub(last,empty)

# for pure black borders & won't waste data encoding border noise ;-)
crop(10,46,-10,-48)
addborders(10,46,10,48)

# Conversion needed for ReInterpolate411
converttoyuy2()
ReInterpolate411()

SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
Weave()
DoubleWeave.SelectOdd()

In a recent post in this thread it suggests to remove DoubleWeave.SelectOdd(). Will that fix the problem?

PS:
could I do this to illiminate the y borders without hurting the interlacing?
crop(10,46,-10,-48)
LanczosResize(720,386)
addborders(0,46,0,48)

And, can the x direction be resized slightly to match the y expansion while retaining proper interlace?

:thanks:

[bb]

Did you input the files with the avs script mentioned above?No, I used the AVI file itself.

I tried it with the avs script above (similar), and with the CCE settings listed in the DV to DVD guide. I burned it to a DVDRW and played it on my standalone, and get a strobe effect in motion areas.That sounds like your field order is wrong. Change it either in the CCE settings or in your AVS script - but not in both!
In a recent post in this thread it suggests to remove DoubleWeave.SelectOdd(). Will that fix the problem?
Most likely, if field order is your problem (which I assume). DoubleWeave.SelectOdd() changes the field order, so deleting this command means to get the other field order. It should work either with or without this command - depending on the field order of your source.

could I do this to illiminate the y borders without hurting the interlacing?
crop(10,46,-10,-48)
LanczosResize(720,386)
addborders(0,46,0,48)Yes, that's all right.

And, can the x direction be resized slightly to match the y expansion while retaining proper interlace?You can resize in horizontal direction without hurting the interlacing as long as your resizing algorithm doesn't touch the pixels in vertical direction. The best way to resize interlaced content is to split the video frames into fields, then resize, and finally weave the fields together to frames again. The scripts in the guide will give you an idea of how to accomplish this.

bb

[henryho_hk]

DoubleWeave.SelectOdd() changes the field order

Will it lead to minor a/v async? Would it be better to use ReverseFieldDominance()?

The best way to resize interlaced content is to split the video frames into fields, then resize

Or should we kernelbob() the content, resize it, and then combine the fields again? Is there any visible difference?

[bb]

Will it lead to minor a/v async? Would it be better to use ReverseFieldDominance()?
Yes, I think so, because it drops a field. I don't know exactly what ReverseFieldDominance() does, so currently I can't comment on this...

Or should we kernelbob() the content, resize it, and then combine the fields again? Is there any visible difference?
I guess the bobbing approach is more accurate, although it's more CPU intensive, too. I can't quantify the difference, but there will be a difference. The problem of the spatial one-line shift between two fields is addressed by the bobbing method; I understand the whole thing as a rounding problem.

bb

[benjaminchoate]

Hello Everybody, feel free to congratulate me on my first post. :D

I've been learning about video for a little while, and decided it was time to start putting it to use converting my miniDV home videos to DVD. This site has been a great resource, many thanks to all the knowledgable people here. I have a (rather cheap) DV cam that I've been using to record various family events with and want to make sure I maximize the already limited quality of the video.

My DV cam records to NTSC (720x480 interlaced @29.970 Hz) DVSD, which I captured to type-2 DV using WinDV. I want to convert it to interlaced DVD (NTSC). I'm currently using this Avisynth script which I found in the DV to DVD guide:

LoadPlugin("C:\Program Files\GordianKnot\AviSynthPlugins\Convolution3D.dll")
AviSource("D:\Home videos\movie.avi")
SeparateFields()
odd=SelectOdd.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(1, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
Weave()
DoubleWeave.SelectOdd()

I'm using the Cedocida DV codec, which is outputting YV12 rather than RGB, so I don't have to reconvert to YV12 for the convolution3d plugin -- also as I understand it, MPEG2 uses the YV12 colorspace, so avoiding unnecessary conversions will help maintain quality (correct me if I'm wrong on any point, please).

Currently I'm feeding that script straight into HCGUI. I'm using all default settings except:

encoding profile: best
Average bitrate: 9608
maximum bitrate: 9608
Interlaced: top-field first
scan method: alternate
aspect: 4:3

I'm recording about 45 minutes of film to a single-layer dvd, os I have plenty of bitrate to spare and am trying to maximize quality. The reason for the odd bitrate is because I read that 9.8 Mb/s is the highest bitrate supported by the DVD standard. I couldn't find out whether that was for just the video, or audio and video both, so I subtracted 192 Kbps from the total just to be on the safe side. :) Clarification on that would be great if anyone knows, as well as tips on a more elegant way of ensuring maximum quality while still meeting DVD standards -- I was thinking perhaps of setting a high Constant Q with a DVD compatible maximum bitrate, but I don't know that that would ensure DVD standard compatibility.

Lastly, and the main reason for posting, is the field order. Finding trustworthy information about the topic is like looking for the Marble in the Oatmeal. :) I'm afraid that by using "DoubleWeave.SelectOdd()" that I might be switching to the wrong field order. On top of that, my avisynth script uses "Interleave(evn,odd)", which I seems to me sets it to bottom field (odd) first? I'm selecting "top field first" in HCGUI because I thought that's what the the NTSC standard called for, but honestly I'm just waaay confused on the whole thing. Any wisdom and tips that can be imparted would be greatly appreciated. I'm kind of a perfectionist and it drives me nuts to think I might be doing this any way that's less than optimal. Thanks again.

[edit] One last bonus question: Even going from an interlaced source to interlaced DVD it seems like I lose a lot of the fluid motion from the dv avis, though it seems that it's as good as it's going to get. I thought maybe it has to do with the fact that I'm viewing it on my computer screen using post-processing deinterlace (I use power DVD to view both my DV-avis, and my mpeg2 clips), but that doesn't seem to be right, because the I can already see the smooth, fluid motion on my computer, in power DVD ... it's only after the conversion that it seems to be lost. Is this a problem inherent in mpeg-2 or something else? The fluidity of the motion is adequate in the mpeg2 that I'm producing and seems to be comparable to any other DVD video that I've seen, but again, I'm a perfectionist and would love to save that "live" look in my DVDs that can be viewed on a standalone player, but it's looking more and more like the only way to do that would be to just copy the dv avis to a DVD and view them on a computer as I am currently, or just lug around my DV camcorder and connect it straight to a TV using S-video. :P Obviously these are less desirable ways of viewing my home movies. :(

[henryho_hk]

Benjamin, when you use Cedocida for YV12 output of interlaced DV clips, remember that you need to set the output mode at "MPEG2 Interlaced". Otherwise, the choma (color) outputs will be wrong for your purpose.

DoubleWeave.SelectOdd() is discarding the very first bottom field and, for a perfectist, it is not very good in terms of a/v sync. I prefer ReverseFieldDominance() which removes 1 line from the top and duplicates the bottom line (or the other way round, configurable).

Since you have plenty of bits to spare in DVD, I think you can leave out the denoisers too.


loadplugin("ReInterpolate411.dll")

clipname="home.avi"

# only if Cedocida not installed or it's a mode 1
#directshowsource(clipname,pixel_type="YUY2")

# for type-2 DV
# remember to disable ffdshow's DV decode
# as we need YUY2, no need to bother about Cedocida's YV12 mode
avisource(clipname,pixel_type="YUY2",fourcc="DVSD",audio=true)

# not needed for Cedocida
#ReInterpolate411()

# switch the video from bottom-field-first to tff
# (remove 1 line from the top and duplicate the bottom line)
# YUY2 only
ReverseFieldDominance()

ConvertToYV12(Interlaced=true)

ssrc(48000)

[benjaminchoate]

Benjamin, when you use Cedocida for YV12 output of interlaced DV clips, remember that you need to set the output mode at "MPEG2 Interlaced". Otherwise, the choma (color) outputs will be wrong for your purpose.
Silly question: How do I pull up the configuration screen for cedocida without going through the motions of starting an encode using it? The dll/inf I installed didn't include any sort of config executable... or do I just have to open a file up in vdub and attempt an encode, alter the settings, then cancel?

[henryho_hk]

How do I pull up the configuration screen for cedocida without going through the motions of starting an encode using it?

well.... i have a silly method....

[Control Panel] -> Icon:[Sound and Audio Devices] -> Tab:[Hardware] -> Item:[Video Codecs]:highlight -> Button:[Properties] -> (Popup) -> Tab:[Properties] -> Item:[Cedocida DV Codec]:highlight -> Button:[Properties] -> (Popup) -> Button:[Settings] -> (Now you can change the YV12 encode/decode modes) -> ok -> ok -> ......

I think everyone prefers your method~~~~

[benjaminchoate]

That works very nicely. :)

Does anyone have any comments about my HC settings? The number for max bitrate, as well as manually setting the avg bitrate/max bitrate to the same thing all feels kind of ghetto to me. Is there a better way to achieve high quality using HC?

[henryho_hk]

Does anyone have any comments about my HC settings?

Just remember to pay attention to the field order.

As for other settings, please ask in another forum. ^_^

[benjaminchoate]

Here's a thought:

What if I used DoubleWeave.SelectOdd(), but then manually offset my audio by 16 ms? Wouldn't that fix the audio problem without having to double up the bottom line? Also, then I wouldn't have to convert to YUY2, then back to YV12...

[henryho_hk]

What if I used DoubleWeave.SelectOdd(), but then manually offset my audio by 16 ms?

Possible. I have not even thought of using DelayAudio(). :helpful:

Also, then I wouldn't have to convert to YUY2, then back to YV12...

DV YV12 is different from MPEG2 YV12 anyway.

[benjaminchoate]

Well it seems to have worked. I loaded my original type-2 avi into virtualdub and demuxed the audio to a .wav. Using audacity, I shaved 16 ms from the beginning of the file and resaved as a wav, then encoded into AC3. I then encoded my mpeg2 using only AviSource(), DoubleWeave.SelectOdd() (with cedocida set to decode to YV12, mpeg-2 interlaced). After muxing I don't notice any sync problems between audio and video.

I suppose technically 1/60th of 1000 ms would be 16.66 (round to 17) but 16 is a nice number... I figure I'm splitting hairs at that point anyway. :)

[odyssey]

I'm SOOOOO confused right now. Those who invented interlacing should burn in hell!

1. Why even use AVIsynth (if no filter is used), instead of just loading the AVI directly into CCE?
2. Okay so i'll try to use the script provided in "DV to DVD"-guide, but many states that DV (at least PAL) is YV12, and i'm using AviSynth2.5 and the Convolution3D plugin only supporting YV12, but getting an error - What is causing this??? (The decoder? And how do I force it to use YV12 all way through?)
3. Why not just deinterlace everything? I tried encoding a DV interlaced file to DVD (directly in CCE to interlaced DVD), but when played on my (cheap) settop DVD player I could definately see the interlacing! However when played in WinDVD it deinterlaced fine.
4. I will maybe use the DVD as source for editing the video later on - Is it still wise not to deinterlace? (And yes I know, I should not encode it whatsoever, but that's what I need right now)

Hopw you can give me some answers...

[bb]

I'm SOOOOO confused right now. Those who invented interlacing should burn in hell!
Well, at least at the time interlacing was invented, it was necessary for technical reasons. Nowadays I prefer progressive video.

1. Why even use AVIsynth (if no filter is used), instead of just loading the AVI directly into CCE?
CCE needs YUY2 input. You may input DV AVI files directly in other encoders, though (e.g. TMPGEnc).
2. Okay so i'll try to use the script provided in "DV to DVD"-guide, but many states that DV (at least PAL) is YV12, and i'm using AviSynth2.5 and the Convolution3D plugin only supporting YV12, but getting an error - What is causing this??? (The decoder? And how do I force it to use YV12 all way through?)
If you decide to use Convolution3D, then use the YUY2 version. The YV12 version has never been finished, thus I don't recommend to use it.
3. Why not just deinterlace everything? I tried encoding a DV interlaced file to DVD (directly in CCE to interlaced DVD), but when played on my (cheap) settop DVD player I could definately see the interlacing! However when played in WinDVD it deinterlaced fine.
You don't deinterlace because deinterlacing introduces artefacts. If you see combing when playing the video on an interlaced device (TV), then you did something wrong. Maybe you resized vertically, or you messed up the field order.
4. I will maybe use the DVD as source for editing the video later on - Is it still wise not to deinterlace? (And yes I know, I should not encode it whatsoever, but that's what I need right now)
I don't see why you shouldn't leave it interlaced - provided that you do it right.

bb

[odyssey]

CCE needs YUY2 input. You may input DV AVI files directly in other encoders, though (e.g. TMPGEnc).I guess not newer versions. I'm able to load in the DV AVI directly without problems
If you decide to use Convolution3D, then use the YUY2 version. The YV12 version has never been finished, thus I don't recommend to use it.For AviSynth 2.5 ??? All I can find for that one is the YV12 version, which barks if it doesn't get YV12 input
You don't deinterlace because deinterlacing introduces artefacts. If you see combing when playing the video on an interlaced device (TV), then you did something wrong. Maybe you resized vertically, or you messed up the field order.Like I said, I just loaded it directly in CCE, and set interlacing options to auto - What could be wrong then? (How do I test interlacing settings in CCE?) Anyway I tried the deinterlacing now, and I don't see ANY artefacts. The quality is yet to be compared...
I don't see why you shouldn't leave it interlaced - provided that you do it right.
I just find it too hard to deinterlace. TMPGENC has a nice previewing feature which are great, but aren't there general rules that apply to interlaced PAL DV???