digitalFAQ.com Forums [Archives] - Trying to make KVCD from DV home movies

digitalFAQ.com Forums [Archives] (http://www.digitalfaq.com/archives/)

- Video Encoding and Conversion (http://www.digitalfaq.com/archives/encode/)

- - Trying to make KVCD from DV home movies (http://www.digitalfaq.com/archives/encode/7520-make-kvcd-dv.html)

Trying to make KVCD from DV home movies

I spent the better part of this weekend trying to make a KVCD from a 90 minute home movie compilation.

I have run into a few snags.

Most of the filters for Avisynth 2.53 require YV12 colorspace. DV Video (at least mine) is in RGB24 colorspace. So try as I might, I cannot use any Avisynth 2.53 optimal script. The one workaround that I thought would work would be to include:

ConverttoYV12() # right after Avisource("c:\video.avi")
# and
ConverttoRGB() # last line of script - I need this line in all my scripts with YV12 source or I get the "solarization" effect in TMPGenc

This did the trick, in that I didn't get any error messages back, however it had a nasty side-effect of flickering lights on the right side of the video.

So I gave up on Avisynth 2.53 and went back to my old 2.08 with the following script:

Quote:

AviSource("D:\My Documents\My Videos\alexa\alexa.avi")
fielddeinterlace(full=false)
LegalClip()
GripCrop(352, 240, overscan=1, source_anamorphic=false)
GripSize(resizer="lanczosResize")
spaceDust() # or PixieDust()
unfilter(50,50)
Convolution3D (0, 32, 128, 16, 64, 10, 0)
mergechroma(blur(1.58))
mergeluma(blur(0.2))
GripBorders()
LegalClip()

Although watchable, the quality is terrible when compared to the original DV avi source. Obviously, since I resized to 352, 240! But I had to do that in order to fit 90 minutes on one CD. Even 352, 480 looked worse. FYI: I used CQ_VBR with a CQ value of 32.7 (from CQMatic - which worked well, fit perfectly on one CD).

So, enough rambling, I have a few specific questions:

1. Can anyone share their experience converting DV to KVCD? (ok, that's a general question).
2. What is the best way to use Avisynth 2.53 with a DV avi source file?
3. What optimal script should I use?
4. What is the best way to de-interlace? I wanted to try incredbile's method from this thread:

http://www.kvcd.net/forum/viewtopic.php?t=8306

but I didn't get very far because I couldn't get anything to work on the RGB source.

Any help would be greatly appreciated!

Thanks,
Icarus

Come on... There must be someone out there who uses Avisynth 2.5 and converts DV to KVCD?

Please, tell me how you do it, what script you use, etc...

I'm getting desperate here... I really can't afford a DVD Burner right now!

Thanks,
Icarus

Quote:

Originally Posted by Icarus3000

I bought a DVD burner mainly for DV conversions and never did DV->KVCD, so I can't offer too much assitance... :( All I use in my scripts are UnFilter and FluxSmooth, to get rid of noise (the CCD in my camera likes to produce it in low light).

Quote:

Most of the filters for Avisynth 2.53 require YV12 colorspace. DV Video (at least mine) is in RGB24 colorspace.

DV is actually closer to YUY2 (in NTSC, at least), so it's too bad that the DV codecs we're using (I'm using the Panasonic one) convert to RGB first. Ideally it would be YUY2, but that won't work with the motion adaptive script.

Quote:

So try as I might, I cannot use any Avisynth 2.53 optimal script. The one workaround that I thought would work would be to include:

ConverttoYV12() # right after Avisource("c:\video.avi")
# and
ConverttoRGB() # last line of script - I need this line in all my scripts with YV12 source or I get the "solarization" effect in TMPGenc

Is it still solarized in VirtualDub without the ConvertToRGB line? And what codec does it think you're using when you load the avs (look in File Information in vdub)? I remember reading once that some YV12 codecs (ATI's, for one) didn't do colourspace conversions properly. This doesn't sound like the same thing, but it's something to look for.

Re: Very frustrated trying to make KVCD from DV home movies

Quote:

Originally Posted by Icarus3000

1. Can anyone share their experience converting DV to KVCD? (ok, that's a general question).

DV is most recorded using BOTTOM Field First! Thats very important when deinterlacing!

So you should try my recommendation but using BFF:

SmoothDeinterlace(tff=false, doublerate=true, blend=true).Selecteven()
(also try one using selectodd() at the end instead of selecteven(), and choose which looks better.

Quote:

2. What is the best way to use Avisynth 2.53 with a DV avi source file?
3. What optimal script should I use?

It depends on the quality your DV Device delivers.

Quote:

4. What is the best way to de-interlace? I wanted to try incredbile's method from this thread:

As I told above, use the TTF-false method in Smoothdeinterlace.
But in case of NTSC you will end up with a 29.976 deinterlaced stream.

So first use this script:

Code:

Avisource("YourDVsource.avi)

ConverttoYUY2()

SmoothDeinterlace(tff=false, doublerate=true, blend=true).Selecteven()

No open the avs in vdub and look at the "File Information" and the mentioned Framerate, it should be still 29.976 but now Framebased.

Now we include the following function to do a conversion from REAL 29.976 to 23.976 FPS

Code:

Avisource("YourDVsource.avi)

ConverttoYUY2()

SmoothDeinterlace(tff=false, doublerate=true, blend=true).Selecteven()

convert60ito24p(2,0) 

# 

############### the function ############# 

# 

function convert60ito24p (clip video, int mode, int offset) 

{ 

work = assumefieldbased(video) 

out = (mode==2) ? interleave( 

\selectevery( 

\layer(trim(work, 1, 0), 

\layer(work, trim(work, 2, 0), "fast"), 

\"fast"), 5, 0 + offset), 

\selectevery( 

\layer(work, trim(work, 1, 0), "fast"), 5, 3 + offset)) : 

\* * (mode==1) ? interleave( 

\selectevery(trim(work, 1, 0), 5, 0 + offset), 

\selectevery(layer(work, trim(work, 1, 0), "fast"), 5, 3 + offset)) : 

\* * (mode==0) ? selectevery(work, 5, 1 + offset, 4 + offset) : work 

assumeframebased(out) 

} 

#################################

And now watch your avs in Vdub.
Its ok? How is the playback, it schouldn't be jerky or stuttering.
Because NOW you got a 23.976 progressive stream and it should look like film material which can be wonderful encoded to KVCD as it now needs not that much bitrate cause of less FPS!

Now below this script you can add your filters as you like.
TemporalSoften, CNR2, Deen("a2d",x,x,x) ... just find out what ends up as best.

(If you gonna use YV12 filters just change the converttoYUY2() at the beginning of the script to ConverttoYV12() as Smoothdeinterlace supports YV12

Thanks for the response Incredible!

It feels like a tease though since I am at work right now, and have to wait until I get home to test it out!

:D

hey, I'm new here and I have a lot of DV to convert, and I'd like it all to be "archival quality". how come you're resizing it so much? How did that script work out for you? Could you post the entire script so I have something to start with?

Also, again forgive my noobness, but is the reason you're having him convert frame rates so he can just use one of the standard templates? Is this less hassle than modifying the templates?

I've been experimenting with quite a few scripts recently.

I still haven't made up my mind yet, but will post my "personal" "optimal" DV script when I find it!

- Icarus

I'm still working on mine, just wondering what you guys have so far.

I posted by "optimal script" here:

http://www.kvcd.net/forum/viewtopic....283&highlight=

What are you using?

Thanks,
Icarus

pretty much the same one as yours. except i can never get blindPP to work so I took that out. However i'm not happy with what i'm getting.
When i use yours i get errors saying i can't use smoothdeinterlace unless it's in yuv2, so i've added that. and I can't get blindPP to work no matter what, i've read what people have suggested with the adding the rgb thing at the end and stuff. mine's ending up just too blocky. what can i do to eliminate the blockiness?

welthqa
If your results are blocky, it usually means that the average
data rate is too low for the frame size. Try increasing the
CQ setting and the max bit rate. BlindPP(cpu=4) requires
loading the MPEG2Dec3.dll. Putting SmoothDeinterlace
right after AVISource works for me in AviSynth 2.53.

welthqa
The AviSynth 2.0 DLLs will not work in AviSynth 2.5 Maybe you are inadvertently trying to use some 2.0 DLLs in AviSysnth 2.5.

which 2.0 dlls are you talking about? the MPEG2Dec3.dll?
also, i can't seem to get prediction to work. seems like calcumatic doesn't load avs or the dv avi that i'm using. cq matic works otherwise, if i guestimate the bitrate, though i admit i'm not sure what bitrate to use.

ohoh, yes, i was using old MPEG2Dec3.dll, i dl the new one, and it works, thanks for that reminder.

now to solve the prediction thing.

Quote:

Originally Posted by welthqa

now to solve the prediction thing.

Do you know how to use CQMAtic for other things than DV avi ? Or you never used it before ?

never really used it before. I've opened projects that I've saved as textproject and the settings from tmpgenc. but the calcumatic doesn't seem to accept dv avi. I know how to change the prediction and the full encode settings. But I guess, no I haven't used it for anything other than DV.

If you are using the Icarus3000 script, you should be able to get fair quality 720x480 video using CQ70, max bit rate 2000, min bit rate 300. You might try that. Remember, CQMAtic does not assure good video quality, only that it will fit on CD.

Hi all !
I'm trying to convert some DV cap's with Avisynth 2.54(tryed 2.53,2.52,2.5- it's not avisynt problem)but Avisource can't oppen My DV files .
Same with VirtualDub - Can't open Avi file .
I know that it's missing codec problem .
Maybe u can suggest wich one is best and where to download it from ???
bman

The DV have to be type 1. Are yours type 1 or 2 ?

bman and others,

If you're having trouble opening your DV footage in avisynth, my advice,
since I do this w/ my DV source (Sony TRV-22) is I open my DV (Type 2)
inside vdub, and then I frameserve into avisynth.

Or, you can just open your DV inside vdub, and framserve into TMPG.
One thing to note, is that vdub does not like Type 1 files. It strip (or ignores)
the audio part, and if you ignore the error/warning message from it, it will
open your source in "video only" mode (no sound)
.
.
But, if you want to open your DV footage directly inside an .AVS script, then
from what I understand, avisynth will only open Type 1 files. This means
that you will have to resort to MUXing your audio at a later step. And, I'm
sure you have that process nailed down :lol:

My prefered method is to first open inside vdub (cause I like to hear the
sound, among other things like some preferenced filtering) and then either
I will frameserve that into TMPG, or I will further process it in an .AVS script,
and then frameserve it into TMPG.
.
.
There is no quality loss in doing it the way I just described, unless you start
incorporating a few ConvertTOrgb(); ConvertTOYUV(); ConvertTOyv12()
etc, because it moves the colorspace around, and each time you call it, it
moves the colorspace around some mores. Like pushing a red paintbrush up
and down a piece of paper. The more you move it, the more color it gets
or whatever :roll:

-vhelp

bman, pansonic's DV codec works fine for me. http://users.tpg.com.au/mtam/install_panvfwdv.htm