DV film to SVCD

[barabas]

I want to make SVCD's from my camcorder DC films with DVD2SVCD. I can capture 19 minutes a time, because the file is then 4 GB big, the limit of Win98 that I use. I heared that the solution is that I have to divide my complete movie into several clips, all 4 GB big, and use a certain avisynth script in DVD2SVCD, so that the clips are used in the right order. This way I get around the 4 GB limit. I don't know how to program the script. Does someone have a tip?
Greetings.

[Mug Funky]

check the avisynth forums, dude.

chaining clips together is easy, and you shouldn't use DVD2SVCD for that - rather just use an avisynth script and serve that directly into your MPEG-2 encoder of choice.

[barabas]

yes that's the solution!!

[omer123]

Hi! I'm trying to convert my dv films to mpeg-2 svcd using avi2svcd and CCE 2.5.So far I have been converting AVI DivX movies to svcd and everything came out great,but...when I covert dv to svcd the movement of people and objects doesn't look like the in the original film,like some frames are missing.Someone told me it is because the dv films are interlaced,and I shouldn't de-interlace. Is that true? Anyway,what do I need to do to get "real" natural movement?
Thanks in advance!:)

[barabas]

When converting DV to SVCD with DVD2SVCD, I get a warning "avisource cannot find a suitable decompressor". I installed dvcodec.exe and canopus file converter to convert DV avi to what actually? It gives several options. But none of the options will solve the warning. Does anyone have a clue?
Regards.

[bb]

@omer123:
Deinterlacing will make your video more compressible which is better for SVCD. There are different deinterlacers available; try Donald Graft's kerneldeint for best results. Some people say that the prefer to let the video interlaced, even at SVCD bitrates. You can try that, too, and see what you like better. Interlaced MPEG-2 needs much higher bitrate for the same quality compared to progressive (de-interlaced) material, though.

@barabas:
GSpot can tell you which codec was used to create your AVI (at least the fourcc). Install the codec, and AviSynth should read the file. Canopus file converter is not a codec (there's a Canopus DV codec though) but a type-1 / type-2 converter. If your DV file is type-1 (fourcc "ivas") then you can use this tool to convert your AVI into a type-2 file. The type-2 AVI will come out a few percent bigger, but Video for Windows applications will be able to deal with it, including the audio. You still need to have an appropriate VfW DV codec installed, though.

bb

[Kika]

At SVCD-Bitrates for interlaced Video, i suggest to use ProCoder for Encoding the Videos (Mastering Quality). It is slow, but good for that.

An alternate Way is to produce an interlaced CVD (Half-D1).

[omer123]

Thanks for your comment.I tried leaving the file interlaced.I learned from one of the forums that I should unchek "progressive frames" and "zigzag scanning order" in AVI2SVCD but I wasn't sure about choosing "NO Deinterlacing" or "Leave interlaced".Anyway,I tried both,and picture came out shaky.Any explanation?:confused:

[omer123]

[QUOTE]Originally posted by Kika
[B]At SVCD-Bitrates for interlaced Video, i suggest to use ProCoder for Encoding the Videos

I tried that,with 2 pass VBR and lower field first (and again with "automatic") and quality wasn't that good.Anything else I should have changed?

[Kika]

Don't use 2pass. It decreases the quality of the resulting Video!
I don't know AVI2SVCD, no Idea what the correct Settings are.

[bb]

Originally posted by Kika
Don't use 2pass. It decreases the quality of the resulting Video!

Well, that depends very much on the the film's length and the desired number of SVCDs...

@omer123:
If you want to stick with AVI2SVCD, and you don't want to deinterlace, you should use "Keep interlaced" on the "Conversion" tab and edit the generated AviSynth script. There should be a "DoubleWeave.SelectOdd()" line after the "Weave()" command. If it isn't there, add it, else delete it, and see whether the motion is still jerky.

bb

[omer123]

well,there was no such line so I will try to add it,see how it goes.How do I get this deintrelancing software (couldn't fine that name you mentioned)

[bb]

Originally posted by omer123
How do I get this deintrelancing software (couldn't fine that name you mentioned)
Which software? If you're talking about the "DoubleWeave.SelectOdd()" command: this is included in AviSynth, no special software required.

bb

[omer123]

[QUOTE]Originally posted by bb
[B]@omer123:
There are different deinterlacers available; try Donald Graft's kerneldeint for best results.

I refered to the deinterlacers you mentioned...Anyway,the "DoubleWeave.SelectOdd()" command really solved the problem.Thanks for that one!

[bb]

@omer123:
The deinterlacers I mentioned can be found at Donald Graft's fabulous filter site: http://neuron2.net/

By the way: please edit your posts, so that the quotes are readable. I guess you forgot to add some closing tags. If you don't know how to do it, I can do it for you.

bb

[omer123]

Yeah...sorry about that.I tried the smart deinterlacer ver 2.7 and the results were good,but one thing bothers me: Everytime I deinterlace my dv stuff the outcome doesn't look like a HOME VIDEO ,but like a movie or a video-clip (peoples movements look "slower",like some frames have been taken out).Is there anyway to solve this or I just have to make peace with that? That doesn't happened when I keep the file interlaced, but then quality is reduced (little cubes here and there).Thank for your patience,BTW.

[Mug Funky]

read up on interlacing dude.. you'll see why the motion looks more filmlike (actually looks better if done right IMHO).

basically, TVs and video cameras scan twice per frame, consecutively. one scan for lines 2,4,6,8... and one for 1,3,5,7...

this means that 2 separate moments are placed in one frame, causing the interlace artifacts. when played on a TV, the interlaced frames are played back in 2 scans as well. so even though PAL is 25 frames per sec, there are 50 scans per sec, each one different. the 2 scans in each frame are referred to as "fields".

when you deinterlace, the frame is stripped down to just one moment, shared across 2 scans. so the motion appears less fluid (but more "filmlike"). usually, from the point of view of resolution, deinterlacing is bad. but if you use a good one that preserves non-moving information in the second field, you don't lose much at all except for some fluidity.

from a compressing point of view, deinterlacing is good - it's easier to compress a regular picture without having to worry about the "mice teeth" all over the picture. there's also less hassle with motion detection and fancy stuff that BB undoubtedly knows waaay more than me about.

so if you can spare the CDs and want your DV to look like "video", then don't deinterlace. if you want better overall image quality at the expense of losing temporal information, then deinterlacing is the way to go.

when i shoot stuff, i tend to deinterlace simply because i wish i was working with film... got a killer avisynth script to do a custom deinterlace too...

[omer123]

Thanks for all the information,it was essential for a begginer like me.Can you share that script with me...?

[Mug Funky]

i should probably start a thread for it... i believe it to be of the calibre of MF's famous scripts.. the really nice but REALLY slow ones.

if you're after quick encodes, this deinterlacer is not for you - i haven't done much more than proof-of-concept, so it could be faster.

ideally somebody could code a nice simple filter for it.. but watch for a new thread in the avisynth user's forum