Telecide / Telecine

[JimMc]

TELECINE and TELECIDE are they the same thing? Or is it a typo

I always used the TELECIDE(PAL) option on DVD2SVCD on PAL projects as the results for the same bitrate were always much better IMHO.

Is the TELECINE option on DIF4U the same thing?

I will be honest, im not too sure what TELECIDE is or does but I like the results.

If there the same thing no problem if there not can a script or option be changed in the big 3 to allow me to get similar result.

[Matthew]

You shouldn't be using telecide if your source doesn't exhibit combing, and even then, for pure interlaced video (extras come to mind) apparently other de-interlacers are better, as decomb was not designed to be a pure de-interlacer as such.

http://www.doom9.org/ivtc-tut.htm may be of assistance.

[JimMc]

Hi Matthew
Thanks for answering.

Ive got to say though that my experiences dont match up to the theory.

I have NEVER had the combing effect appear on any of the pal titles I have done (not sure how many but quite a few). I HAVE followed the guides and LOOKED for it.

My use of Telecide came about by "giving it a go" I used all the settings in DVD2SVCD and judged the end results, SAME FILM, SAME BITRATE, SAME END FILE SIZE. TELECIDE(PAL) won by a mile. It has also, always produced excellent quality extras at far lower bitrates than I would have thought possible.

THE GUIDE YOU POINTED ME TOO.
Cheers for that Ive just had a quick read and one bit I noticed has gone some way towards helping.
3. Use another AviSynth filter, such as Telecide.
So now I know where to start playing and rewriting for a way to innitiate it. UNLESS YOU OR SOMEONELSE wants to give me a shove in the right direction:D

My way might not be the technically right way, but hey its the end result that counts.
JimMc

[Matthew]

I wouldn't expect you to see combing on PAL movie DVDs (extras aside).

Anyway, I have little idea about deinterlacing, however AFAIK it damages progressive material (in that it makes the output less like the source than if no deinterlacing were applied). So perhaps decomb is acting as a quasi blur filter or something like that, which would reduce pixellation but also reduces detail. Now if that floats your boat, fine, but there are other things specifically designed for that - e.g. CCE's noise filter or avisynth's blur filter.

Might be a good idea for you to post your thoughts about telecide improving the quality of output for progressive PAL material in the Avisyth Usage forum ;)

[Kedirekin]

If I'm not mistaken, Telecide and DeComb share a common heritage.

The filter has a threshold - it won't apply any deinterlacing if it can reconstruct a frame that is 'progessive enough'. If the frame is progessive to begin with, it would almost certainly fall below the threshold and bypass deinterlacing.

Also (again, if I'm not mistaken) telecide is ideally suited to reconstructing progressive frames on field-shifted progressive PAL video - though there may be better more efficient ways to do it (swap field order in DVD2AVI).

However, I don't see any way that Telecide would improve material that is already progressive. It basically boils down to a very expensive no-op.

FYI: Telecine is a process that duplicates field in 24 fps film material to make it suitable for playback on an NTSC television. In effect, it turns 4 frames into 5.

Telecide is a filter than can often reverse telecine (provided the telecine method used was 3/2 pulldown), but it has other uses - as described above.

[Matthew]

Originally posted by Kedirekin
The filter has a threshold - it won't apply any deinterlacing if it can reconstruct a frame that is 'progessive enough'. If the frame is progessive to begin with, it would almost certainly fall below the threshold and bypass deinterlacing.

mmm...yes you'd be right, silly me was thinking in terms of FieldDeinterlace() with "full" not set to false. But of course this wouldn't apply to telecide with postprocessing enabled.

[JimMc]

WOW!

Last 3 Posts- Way over my head!
But Ive been reading up some of what you have wrote-give me 3 of 4 years, and I might be able to hold my own in this conversation:D

On a more serious note I have taken in a bit, and am at this moment doing a rerun with more "standard" settings for the material Im using.

"FLOATING MY BOAT", Just to expand on what I posted previous, Surly the end result of what we are trying to do is to get 60% of a dvd to play on our screens the same as the 100% original. Now I believe what I have achieved doing it the "wrong way" is pretty damn close.
Ive got a BIG HQ CRT TV which really does show up the bad bits, the halos the graineyness the lot, My aim is alway to get the picture as close as possible to the original, I would have to say thats what Ive been achieving, even having to check the box on a few occasions to make sure I WASNT playing the original.

BUT Im willing to learn and try other things, my way had its problems mainly down to having to reencode things more than once as my calculations on AVE bitrates with DVD2SVCD left a lot to be desired :mad:

Thanks Guys for your comments/Help.
Im learning a few bits. I just hope at my age by learning somthing new I dont forget the stuff I used to know, like how to drive the car!
JimMc

[Matthew]

mmm...well might as well explain the situation as I understand it then...

Movies are almost always shot on film at 24 fps, which makes them progressive. When placed on PAL DVD, the movie is sped up to 25 fps so everything plays 4 percent faster (and movie is 4 percent shorter). However, sometimes the video stream is "padded" to 25fps so that the length of the film remains constant. The result of this is a movie that appears to be interlaced. As Kedirekin said it's "field-shifted progressive PAL video".

Now, what telecide will do is some field matching (or something, dunno the technical details -at all-), so that the progressive frames are reconstructed rather than deinterlaced. Apparently the result is duplicate frames, which can then be deleted.

Anyway, this field matching isn't 100 percent and some interlaced frames sneak through, so what telecide's postprocessing does is detect these frames and then de-interlaces them. That is, it looks at each frame and decides whether it is interlaced, if yes it de-interlaced if no it leaves it alone.

So, when you use telecide on a progressive stream, what should happen according to Kedirekin is that telecide doesn't do any reconstruction of progressive frames because there's none to be done, then it looks at each frame and de-interlaces it, *IF* it's detected to be interlaced. But, as your source is progressive it won't do any de-interlacing (short of the odd mistake here and there).

As an aside, my FieldDeinterlace comment relates to that fact that using FieldDeinterlace(), which is a sister command of telecide, de-interlaces each frame. This would damage progressive frames if you used it.

FieldDeinterlace(full=false), however, does what telecide's post-processing does - detects whether a frame is interlaced and only then de-interlaces it (good for hybrid interlaced/progressive material I suppose).

Anyway, what it boils down to is this. Either:
1) You are imagining things and telecide does nothing to your stream; or
2) For some odd reason telecide does affect the stream in a significant way.

If 2) then you are better off achieving this effect using some other means. Lets say, for example, telecide is destroying detail. This makes your video look blurrier and hence easier to encode, which means less pixellation. That's a good thing at low bitrates and why SVCD is lower resolution than DVD. But it's better to achieve that effect using the tools/settings designed for it (blur filters, anti-noise filters etc).

[JimMc]

Hi Matthew.

Im getting there, Ive spent most of the weekend running various things thru DIF4U.
Results much better, but this is mainly due to me deviating from the guides a touch. My biggest problem has been following the guide blindly and expecting the program to Do it all for me.

I have had to put in much more HUMAN intervention than I thought I would have to, relying much more on information gained from bitrate viewer etc before I start.

My results so far.

3 PROGRESSIVE sources- No problems at all good clean results. DIF picked the right settings Didnt have to change a thing. (Actually I did have the wrong settings selected once by DIF4U seconfd run thru of same source right settings selected!!!)

Interlaced, these were the 2 originals I ran thru, which sparked off this post, I found I had script changes to make, and also make alterations in DCCE, results much closer to the original.

Jim
Now getting right results the right way!!!