Log in

View Full Version : I thought decomb was supposed to give me a 23.976 fps?


lancer
31st December 2002, 00:34
I'm using the new 405 decomb legacy dll. not the 406b one mr graft just put out. I thought I had this right in my head but now I'm a bit confused.

I captured a 29.97 fps NTSC source to an avi file. then after running this setting on decomb as part of an avisynth script as a test

LoadPlugin("decomblegacy.dll")
a = OpenDMLSource("side1testfirst bit1.avi")
b = OpenDMLSource("side1testfirst bit2.avi")
c = OpenDMLSource("side2huffyuv2.avi")
d = OpenDMLSource("side2huffyuv3.avi")
e = OpenDMLSource("side2huffyuv4.avi")
f = OpenDMLSource("side2huffyuv5.avi")
g = OpenDMLSource("side2huffyuv6.avi")
h = OpenDMLSource("side2huffyuv7.avi")
i = OpenDMLSource("side1test.avi")
AlignedSplice(a, b, c, d, e, f, g, h, i)
ConvertToYUY2()
Telecide(guide=1, post=true)
Decimate(cycle=10)
Crop(0, 88, -0, -72)

i end up with an avi file with a framerate of 27.353 fps.

I thought decomb was supposed to make it 23.976? or have I got a setting wrong.

should cycle be 5? help. I've got my file looking like the laserdisc it is taken from but the framerate is obviously off.

my understanding is

capture at 29.97 to make my starting avi
decomb to 23.976 to make my interim avi
tmpgenc as a progressive 23.976 source with the 3:2 pulldown flag set.

what step have I missed?

man I thought I had this licked.

scmccarthy
31st December 2002, 00:45
Decimate(cycle=5)

which is the default

gives 23.976.

This tells decimate to drop or blend every fifth frame.

Stephen

scmccarthy
31st December 2002, 00:51
@lancer

How did you arrived at Decimate(cycle=10)? This drops every tenth frame, only half the amount of frames you need to drop to go from a frame rate of 29.97fps to 23.976fps.

lancer
31st December 2002, 00:55
bollocks

so cycle 10 got me halfway yeah. what a ninny. should have looked at the decomb instructions a little more closely.

oh well. if that's the only hurdle I got left to cross I'm happy but something tells me this is only the beginning, I've got menu authoring to put up with yet.


oh joy. I can't wait.

scmccarthy
31st December 2002, 01:17
@lancer

I can tell you are from England now, since that is a swear word nowhere else.

I just did the math and you should use a frame rate of 26.973fps with decimate(cycle=10) to maintain sync with the audio. Instead of Decimate(cycle=5), Decimate() is fewer keystrokes and the same thing since it is the default.

Do you actually have NTSC video in England rather than PAL?

Stephen

Guest
31st December 2002, 07:04
Originally posted by scmccarthy
I can tell you are from England now, since that is a swear word used nowhere else.
I figured it out from his Location field. :)

And isn't bollocks used in Australia too?

scmccarthy
31st December 2002, 08:05
@neuron

I actually really truely etc..(I overdoit don't I) don't think so.

And I have listened to the original soundtrack of MadMax, newly released on DVD in the US. I think that makes me an expert on the subject.:p

Stephen

PS I have the RoadRants SubRip files to that movie too, if anyone is interested. It was a particuliarly hard one to get right and I did it nearly perfectly.

lancer
31st December 2002, 09:43
yeah england. nope no NTSC here but us chickens. good old PAL, higher res, 4% speedup so gotta go NTSC for the right source sound.

in my ignorance I was just fiddling with the decomb cycles . didn't realise that the cycle determined my final frame rate. I stupidly thought it just gave you 23.976 no matter what if you fed it a PAL or NTSC signal.

scmccarthy
31st December 2002, 16:33
I stupidly thought it just gave you 23.976 no matter what if you fed it a PAL or NTSC signal.
This gives me chart blanche to talk about two of my fav subjects. IVTC and audio.

But instead of IVTC I'll just point out that the telecining process takes 24fps and converts it to 30fps fields from two different frames into two frames, so you end up with three original frames followed by two interlaced; and five frames for every four you had originally. In order to compensate for the greater number of frames and staying in sync with the audio, the frame rate is changed to 30fps

And here is where we get to the audio part, For NTSC the framerate and the audio are slowed in sync to 29.97fps and for PAL the framerate and the audio are sped up to 25fps. That is speeding it up a lot more than slowing it down, but neither is exactly the original speed of audio.

I use avisynth to change the framerate with the audio in sync like this:AssumeFPS(24,true)
True changes the sample rate of the audio to match the change in frame rate from what the d2v file says it is to the one given in AssumeFPS. ConvertFPS does not do this, that is one of the differences between the two.

Alternately, I use BeSweet to speed the sample rate 999 to 1000 for NTSC. There are presets in BeSweet to use for PAL.

Once the audio sample rate is changed, it won't compress to mp3. I use SSRC in Besweet to convert the sample rate back to 48 or 44.1 kbps. Often, I compress the audio after I compessed the video. But after the additional hassle, you get a movie in true 24fps whether you started with PAL or NTSC. You can use Besweet to do it all in one step, however. Then put in the avs file this line instead:AssumeFPS(24,false)
This is something I meant to post to the audio forum sometime, so it is fine with me if someone wants to move this to that forum now.

Stephen

scmccarthy
31st December 2002, 16:43
Also, I wonder why it is called telecide(), that name implies it does IVTC, but Decimate comes in the finish the IVTC job. Doesn't telecide actually only perform deinterlacing? I know how to use it, but why there is a separate telecide and field deinterlace is a little confusing to me.

Stephen

Cyberia
31st December 2002, 18:09
Yea, I wonder that about the name also. Why is it called "telecide", and not "decomb"?

But then again, I didnt write it, so I don't get any say in the name :) He could have called it "fragglerock.dll"

Belgabor
31st December 2002, 18:16
Probably because it decides on the right frames which decimate removes then?

Si
31st December 2002, 19:59
AFAIK telecide == telecine killer

c'est vrai n'est pas Don?

Happy New Year

Simon

Guest
31st December 2002, 20:22
Originally posted by siwalters
AFAIK telecide == telecine killer
You win the cigar.

Guest
31st December 2002, 20:53
Originally posted by scmccarthy
why there is a separate telecide and field deinterlace is a little confusing to me.Originally there was Telecide, which did field matching only and no postprocessing, and there was FieldDeinterlace(), which did postprocessing. For performance reasons, the postprocessing was integrated into Telecide. But FieldDeinterlace was kept for standalone deinterlacing.

Metalazzo
31st December 2002, 21:16
Originally posted by siwalters
AFAIK telecide == telecine killer

c'est vrai n'est ce pas Don?

Happy New Year

Simon

;)

scmccarthy
31st December 2002, 21:57
Originally there was Telecide, which did field matching only
Certainment, c'est vrai, mais...

That implies that telecide does IVTC, but it only does that in conjunction with decimate. Now I am beginning to sound like I don't like the way you separated the fuctions, but I am willing to assume that there are advantages to doing it that way. If something works, I happy to learn the complexities of it without questioning it. But while I can learn each functions separately (ie, telecide, fielddeinterlace, and decimate) and how telecide() works with decimate(), I am not clear on how or what telecide might comunicate with decimate. In fact, I always assumed that they work completely independantly of one another.

Also, does 'field matching' mean determining which are the two interlaced frames in every five frames of telecined material and deinterlacing them. I am unsure of that terminology.

Back to the first point, I already saw the connection between telecide, homicide, and pesticide. So, originally telecide was for preparing telecined material for fieldeinterlace and decimate, but now you only need decimate with telecide?

Stephen

Guest
31st December 2002, 22:36
Originally posted by scmccarthy
That implies that telecide does IVTC, but it only does that in conjunction with decimate. Now I am beginning to sound like I don't like the way you separated the fuctions, but I am willing to assume that there are advantages to doing it that way.Telecide kills the combing produced by telecining. If you also want to get rid of resulting duplicate frames, then you add Decimate.

Note that some applications require Telecide but not Decimate, e.g., field-shifted PAL.

If something works, I happy to learn the complexities of it without questioning it. But while I can learn each functions separately (ie, telecide, fielddeinterlace, and decimate) and how telecide() works with decimate(), I am not clear on how or what telecide might comunicate with decimate. In fact, I always assumed that they work completely independantly of one another.They did work independently, but I am just now adding a communication channel between them, because it can help in certain situations. For example, Telecide has an idea about which frames are from 3:2 sequences (film frames). This knowledge can be used by Decimate, for example in Decimate(mode=1) and the upcoming Decimate(mode=3), you want to decimate film and non-film sections differently.

Also, does 'field matching' mean determining which are the two interlaced frames in every five frames of telecined material and deinterlacing them. I am unsure of that terminology.The process of matching is well explained in the Decomb help file Appendix. It is simply the process of finding two matching fields for each frame to be output. This can be done in the absence of any pattern guidance.

Back to the first point, I already saw the connection between telecide, homicide, and pesticide. So, originally telecide was for preparing telecined material for fieldeinterlace and decimate, but now you only need decimate with telecide?Yes.

manono
1st January 2003, 03:01
scmmcarthy-

I am willing to assume that there are advantages to doing it that way

What if you have an NTSC R1 DVD that was "telecined" off of a 25fps PAL source? Then you'll need to Decimate(6) (maybe with AssumeFPS(25) added). I've seen plenty like that.

What if you have a silent film originally at 20fps? Then you'll need to Decimate(3) (maybe with AssumeFPS(20) added). I've seen plenty like that, and with other unusual frame rates also.

No other IVTCs can handle those kinds of source materials properly.