Cinemacraft SP and 3:2 Pulldown - Beginner Questions

[dvdboy]

I've had a client supply me with a film on Digi-Beta. They contacted me before had and said the film was in 23.98fps, and I requested the film be laid off to digi-beta at 29.97fps with 3:2 pulldown.

I've recieved the tape and captured it as an AVI file.

Can someone please give me a few pointers as to how to encode this correctly in CCE SP?

I've gone through the obvious steps, and selected pulldown from with CCE SP. I've opened up the settings window and when I step through the film, I can see a clear pulldown sequence, but I kind of got lost at this point, chickened out and used the 2-pass automatic detection option. Is this ok?

Opening the m2v file up in DGIndex, when playing the stream it is correctly flagged as FILM, although for some reason the colorspace is set to BT.709 not BT.609...

How should I look to encode the tape correctly for 3:2 pulldown, and how do I ensure there are no cadence breaks?

I've used AVISynth before to IVTC an AVI file - is this a better way (feed CCE-SP a 23.98fps AVI and get it to add the pulldown).

Any help would be greatly appreciated!!

DVD-BOY
(Wishing there was a CCE-SP course he could go on...)

[PhillipWyllie]

I wouldn't get CCE to add the pulldown if I were you. Ask for the original source and capture at 23.976,then encode at 23.976(progressive),then use Neuron2's rather excellent pulldown tool to turn the mv2 into 29.97(interlaced),without adding any frames or re-encoding.

note: DGPulldown requires pregressive video streams, so if your source is interlaced and you want to keep it that way then the above method isn't for you. I'd still not trust CCE, but use avisynth instead.

[dvdboy]

Thanks Philip,

So if I have an 29.97fps interlaced video file, my best bet would be to use an AVISynth Script to IVTC it back to 23.976, feed this to CCE-SP to create a 23.976 MPEG-2 stream, which I then feed to Neuron2's pulldown tool to get a 29.97fps stream to author with?

DVD-Boy

[FlimsyFeet]

I wouldn't get CCE to add the pulldown if I were you.
Any reason why not? Just interested because this is the way I always do it.

[neuron2]

So if I have an 29.97fps interlaced video file, my best bet would be to use an AVISynth Script to IVTC it back to 23.976 You cannot apply IVTC to an interlaced source, but only to a telecined source.

[dvdboy]

Apologies, I think I'm using the wrong terms from memory. If I have a 29.97fps interlaced master, which has an obvious pulldown pattern, my intention would be to use an AVISynth (Telecide and Decimate from memory) to remove the redundant frames, and recover the 23.976fps master before encoding this through CCE.

[neuron2]

If I have a 29.97fps interlaced master, which has an obvious pulldown pattern That's like saying I have a black cat, which has white fur. The correct way to describe it is "hard telecine".

my intention would be to use an AVISynth (Telecide and Decimate from memory) to remove the redundant frames, Telecine duplicates fields, not frames.

and recover the 23.976fps master before encoding this through CCE. Assuming you have hard telecining, then yes, your intended process is fine. The only thing to be aware of is that you'll need to disable DVD compliance when you encode in CCE, because the encode will be non-compliant for DVD until you apply pulldown in the next step.

[ettin]

Im just curious as to why, if you have a 24fps source, would you telecine it to 30, then IVTC it back to 24, then re-telecine it to 30? Is there a specific reason for doing this?

[neuron2]

Apparently he wants to edit/filter it.

[PhillipWyllie]

Thanks Philip,

So if I have an 29.97fps interlaced video file, my best bet would be to use an AVISynth Script to IVTC it back to 23.976, feed this to CCE-SP to create a 23.976 MPEG-2 stream, which I then feed to Neuron2's pulldown tool to get a 29.97fps stream to author with?

DVD-Boy

What i meant was to get the original unedited source which is at ~23.976. If your client will only give you 29.97 fps video you don't need to bother with pulldown.

[dvdboy]

Thanks Neuron / Phillip,

As you can tell, this is not something I deal with a lot, and so although I know what I'm talking about in my own head, I'm obviously getting the terminology completely wrong when describing it to you guys.

It was not possible for me to get the film as a file at 23.976fps - The film is in the US and I'm in the UK - I've had to receive it on tape and digitize it at my end - my final destination is an NTSC DVD.

I'm assuming that the "hard telecine" shouldn't have any cadence breaks if it was applied at the point the film was recorded to tape.

So if I then use AVISynth to remove the hard telecine, encode the film as a not DVD compliant MPEG-2 file, then run it through pulldown, I will have done "my best" with this encode? Do I take it that CCE-SP does not do a good job at encoding this automatically - does it infact encode a hard telecine into the stream as opposed to using repeat flags?

Again, thankyou for all your help - it seems there is a distinct lack of courses you can go on to learn this stuff, so picking it up as you go can cause so 'errors' in your knowledge. Your patience and help is very much appreciated.

Many Thanks

DVD-Boy

PS: Neuron2, why would DGIndex be reporting the Colorometry as BT.709 - I didn't think I had an ability to set that within CCE-SP?

[neuron2]

I'm assuming that the "hard telecine" shouldn't have any cadence breaks if it was applied at the point the film was recorded to tape. Probably, but it doesn't really matter because the Avisynth IVTC filters adapt to cadence breaks.

So if I then use AVISynth to remove the hard telecine, encode the film as a not DVD compliant MPEG-2 file, then run it through pulldown, I will have done "my best" with this encode? There are too many unknowns to answer that. Maybe some filtering could help. Without seeing the source how can we know? Using soft pulldown will allow you to assign more bits to the actual video instead of to field duplicates, but again, there are too many unknowns for us to know if that theoretical benefit can help you; maybe the video is short enough that you could brute force a high bitrate with hard telecine and still fit it on the DVD. I think generally it is a good idea to encode progressive and use soft telecine where possible.

Do I take it that CCE-SP does not do a good job at encoding this automatically - does it infact encode a hard telecine into the stream as opposed to using repeat flags? Don't know, but I would doubt it.

PS: Neuron2, why would DGIndex be reporting the Colorometry as BT.709 - I didn't think I had an ability to set that within CCE-SP? As stated in the DGIndex user manual for the Colorimetry field of the Info dialog:

"Displays the colorimetry scheme used by the stream. Note that if the stream does not declare the colorimetry, then ITU-R BT.709* is reported for HD video, and ITU-R BT.470-2* is reported for SD video. The * character indicates that the stream did not declare the colorimetry."

So if DGIndex reported BT.709 without the *, then the stream actually claimed to be BT.709. If the * is present, then it is HD video.

[dvdboy]

Probably, but it doesn't really matter because the Avisynth IVTC filters adapt to cadence breaks.

This, I assume is using Decombe rather than the Telecide / decimate method which seems rather arbitrary.

There are too many unknowns to answer that. Maybe some filtering could help. Without seeing the source how can we know?

Sorry, it's a cleanly sourced (HD CAM I believe) film. Running time is in the region of 20 minutes, encoded at 5.5Mbps. Certainly to my eyes it looks really nice. I guess I was confirming that, technically speaking I would have done the "best job" in terms of film content on an NTSC DVD - I just don't want some reviewer to turn around and say "What muppet encoded this???" :)

Don't know, but I would doubt it.

I guess this comes back to my original post - if CCE-SP does a good job at "Inverse 3:2 pulldown" using "Auto 2-pass detection", and produces a correct 'soft telecine' do I need to worry about removing the 'hard telecine' from the AVI file?

As stated in the DGIndex user manual for the Colorimetry field of the Info dialog:

"Displays the colorimetry scheme used by the stream. Note that if the stream does not declare the colorimetry, then ITU-R BT.709* is reported for HD video, and ITU-R BT.470-2* is reported for SD video. The * character indicates that the stream did not declare the colorimetry."

So if DGIndex reported BT.709 without the *, then the stream actually claimed to be BT.709. If the * is present, then it is HD video.

Firstly, my apologies for not reading the manual - n00b error 1!

Looks like N00b error 2 is that the it was version 1.49 that was reporting this information (BT.709 without the *). I've just upgraded to 1.52 and it is correctly reporting the stream as BT.470-2

Video is registered as FILM
Frame Type is registered as PROGRESSIVE
Field Repeats seem to be going up nicely :)

Again, thank you for your help, patience and contribution to these forums. If I could buy you a drink, I would :)

[neuron2]

This, I assume is using Decombe rather than the Telecide / decimate method which seems rather arbitrary. Decomb is the name of the package that contains Telecide() and Decimate(), so they are the same thing. What do you mean by "seems rather aribtrary". Telecide() and Decimate() together form a fully adaptive IVTC mechanism.

I guess this comes back to my original post - if CCE-SP does a good job at "Inverse 3:2 pulldown" using "Auto 2-pass detection", and produces a correct 'soft telecine' do I need to worry about removing the 'hard telecine' from the AVI file? CCE may be apply to apply pulldown very well, but not be able to remove pulldown reliably. I don't know specifically about the capabilities of CCE in that regard, but personally I would do it using Avisynth filters (either Decomb or TFM/TDecimate).

[dvdboy]

Decomb is the name of the package that contains Telecide() and Decimate(), so they are the same thing. What do you mean by "seems rather aribtrary". Telecide() and Decimate() together form a fully adaptive IVTC mechanism.

From memory, for example, Decimate (x) removes every 'x' frame from your video, so after you've used Telecide to de-interlace your video, decimate removes the duplicate frame - but how does that get affected if your cadence break changes - wouldn't it be a different frame you wish to 'decimate'.

Knowing that this is your software Neuron2, I assuming I've made a huge assumption about something, or got the complete wrong end of the stick in terms of theory, so please be gentle :o

[neuron2]

Well, if you don't bother to read documentation, then you will be at risk of spouting utter nonsense, as you have done here.

Decimate(5) does NOT remove every 5th frame as you claim. It removes the frame most like its predecessor in every cycle of 5 frames. It finds and removes the duplicate.

If it did what you claim it would be the same thing as SelectEvery(5,0,1,2,3) and would have no reason to exist.

Gentle enough? :)

[dvdboy]

Perfectly Gentle :)

I shall now retire, tail between my legs and read the documentation, although it all sounds very clever.

:thanks:

[FlimsyFeet]

So if I then use AVISynth to remove the hard telecine, encode the film as a not DVD compliant MPEG-2 file, then run it through pulldown, I will have done "my best" with this encode? Do I take it that CCE-SP does not do a good job at encoding this automatically - does it infact encode a hard telecine into the stream as opposed to using repeat flags?
I still don't understand why you would use an external application to add pulldown flags when CCE, at least, later versins anyway, can do it for you. Select a framerate of 23.976, and CHECK "DVD complaint", and it will put the repeat flags in to give you a 29.97 stream (not hard telecined).

[neuron2]

I agree. Perhaps the OP just heard that "CCE doesn't do 3:2 good" and confused removal of hard 3:2 with application of soft 3:2.

[ettin]

Apparently he wants to edit/filter it.

you cant edit and filter at 24fps?

[dvdboy]

I agree. Perhaps the OP just heard that "CCE doesn't do 3:2 good" and confused removal of hard 3:2 with application of soft 3:2.

Still off reading the documentation :)

TBH I hadn't heard anything good or bad with regards to CCE-SP handling of 3:2. It was more of a case that the manual setting of Inverse Telecine looked quite complicated, and I wasn't too confident in how to use it (The manual was a bit head twizzling with regards to A1,A2,B1,B2 etc) and I was concerned with how good a job the Auto-2-Pass detection would do - it seemed too easy.

At present the Auto-pass looks good on a spot check, although it hasn't been properly QC'd as of yet. For piece of mind though I will probably remove the hard telecine with decombe and then use CCE-SP to encode and apply soft telecine to the file. At least with AVISynth I can step through the video and be happy with the results before encoding.

Again, I don't know if CCE-SP auto inverse telecine is better or worse than AVISynth / Decombe, but the support network here is a little better. Neuron2 is far more forgiving of my mis-use of terms and ignorance than the possible English / Japanese language barrier of explaining this to Cinemacraft :)

I just wish there were some proper courses with regards to all of this video theory / tool usage.

DVD-Boy

[SquallMX]

CCE-SP auto inverse telecine is good enough (using 2 pass mode), just make sure that your AVI is Top Field First, the only "bug" that i've found is that sometimes detects low motion 30i has 30p so CCE encodes interlaced video has progressive, my prefered settings:

"Progressive frame pairing: 60"
"Repeat Field Tolerance: 2" (For 24p, 4 for Hybrid material)
"3:2" (2:3 don't work correctly)

Cheers.

[kolak]

Just feed your video into CCE and use Remove pulldown option. CCE does it very well (you can even check it and if it's needed manualy correct it).
CCE will remove pulldown, encode progressive 23.976 source and add pulldown flags to make DVD compliant mpeg.

Andrew

[dvdboy]

Hi Andrew,

How do you go about manually correcting it - The manual settings as described in the manual are the part I originally got lost at - hence my topic on wondering how good automatic is.

Many Thanks

DVD-Boy