"24p Pulldown Removal" or "How do I know which tape is 24fps?"

[ChrisKron]

Good day to you,

I wonder if you could help me with my little problem here.

So, my old Canon HV20 Camcorder (1080 60i) has a 24p-mode which internally records 24fps, but turns them into 29.97fps and also gives out 29.97fps when transferred to the PC. So all videos I get with "HDVSplit" are 29.97fps, no matter if actually having been recorded in 24fps.
It all happens through a so called 3:2 or 2:3 pulldown. When you pause the video you see 3 full frames, then 2 interlaced (or blurry images).

I used to use a software from "Cineform" which had an option "3:2 pulldown removal" and transferred my footage with 24fps-progressive to my PC.
When I use "HDVSplit" I always get 29.97fps (as stated above), also the video is interlaced.

What would be the correct way to get TRUE 24P? And more importantly, how do I know which of my MiniDV-tapes have been recorded in the 24p mode? There is no apparant indication to my knowledge...but there must be a way. It was THE selling point of the HV20 back in the days. How did professionals work with this camera if you don't know what framerate you filmed at on a specific tape?

Is there a software that I need to use to a pulldown removal option while transferring a tape, or do I have to transfer my tapes to 29.97fps and then afterwards do the pulldown removing? But the most important issue remains: How do I know which tape is 24p and which is 29.97fps?

[jagabo]

DV camcorders usually used 2:3:3:2 pulldown to create 29.97 fps interlaced video from 24p. If you look at that in an editor without deinterlacing you will see a repeating pattern of 4 progressive frames (no combing) followed by one interlaced frame (with combing). To reduce that to 24p you want to discard the interlaced frame from each group of 5 frames.

[JN-]

Apply IVTC to get it back to 24p after capturing to computer. On page 38 of the manual it says it can record to 24p also. I’d imagine you would have to just do the firewire capture first then apply inverse telecine, or, perhaps playing the tape in the camcorder first and then maybe the frame rate can be displayed on camcorder screen with other info ?

If you search IVTC here you will get good results, handbrake etc. I also made an IVTC util, batch file, community user poisondeathray here helped me immensely.

[jagabo]

Yes, the conversion from 24p to 29.97i (59.94 fields per second for display on a TV, packaged as 29.97 interlaced frames per second in digital form) for TV is called a Telecine. Reversing that process is called an Inverse Telecine.

[Bwaak]

DV camcorders usually used 2:3:3:2 pulldown to create 29.97 fps interlaced video from 24p.

I don't know what do you mean by "usually", but Panasonic, for example, calls 2:3 pattern "24p Standard", while 2:3:3:2 is "24p Advanced".

The HV20 uses 2:3 without flags, so the software that relies on flags sees it as 29.97i.

The HV40 has two modes: 24F and 24P, one is 2:3, another is handled by most software as native 24 fps, but I guess it is the same 2:3 with flags.

[Sharc]

The HV40 has two modes: 24F and 24P, one is 2:3, another is handled by most software as native 24 fps, but I guess it is the same 2:3 with flags.

AFAIK 24P and 24PA are synonyms for pulldown, 24P for 2:3 and 24PA for 2:3:3:2 aka "Advanced Pulldown". I don't know what 24F stands for. Maybe it's 24Frame in interlaced PsF (2:2 pulldown) format?. The acronyms are often brand specific and not very clear.
24PA aka 2:3:3:2 pulldown is "usual" for DV (SD DV), while 2:3 pulldown is generally more common outside of SD DV (film broadcast in NTSC, film on NTSC DVD) as it minimizes judder.


[Attachment 68567 - Click to enlarge]

Edit:
To me it is not clear how the HV20 stores the HD "24fps" on the tape. Is it really 24fps progressive frames, or is it always stored as 59.94 fields per second using either 2:3 (24P) or 2:3:3:2 (24PA) pulldown, even hard telecined. Or is the pulldown applied by the playback software only? But does it really matter? Applying IVTC on telecined material will return the 24fps progressive frames.

Edit2:
Manual of the HV20 page 101 (Footnote 1):

[Attachment 68568 - Click to enlarge]
So it seems that for the HV20 "24fps" is always stored as telecined 59.94 fields per second (=60i in Canon terminology) according to the manual. Therefore the HDVsplit will transfer it accordingly, and the InverseTelecine to return the 24 fps progressive frames has to be done in post.
The only purpose of the "24fps" setting is to give it a film look.

Edit 3:
Maybe the HV40 supports 2:2 pulldown with its "24F" (F for FrameMode) setting, means interlaced with both fields from the same instant in time, also called PsF (Progressive segmented Frame). It's just my guessing though.

[JN-]

Yes, good spot, page 101. I was thinking that maybe anything stored on tape is probably only in fields, not likely to be Progressive. So if the Camcorder displays frame rate as 60i in playback mode then it can be IVTC'D to 24p.

Looking at page 57, on-screen displays, I don’t think it displays the frame rate.

[Sharc]

And here the difference to the HV40 model (@Bwaak, post#5) which can apparently store 24fps progressive video with its additional "24F" setting (Manual page 106):


[Attachment 68572 - Click to enlarge]
(I would have assumed that it is stored as PsF rather than native 24p, but this doesn't really matter.)

I think this should clarify the OP's question. The HV20 does not have this setting.

[jagabo]

Whether the file is 2:3 or 2:3:3:2 pulldown the procedure is the same. Field match and decimate. In AvISynth it's

AviSource("filename.avi")
TFM()
TDecimate()

[JN-]

"how do I know which of my MiniDV-tapes have been recorded in the 24p mode? " Maybe shoot and capture two small test clips with and without the 24p in camera mode setting. Then use the "Mediainfo" util to compare/inspect the two files.

[Sharc]

Watch a scene with motion on the PC, using a player without deinterlacing:
- Those recorded in "24 fps mode" will show the typical telecine pattern, means (2 combed + 3 clean) for 2:3 pulldown, or (1combed + 4clean) for 2:3:3:2 pulldown.
- Those recorded in normal 1080i video mode will have all frames combed.

[Bwaak]

I think that the question whether file-based video is native interlaced or native progressive is moot - it may be coded as fields or frames yet the flags would indicate to treat fields as frames at one point or as fields at another, building a certain sequence. But when recorded on tape, I think it physically is recorded as interlaced, after all one track is one field. No, I take it back, with DV/HDV one frame is 10 tracks for 60 Hz, and 12 tracks for 50 Hz. Anyway, if it were recorded as 24 fps native, then it would use lower bitrate and the tape would fit more footage. With DV/HDV bitrate and duration are the same no matter the chosen frame rate, so I think it is still interlaced on tape, but with flags, and when it is sent over Firewire it is converted to native progressive. Or not, as MediaInfo actually reports 2:3 pulldown for 24F footage from the HV40. All this is academic anyway, as long as the editor/converter can handle it as native 24p.

But the HV20 does not write flags, and it is not 2:3:3:2, it is 2:3 only AFAIK, because it is a consumer camcorder after all.

[JN-]

“I don't know what 24F stands for.” I believe it means that its recorded directly to tape as 24 frames.

[Sharc]

Yes, according to the manual of the HV40 page 106 (see post#8).
The HV20 doesn't have this option though.

[JN-]

Yes, according to the manual of the HV40 page 106 (see post#8).
The HV20 doesn't have this option though.

I missed that post, thanks.