DV Info Net - Understanding 24p

DV Info Net (https://www.dvinfo.net/forum/)

- Techniques for Independent Production (https://www.dvinfo.net/forum/techniques-independent-production/)

- - Understanding 24p (https://www.dvinfo.net/forum/techniques-independent-production/108000-understanding-24p.html)

Understanding 24p

This is a continuation of a discussion we were having in another thread pertaining to the Sony Z7 and S270. I'll throw in some quotes to bring everyone up to speed.

Quote:

It's like the V1u, DVX100, HVX200, etc.: it can record in both interlace and progressive. Keep in mind, with the exception of the HVX200, in the sub-$15,000 category, all 24p cameras are packaging that 24p signal in a 60i stream via the 2:3 pulldown.

The HVX200 will record in 720p 24pn (native), no pulldown. I did a short film recently in that setting on the HPX500.

Quote:

Is the HDV on the card being recorded as full progressive frames?

Lets say you're recording 1080P to both card and tape at the same time (at 24 or 25 fps).. currently, recording HDV to tape in this manner will mean it'll be done with 3:2 pulldown (in a 60i or 50i stream). No surprise - same deal over HD-SDI.

However, if it's recording to flash cards as well at real time, well, would it/could it record as true progressive frames (a la HVX-200) so it wouldn't require that extra step of removing the pulldown once on the computer?

Quote:

I'm not saying that's bad, but other than the fact that the HVX200 is putting the video on a P2 card, the process it uses to achieve 1080 24p is not unique among cameras under the 15k price tag that the writer made reference to. In fact, it seems very similar to the JVC HD series process.

Being we are all trying to learn the facts, the little details can be important. And as I say that, if one of the HVX200 wizards has a better explanation, or can explain where I am misguided I would welcome that info. I really admire this camera and want to understand all of this stuff.

If the HVX200 records 1080 24p at full resolution, to tape or P2, I will buy one. But so far, I haven't found that to be the case.

Quote:

There's no quality difference between 1080/24PA and (if it were to exist) 1080/24P native. It's just a matter of removing some extra frames.
Applications like DVFilm Maker, RayMaker, and plugins like Raylight for Vegas can remove the pulldown without any recompression or loss of quality, they simply skip over the redundant frames.

Quote:

The only thing I was pointing out, is that the HVX200 performs pulldown to achieve 1080 24p... not that the image quality suffers because of it. It is simply the process that it uses to produce 1080 24p.

This new camera from Sony, appears to operate in a very similar way to the HVX200.

That's all.

Quote:

But that's not an accurate statement. Rather it acquires 1080 at "true" 24p and in order to make it compatible with older editing systems, it then adds pulldown to convert it to 60i. But removing the pulldown restores the original 24P.

Quote:

Nevermind... I see it. It starts with 1080 60p Native. All other streams are derived from that. Final output depends on medium and compression...

Quote:

Another untrue statement. Unfortunately Panasonic is guilty on this one too so I understand everyone's confusion. When the camera is shooting 24P it really is sampling the sensor at 24P. A simple experiment proves this is true. At 24P with the shutter speed set high (around 1/250 for example) you pan over an object rapidly, and then look at the results frame by frame. If it was converting 60P to 24P you would see either blended frames or slightly uneven steps. You don't. It's perfectly even and clean, because it's sampling at 24P just as it should.

So let's continue this shall we?

Marcus,
I was holding off asking this question in the other forum because I didn't want to hijack the Sony thread with some Panasonic discussion... so I'll ask it here.

Am I correct in understanding that you're saying the HVX is adding some kinda pulldown pattern to the 24pn modes? Explain to me why they would do this when this 24pn mode is supposed to be JUST 24 frames on a card. They include the ability to shoot in a 24pa mode, so why the need to muddy the waters with a 24pn mode if it doesn't actually shoot only 24 frames every second without some type of pulldown.

No pulldown in the 24p native mode, that I'm aware of. I never had to remove a pulldown.

However, and this will throw you WAY off, but if you take Canon HDV 24F footage, capture it using Final Cut Pro's HDV 1080p24 codec, it will indeed remove the pulldown upon capture. I couldn't believe it when I tried it, but it did it!

And with FCP 6.0.2's release, apparently, something is addressed with V1u 24p footage, and I think JVC, too. Someone in the know mentioned earlier this fall that an update will come out to address some JVC/Final Cut Pro issues.

Heath

Quote:

Originally Posted by Heath McKnight (Post 775706)

No pulldown in the 24p native mode, that I'm aware of. I never had to remove a pulldown.

So why then would Marcus say the following (talking about the HVX I had assumed)

Quote:

There's no quality difference between 1080/24PA and (if it were to exist) 1080/24P native

and

Quote:

Marcus?

24pa means advanced, where special "flags" are put up to help with the removal of pulldowns. It's VERY deep geek. Read this:

http://adamwilt.com/24p/index.html

heath

ps-In particular, this:

http://adamwilt.com/24p/index.html#When_to_shoot

I'll email some of the guys about our new discussion thread.

Heath

Quote:

Originally Posted by Heath McKnight (Post 775706)

However, and this will throw you WAY off, but if you take Canon HDV 24F footage, capture it using Final Cut Pro's HDV 1080p24 codec, it will indeed remove the pulldown upon capture. I couldn't believe it when I tried it, but it did it!

Heath, not to stray off topic, but this doesn't work for Canon's HV20 footage shot in the 24P mode now, does it?

Quote:

Originally Posted by Ethan Cooper (Post 775697)

No

Pulldown is added whenever you shoot 24 but the camera is recording
more frames than that for compatibility purposes like for 1080i60/24PA
for example.

The real issue is how the 24P is created to begin with. The answer is:
the sensor is sampled at 24 (actually 23.976) fps.

I haven't tried it with Canon's 24p stuff. But I'd like to try!

heath

Quote:

Originally Posted by Marcus van Bavel (Post 775681)

Another untrue statement. Unfortunately Panasonic is guilty on this one too so I understand everyone's confusion. When the camera is shooting 24P it really is sampling the sensor at 24P. A simple experiment proves this is true. At 24P with the shutter speed set high (around 1/250 for example) you pan over an object rapidly, and then look at the results frame by frame. If it was converting 60P to 24P you would see either blended frames or slightly uneven steps. You don't. It's perfectly even and clean, because it's sampling at 24P just as it should.

Calling that an untrue statement is not correcting me... it is correcting Panasonic. That is exactly how they state in their literature that it works. (The HVX200 brochure). Their document repeatedly suggests that everything is captured at 1080 60p, but they don't record 1080 60p, and all other streams are derived from that. Not MY words, Panasonic's.

I don't dispute the results of your work or your expertise for that matter, but this is why it gets confusing trying to understand this technology.

And I would also like to clarify for all, it is not my intention to argue with Heath, Marcus, or anyone else. I am not an expert, but the Panasonic literature indicates just what I've posted. I don't have the expertise like Marcus and others to perform tests that disproves or proves anything. I am like so many others here, just trying to understand and apply all of this info into my needs. So I will recap what I have learned thus far:

The HVX200, (according to Panasonic specs and info), Captures at 1080 60p, and all other streams are derived from that. (HVX200 brochure page 4)

It writes to tape using a 1080 60i stream with 2:3 pulldown.(HVX200 Brochure pg 5, note 2)

As to the importance of how the original 24p is derived from a native progressive source, I am enlightened. I had always believed the 24p was derived from the 1080 60i to begin with. It is only "packaged" in a 60i stream to support the limitations of writing to MiniDv and the P2 card.

Thanks for all of the input, and explanation.

HVX's DVCPRO HD doesn't go to tape; there's a tape transport system for DV/SD only. They went with P2 because it's cheaper to use those cards than a DVCPRO HD tape transport system, which would quadruple the cost of the camera.

heath

Quote:

Originally Posted by Heath McKnight (Post 775965)

That's right. Typing too fast. It is all as clear as crystal. Thanks.

From what I gather, there's a difference between how the Canon XHA1 records 24fps to tape and how the Canon HV20 records 24fps to tape. Someone care to break the difference down?

The HV20's cadence can break on each clip, and it doesn't record any flags, so the only way to extract the frames is by analyzing the clips. Cineform's tools do this in realtime. I'm not aware of a realtime process for Macs, unless this new FCP update can do it. Waiting anxiously to hear definitively!

Hey, Cineform's now on Intel Macs! So that'll be something to check out.

Heath

Quote:

Originally Posted by Heath McKnight (Post 776632)

Hey, Cineform's now on Intel Macs! So that'll be something to check out.

Heath

The CODEC is available on Intel Macs.

Pulldown is removed in the CineForm Capture Utility. I don't think they are shipping this yet.

-----------------------

The HV20 uses the same 2-3 pulldown as the Sony V1 in the "24" mode. No 2-3 MPEG-2 flags, just a 2-3 cadence. The A frame need not be the first frame. For this reason, I suspect FCP 6.0.2 will not work with "24P" mode.

When shooting 24fps with the V1 you should/must shoot using "24A mode.

"A" mode does NOT mean ADVANCED. It has nothing to with the HVX200's ADVANCED mode. It means "A frame." The first frame of every clip is always an A frame. This means the cadence need not be anaylized. So it EZ to find the 2 judder/split frames.

6.0.2 checks the MPEG-2 data stream. If the MPEG-2 flag indicates the stream is INTERLACE, capture proceeds as 1080i60. If the MPEG-2 flag indicates the stream is PROGRESSIVE, capture converts the two judder frames to two duplicate frames. Thus, every 5 frames there will be a duplicate frame. The file remains 1080i60.

Once in a 1080p24 Sequence, ONLY the 24p frames can be seen.

Canon's 24F mode is a 3rd way of recording 24fps in 60i. The 24 frames are MPEG-2 flagged in a 2-3 cadence. So the recording can be 60i. FCP reads the MPEG-2 flags and puts just the 24 frames into a file.

XDCAM HD and the EX1 in HQ mode do the same -- I think.

JVC uses a 4th way -- Repeat flags that create a 2-3 cadence in a 60p data stream.

The Cineform codec is in beta for Intel Macs, but not their applications. It's their apps that do the pulldown. At least that's my understanding having not actually used them.