8GB 20 minutes of 720p 24fps?

[braw]

A 4GB (AJ-P2C004) card now holds some four minutes of DVCPRO HD. The good news is that not only will an 8GB card hold eight minutes of 720p/60, it will hold 20 minutes of 24p.

So is this an assumption or has it been confirmed we will have no recorded junk frames for 24p?

Source (http://videosystems.com/mag/video_nab_14/index1.html)

[Luis Caffesse]

As far as I know it is still an assumption.
Although I would say that it is a safe assumption to make.
It makes absolutely no sense to record duplicate frames to P2 cards, and seeing as P2 size is a major obstacle for Panasonic to overcome, I would shocked if they didn't have the foresight to do away with the junk frames.

[Drew599]

As I understand it 1080P/24P will have the extra frames in the footage and the 720P/24P will not. Sorry I couldn't quote something but that's what I remember reading.

[reservoir]

It would make sense to me anyway. Write to tape in the 60i format with duplicate frames, but drop any *junk frames* when writing to P2 to save space. That makes sense and is logical!!

~reservoir~

[Barry_Green]

I went on record over at the Cow, guaranteeing that there will be no duplicate frames recorded on P2 when shooting 720p. Now, the guarantee is that I'll fork over $100 to the first person (and only the first person) who proves otherwise. But as an ex-programmer and as a current video guy, I feel very confident in saying: ain't no way they'll waste P2 card space storing redundant, unnecessary "padding" frames! No way. Won't happen.

I do not know this for a fact, but I believe the Panasonic engineers are kind of smart and I feel very confident in saying that there's no way they'll do something as dumb as waste 2/3 of the space on a P2 card by recording duplicate frames! I do predict the duplicate frames will be streamed out the firewire port (assuming that it allows streaming out the firewire port), because I think they will keep the 100-megabit stream intact. But I can't fathom them recording duplicate frames on the P2 card in 720p mode. Maybe in 1080i, but definitely not in 720p.

[Gary_McClurg]

Barry, how bout duplicate frames in 1080p?

[braw]

I thought it would be smart and easy to only store useable frames on the card, and have pulldown directions in the header so it could duplicate on the fly the doubled frames as yuo import it.

Not sure if that can be done with this medium, but it seems easy and keeps the stuff compatible with what's already out there while optimizing storage space.

[Tzedekh]

As far as I know it is still an assumption.
Although I would say that it is a safe assumption to make.
It makes absolutely no sense to record duplicate frames to P2 cards, and seeing as P2 size is a major obstacle for Panasonic to overcome, I would shocked if they didn't have the foresight to do away with the junk frames.
Although she technically didn't say data is recorded to P2 at 40 Mbps, Jan Crittenden has effectively answered this question (here (http://www.creativecow.net/forum/read_post.php?postid=111206508018739&forumid=193&highlight=40%2Cmbps)).

[Barry_Green]

Barry, how bout duplicate frames in 1080p?
Well, that's the wild card. I mean, the very fact that they're doing 1080p means they're inventing something new, because DVCPRO-HD doesn't have any provision for 1080p at all (unlike with 720p, where the support for variable frame rates was built in at the beginning).

So, I'm certain they'll use a pulldown scheme to implement 24p within a 60i stream. If it uses 2:3 pulldown, you *cannot* drop duplicate frames, because there *are* no duplicate frames.

If instead it uses 2:3:3:2 pulldown, then yes, there's one frame that could be dropped, but technically it is *not* a duplicate frame; it's a frame that's made up of one field from the preceding frame and one from the following frame. So in 2:3:3:2 there are technically no duplicate frames at all, so will it drop it? I don't know. There is an *unnecessary* frame, yes, but not a duplicate, so the technology is a little more complex.

But if I was programming it, I'd find a way to just not record that frame. However, for compatibility with the 100mbps data stream, there may be reasons why they can't do it. I don't know what, but there may be some. However, it's not that big of a deal; in 720/24p you get 150% more recording time by dropping duplicate frames (2.5 times as much). In 1080/24p, dropping the "c" frame, you'd get 25% more recording time. It'd be nice to have that 25%, but it's not nearly as significant as the difference for 720/24p.

[thisiswells]

Thing is, we don't yet even know if the camera will have a variable-scan system
besides 24p, 30p, and 60p. The question is what happens between 30p and 60p?

An easy way to get Barry's money is to find out how 24p will be implemented in the
SPX800 camera. It is doing 24p in a P2 card and while there isn't a provision for 24p
without a 30p stream in PRO50, thus requiring a pulldown in post... neither is there a
provision for 720p24 without doing a pulldown from 720p60, thus the only way of doing
24p in *any* Panasonic product is to do a pulldown process in post from the original
footage. Of course, new tech changes things, and the SPX800 is the new tech.
So, find out how 24p will be implemented there and you've got your answer.

: )

[thisiswells]

My prediction is there will be two recording formats in P2/720p:

50Mbits format up to 30p stream
100Mbits format for up to 60p stream

That way, the editing software only has to include a single new version of DVCPRO-HD Codec.
And I shall call it, drum roll please. . . . . DVCPRO-HD50.

Under this definition, the DVCPRO-HD50 format will *have to pad* 24p into 30p.
I will gladly accept your offer. Barry, you're on. In a friendly way, of course.
: )

[thisiswells]

Brian’s Predictions on the suspected DVCPRO-HD50 video recording system

720p30 achieves alternate framerates using pulldown
Four uncompressed monophonic audio tracks, down from eight in DVCPRO-HD

= 50Mbits/Second High-Def recording system

NOTE: DVCPRO-HD50 IS NOT REAL. IT IS NOT BASED IN FACT. IT IS A GUESS.
PLEASE, DO NOT DISCUSS THIS *GUESS* IN ANY WAY OUTSIDE OF THIS EXACT THREAD.

[satellitebunny]

I'm kind of skeptical about the camera recording only 24 frames in 720p. If it would do so, it would be easy for the camera to pad on the fly to get the full 60p 100Mbit/s stream out of the firewire. But, once you plug the P2 card on to your laptop, then your NLE should be aware of this non-standard DVCPRO-HD 24p non-padded stream. And that would require some more updates to the NLEs.
I'm not saying it couldn't happen, but I'm just guessing that it might be awkward to break the 100 Mbit/s standard. But, the thing thisiswells is suggesting about the new 50 Mbit/s DVCPRO-HD codec variant, that might be just the way to implement it, but I wouldn't bet on it.

I hope you all are right, and my guess is wrong.

The 24p 1080p is propably going to be 80 Mbit/s with pulldown on a 60i stream and the 1080/30p and the "PAL" 1080/25p is propably full 100Mbit/s without any kind of unnecessary data. So, again it would make the 25p "PAL" framerate a little better than it's "NTSC" counter part. But only by a hair. (Unless you don't want to slow down the 25p to 24p. I really see no reason to be using the PAL version in an NTSC-land this time though.)

[Barry_Green]

Ah, but -- and someone correct me if I'm wrong, but I don't think I am -- Panasonic/Apple released a DVD with some sample DVCPRO-HD footage on it, when FCP-HD was introduced. And, on that disc, there were 720/24p files, and they were encoded as 40-megabit with no pulldown frames!

Which means the editors already know how to handle it.

No chance for a new limited-audio 50-megabit format. Nope. It's gonna be full DVCPRO-HD, with the only difference being whether inactive/"pad" frames are recorded, which I'm saying they won't be...

[thisiswells]

After sitting on this, I am even more convinced.

Let's examine completely variable frame rates on a camera. You can do anything between
time lapse and 60Fps. So, you record everything one-to-one on a frame-by-frame basis. If
the camera is set to 24Fps, it only records 24Fps. Well, this could easily be at least 56 variations
of the same Codec, assuming the limit is 4Fps to 60Fps. That's not a big deal. So, you only
record discrete frames... But, think about how you would play them back? Currently there is
specialised hardware (Panasonic FRC27) or software (FRC in FCP4HD) that will perform this.

If the camera is totally variable frame rate, then you would have to instruct it on how to
do playback. Here is why. Filmmakers would want to record at 60Fps and playback at 24Fps.
Video shooters for TV would want to playback at 60Fps. This becomes really complicated if
you wanted to shoot 40Fps. But, wanted to playback at 24Fps. Or, record at 27Fps and
playbackat 27Fps instead of 30Fps or 24Fps. Point is: Where do you draw the line between
reasonable and too much? I can't imagine ever having to record 27Fps and have the choice between
different playback rates, but really for any of this to work it would have to be there
as an option, at least. I don't think it's reasonable to build a camera and software to encode individual,
discrete frames and set a variable playback framerate on the fly.
I just don't see it happening.

So, let's look at other another option.
The reasons to record at 60p are (1) "that reality look" (2) slow motion
The discussion of not recording pad-frames for 24p or 30p in a 60p stream is relevant.
Why waste 2.5 times the bandwidth when it simply isn't necessary? It isn't. You won't.
That's why I think there is going to be a separate, reduced bandwidth 30p version.
It sets a necessary restriction on the recording and playback systems for better integration.
Really, if you *could* do any frame rate, this would be very complicated to edit without a
standard. You would have to tell your system how to play it back. Presuming the camera
would record framerate in metabata, the editor would still need to be instructed for playback speed.
Could you batch tell it that "all these one's are 24p, so make em' 24p" or "all these
ones are 27p" ? I don't think that is reasonable. It's too much. It's too congested. Too many choices.
It's too confusing to explain. For these reasons, I think there will be a 2nd version of the DVCPRO-HD codec
at 50Mbits... It creates a standard playback speed.

The way I see it, most will want to record at 24p or 30p most of the time -and- for that, they
will provide us with a second 50Mbit version of DVCPRO-HD that only records up to 30p.
Anytime we would want to go into 60p, then that would be a full bandwidth 100Mbit.
So, yeah -- it's a small compromise to accept a 6-frame padding to do 24p within 30p,
but it isn't ridiculous like doing 24p in 60p, either. It is reasonable and it is a standard.

thoughts?

[thisiswells]

I wish to add that if I am dead wrong, I would like to see a really good implementation
of doing variable frame rate playback in camera at necessary playback speeds and a
reasonable post-production workflow. Actually, I really hope that I am wrong.

FWIW, I bet there will not be any framerates besides 24, 30, and 60.
This is of course, fine as my desire/need for 32,40,48 is fairly seldom. 60p is cool enough.
And, 60p is probably there for the reality and sports, anyways.

brian wells

[thisiswells]

No chance for a new limited-audio 50-megabit format. Nope. It's gonna be full DVCPRO-HD.

I hope so. If that is the case then 24p would have four stereo tracks embedded, then.
Please, correct/edit this post if it is incorrect. Stereo tracks on DV and HD100 are to
be recorded in a 6mbps dedicated audio portion on tape. This is uncompressed 16/48.
Well, wouldn't that indicate 6mbpsx4 Stereo tracks=24mbps? I think so. This is the
"missing variable" in the logical deduction of where we came up with a 24p/40mbps
recording. If it is still going to be the full 8-mono or 4-stereo tracks, then that only
leaves 14mbps for video if only 40mbps is recorded. That is never going to happen.

It may be more accurate to say that 4Fps could be 30mbps and that 24Fps could be
more like "under 60mbps" (too tired to work this one out) but it isn't anywhere
close to 40mbits as we had previously thought... Or, reduce the audio tracks in the
recorded format, and save space. LMK what you all think. I'm going to sleep now.

brian wells

[Barry_Green]

I can't see any logic to what you're saying. Reason being -- this camera isn't a tape-based camera, it's more likely to be a "camputer", as they call the SPX800. Think about Windows Media Player -- you can encode a file in .wmv format to play back at 12fps, or 24fps, or 30fps, or whatever you want. The computer just plays it back at whatever rate the file tells it to. I'm sure the HVX will be just as "aware". And, again, re-examine what I said earlier about the demo DVD for FCP-HD: I'm certain there are 40mbps 720/24p clips on that disk, which means the computer already knows what to do with it.

As for playing back at different frame rates, you can do that today with editing software. I could import a 30P .AVI into Vegas, and tell it to play back at 80% speed, and it'll lay down frame-accurate slow-mo video into my 24p timeline. I'm sure you could do the same thing (with the proper percentages applied) to your 27p footage (90% speed, presumably) or with whatever frame rate you wanted.

The only restriction on playback rates is for analog or HDMI output... there the system will have to conform to some sort of ATSC standard in order for the playback to be viewable on an ATSC-compatible display. And the solution is simple; it just pads the playback rate. If the file is encoded with a header that says "24p playback", then the camera will know to do 3:2 pull-up on playback.

This really seems like a very simple thing to implement, and IMHO doesn't necessitate any sort of need for inventing some new 50mbps variant. The files are computer files. The camera is a computer (or a "camputer"). It's not bound by tape and tape's restrictions. It can play back files, at whatever the appropriate frame rate is, just like quicktime or windows media files.

[Barry_Green]

I hope so. If that is the case then 24p would have four stereo tracks embedded, then. Please, correct/edit this post if it is incorrect. Stereo tracks on DV and HD100 are to
be recorded in a 6mbps dedicated audio portion on tape. This is uncompressed 16/48.
Well, wouldn't that indicate 6mbpsx4 Stereo tracks=24mbps? I think so. This is the
"missing variable" in the logical deduction of where we came up with a 24p/40mbps
recordding. If it is still going to be the full 8-mono or 4-stereo tracks, then that only
leaves 14mbps for video if only 40mbps is recorded. That is never going to happen.

The audio portion of DVCPRO-HD is 8 tracks of 16-bit 48khz sampling. All told, all together, they occupy about 6 megabits of the 100-megabit data stream. That's not 6 megabits per stereo pair, it's 1.5 megabits per stereo pair. So 4 pairs, i.e. 8 uncompressed tracks, = 6 megabits, not 24.

[thisiswells]

Well, then that's really good news. Cool.

[Aaron Koolen]

The audio portion of DVCPRO-HD is 8 tracks of 16-bit 48khz sampling. All told, all together, they occupy about 6 megabits of the 100-megabit data stream. That's not 6 megabits per stereo pair, it's 1.5 megabits per stereo pair. So 4 pairs, i.e. 8 uncompressed tracks, = 6 megabits, not 24.

Shit, 8 tracks? How are they going to fit all those XLR plugs on it then? :)

Aaron

[satellitebunny]

Ah, but -- and someone correct me if I'm wrong, but I don't think I am -- Panasonic/Apple released a DVD with some sample DVCPRO-HD footage on it, when FCP-HD was introduced. And, on that disc, there were 720/24p files, and they were encoded as 40-megabit with no pulldown frames!

Which means the editors already know how to handle it.



If this info is correct, then I'll have to agree with you, that it will propably be recording 720/24p without the padded frames. That sounds great, as you'll get much more minutes on a P2 card that way.

Another question that remains unanswered is: What about external harddisk/firestore type recording? Will it still be without the padded frames, or will it be the full 100 Mbit/s? This we propably won't know until NAB, or not even then...

[Barry Goyette]

I don't know if this clears this one up or not...

From Jan Crittenden Livingston (on dvinfo.net)

"While I cannot disscuss how this camera works, I can discuss the concept of the data remaining in the DVCPRO HD signal once the extra frames have been removed. DVCPRO HD has 100Mbps at 60 frames of information. The compression on each frame is equal and the same, d optimized for the compromise of intraframe compression. If we look at the fact that there is 60 frames per second and there is 100Mbps, we have a formula. If we only use 24 of those frames, then we are using a little over 1/3 of the data or about 40Mbps. When working with FCP over firewire, you can reduce the data load from 100Mbps to 40Mbps just by casting off the exta frames."

[thisiswells]

I think they will integrate a "final cut pro-like" frame remover as it saves data to P2, then
re-instert duplicate frames upon playback to NTSC, just like Barry said.

[Barry_Green]

Shit, 8 tracks? How are they going to fit all those XLR plugs on it then? :)
The format supports 8 tracks of uncompressed audio, yes. And DVCPRO50 supports four tracks of uncompressed audio.

Now, whether we'll have four (or eight!) connectors remains to be seen. I'm *really* hoping for four... I think that would be the bomb. Even if it's something like an external box that connects through a multipin connector... man, that would be sweet!

Not counting on it though... the VariCam and the SDX have just two XLR ins (plus the microphone XLR in)...

[Gary_McClurg]

Barry, always the man with the answers.

I have no clue if this is possible, but could some one make an external box like your saying and run a mixer through firewire so you could have each channel with a different mic.

[Luis Caffesse]

I have no clue if this is possible, but could some one make an external box like your saying and run a mixer through firewire so you could have each channel with a different mic.


Intriguing idea, but I would doubt it.
That would mean inputing something through the firewire while shooting, at which point the camera is usually streaming the signal out of the firewire port.

I don't think you can input information through the firewire port on any camera while in 'camera' mode...it seems to only take in data while in 'vtr' mode.
(then again, so far the HVX doesn't seem to be like most other cameras).

[thisiswells]

Those tracks are really there for post production and surround sound mastering.
With that said--yeah there is an interface for the big cameras, but they aren't eight tracks.

Referencing Wolf Seeburg's book, 24p: Sound and Video Assist" Page30. The Sony CineAlta
will take an additional two tracks (20bit at 46K, yes 46K) for a total of four inputs using the
HDCA-901 adapter with two XLR inputs. The information is less clear for VariCam, as Wolf
has not used VariCam as much (or chose to dedicate his book to Sony 90% and Panny 5%)
because it isn't really clear what Panasonic offers. It hints at the idea of inputs 3+4 using
AES, but as I have read it, it just isn't real clear.

Regardless, it seems like the full audio capabilities are recognised in fully capable VTR's with
AES inputs or possibly individual analog XLR inputs. There really isn't a field camera that
uses all eight tracks, even with the existing breakout boxes.

brian wells

[thisiswells]

It is fair to note Wolf Seeburg is the "dual system audio" advocate in Hollywood...
So his assessment of the audio capabilities on an HD camera may be skewed a tad.