View Full Version : Data rate of DVCPRO50 at 24p?


Edwin Hernandez
November 5th, 2005, 09:42 AM
If DVCPROHD @ 720p, 24p is 40 Mbs, what's the data rate of DVCPRO50 @ 24p?

If most of my stuff is gonna be aired in the short term in SD (TV spots and corporate video), does it make sense to shoot DVCPRO50 @ 24p, instead of DVCPROHD? How long will I be able to capture on each 4GB P2 cards at that rate?

-EDWIN

Jim Exton
November 5th, 2005, 01:11 PM
DVC Pro 50 is 50 Megabits per second.

A 4 GB P2 will store 8 minutes.

For SD only work, yes, I would recommend DVC50.

But I think that is going to be a decision you are going to have to make based on your budget, etc.

Edwin Hernandez
November 7th, 2005, 08:49 AM
Jim: Pardon my ignorance, but I still don't understand why DVCPROHD 720p at 24p has less data rate than DVCPRO50 at 24p?

Barlow Elton
November 7th, 2005, 10:31 PM
Jim: Pardon my ignorance, but I still don't understand why DVCPROHD 720p at 24p has less data rate than DVCPRO50 at 24p?

The data rates are the same, but it has much more compression per frame. Correct me if I'm wrong here, but DVCPRO50 24p is less compression per frame (3:3:1 NTSC) than DVCPRO HD 720p 24p (6:7:1), but at a much smaller pixel (720x480 vs. 960x720) resolution. Technically, the bit rates per frame are similar (talking "active" frames here, i.e. the non-pulldown 24fps) but the compression ratios are different. There simply is less "data" at 40 megabits for 24 720p frames then there is for 24 480p DVCPRO50 frames.

It's really just semantics. "Data rate" can mean different things in video, depending on the format, compression scheme, resolution etc.

Thomas Smet
November 8th, 2005, 09:37 AM
480i and 1080i modes are interlaced so 24p video shot in these modes must use a 3:2 pulldown truning them into a 29.97 video. Even though the in between frames a field duplicates it still sees them as 30 frames so the datarate is exactly the same.

720p is native progressive so when it puts 24p into a 60p stream it just duplicates frames to make it fit. On tape it just exactly the same as interlaced video. No matter what all frame rates of 720p took up the same amount of space on tape. With P2 however they don't need to conform 24p to 60p so the card actually only records the 24p.

I guess it is much easier to duplicate progressive frames on the fly during output than it is to add 3:2 pulldown.

Luis Caffesse
November 8th, 2005, 02:59 PM
This thread seems to be getting confusing.

Just to be clear, datarate does not mean many different things, it means exactly what it sounds like: the amount of data recorded in a given amount of time. Granted, many things can affect the datarate, but that doesn't change what 'datarate' means.

It is true the the compression is essentially twice as much in DVCProHD as it is for DVCPro50.

There is also another key difference between the codecs (which was already touched on).

DVCPro50 is a 30fps format
720 DVCProHD is a 60fps format.

DVCPro50 has a datarate of 50Mb/s
DVCProHD has a datarate of 100Mb/s

That means that for 1 second of DVCPro50 you have 50Mb of data.
That 1 second of footage = 30 frames.

But in 720 DVCProHD, 1 second = 60 frames.


So if you do the math you'll see that each frame of footage = 1.667 Mb
(60 x 1.667 = 100Mb/s) 720 DVCProHD
(30 x 1.667 = 50Mb/s) DVCPro50


BUT, and here is the part where I think the confusion comes in, 720 DVCProHD is a natively progressive format. There is no such thing as interlaced 720 DVCProHD.

On the other hand, DVCPro50 is a natively interlaced format. Everytime DVCPro50 is recorded it must be interlaced, so when you shoot DVCPro50 at 24fps the signal has to have a pulldown added to make the video conform with the 29.97 NTSC standard. This works exactly the same the DVX100 does now (as Thomas explained above).

So, because of the pulldown which is added, you are always recording 30fps when recording in DVCPro50, which means you are always using the full 50mb/s datarate.

But, when recording 24 frames in 720 mode onto P2 cards the camera only records the 24 frames per second that are needed (no pulldown, no duplicate frames). Because of that you wind up with a 40Mb/s datarate.

That's the long answer.

The short answer is that in 720 mode you are only recording 24 frames per second, and in DVCPro50 mode you are always recording 30 frames per second.

Does that make sense?

Once the DVCPro50 footage has been converted to a 24P file in post, the datarate of the two would be the same (40Mb/s)

(24 frames per second) X (1.667Mb per frame) = 40Mb/s

______________

You'll notice that I kept referring to "720" DVCProHD.
That's because the DVCProHD codec also has the ability to record 1080.
However, in 1080 the codec is interlaced and records only 30 frames per second (just like DVCPro50).
But you still have the same 100MB/s datarate.

So, just like in the DVX the 24frame signal has a pulldown added and a 30fps signal is recorded.
So, if you shoot 24P in 1080 DVCProHD you'll have to record the full 100MB/s of data.
(just like you have to record the full 50Mb/s of data in DVCPro50)

But once you remove the pulldown in post and get your progressive 24 frame file you will have a datarate of 80Mb/s (twice that of 24fps recording of both 720 DVCProHD and DVCPro50)

Edwin Hernandez
November 8th, 2005, 10:08 PM
Luis: Thanks for a very detailed answer.
-EDWIN

Barlow Elton
November 9th, 2005, 08:53 PM
Thanks for clarifying what I couldn't, Luis.

On an added note: Blackmagic HD cards will record HD-SDI input into the non-100mbps flavors (720 or 1080i) of DVCPROHD, just like a P2 card will. I captured an entire features' 35+ hours of HDCAM 24p footage into DVCPRO HD 1080 24p onto a 2 TB raid. The bit rate was 80mbps. Makes a good HD offline format for HDCAM.

Luis Caffesse
November 10th, 2005, 12:08 AM
Thanks for clarifying what I couldn't, Luis.


Your explanation was fine, but I figured it's come up more than once on these boards so it would probably be best to try to spell it out in detail for anyone else who might be confused.

Nice note on the Blackmagic cards.
Brings up an interesting point though (not to change subjects of this thread)... as things stand now we remove duplicate frames and/or pulldown upon capturing (i.e. 24P & 24Padvanced on the DVX).

If we record 1080P on the Firestore we wind up with a 30fps file on our hard drive. Are we going to have to add an extra step in post to remove the pulldown?

I someone utters the word 'render' I swear I'm going to scream.
Hopefully there will be some way to strip those frames upon importing to your NLE (fingers crossed).

Barry Green
November 10th, 2005, 01:48 AM
If we record 1080P on the Firestore we wind up with a 30fps file on our hard drive. Are we going to have to add an extra step in post to remove the pulldown?

I someone utters the word 'render' I swear I'm going to scream.
Hopefully there will be some way to strip those frames upon importing to your NLE (fingers crossed).
It is my understanding that the pulldown is done exactly the same as on the DVX, meaning that there are flags in the data stream that identify the pulldown cadence. Some programs can either strip the frames out on importing, or just ignore them. FCP can work with 24PA footage without having to remove pulldown, but for 24P 2-3 pulldown I guess it has to have a separate process to remove the pulldown. Vegas just "knows" and removes pulldown on the fly. Presumably they will be able to implement identical processing for the HD version, as the pulldown is handled identically to how the DVX does it.

Luis Caffesse
November 10th, 2005, 07:27 AM
Whew, thanks for the good news Barry.
Glad to see you didn't have to use the "R" word.

Barry Green
November 10th, 2005, 12:23 PM
Well, depends on the editor... it's my understanding that if you're using an editor that requires the pulldown to be stripped out, then yes, the dreaded "R" word *could* come into play. If you're using a 24PA (2-3-3-2) file, and for some reason your editor can't handle with just skipping the duplicate frame on the fly, then it could be removed and a new file written to disk that contains only the original 24p source frames, with no "r'ing" necessary.

However, if you're using 24P mode (2-3 pulldown) and your editor insists that it needs a raw-frame source file, then yes, it would need to be "r'ed"... the pulldown gets extracted, the individual frames get reconstructed, and then that would get written out to disk. And that re-writing out to disk is going to cost some quality, depending on how the file gets written out (if it's written out as raw uncompressed, no quality loss will happen, but if it's written out using a lossy codec, obviously some quality hit will occur).

It's certainly possible to avoid this whole process, as Vegas has demonstrated because it allows on-the-fly dynamic pulldown removal from 2-3 files. But just because they can do it doesn't mean everyone else does.

So, to recap: the dreaded "R" *should* be completely avoidable if using 2-3-3-2, even if the editing program can't handle the active-frame flags and insists on you creating a new 24p-frame-only file. But using 2-3 pulldown may involve having to resort to the "R" depending on how the NLE chooses to deal with 2-3 pulldown.