Jump to content


Photo

Clarifying the data rate


  • Please log in to reply
10 replies to this topic

#1 Mark Allen

Mark Allen
  • Basic Members
  • PipPipPipPip
  • 591 posts
  • Director
  • Los Angeles

Posted 19 April 2006 - 05:44 AM

at 1080, in theory the HVX does 100mbits. This interests me because I need to shoot some GS and while I've done keying with DVCPRO50 - It wasn't quite enough. I'm thinking the extra 50 would solve the issue.

That said, I have a concern. A respected colleague has told me that, in fact, at 30p, the data rate is 50mbits, at 24p it's 40mbits.

When is the data being encoded? is it at the end of the stream or the top? is it really throwing away the data after it has spent its data bandwidth? That would be... really poor design.

How can I get the 100mbits out of this camera? My end product is Standard Def, but I don't mind shooting HD and shrinking down - that always does decent things for a key as opposed to the other way around of course.

Does anyone have the link for this information or know the answer? Thank you.
  • 0

#2 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 19759 posts
  • Cinematographer
  • Los Angeles

Posted 19 April 2006 - 06:28 AM

DVCPRO-HD is 100 Mb/sec for 720/60P. So if you are shooting 720/24P, it's only like 40 Mb/sec.
  • 0

#3 Phil Rhodes

Phil Rhodes
  • Sustaining Members
  • 11937 posts
  • Other

Posted 19 April 2006 - 07:22 AM

Actually, it's exactly 40mbps.

A shame, when the P2 cards are capable of so much more.

Phil
  • 0

#4 Alessandro Machi

Alessandro Machi
  • Sustaining Members
  • 3318 posts
  • Other
  • California

Posted 19 April 2006 - 08:07 AM

Seems weird to me that the mbps would be the same for a wide shot versus a close up shot. Also contrast levels could also skew the quality one is getting, as could motion in the shot. Is the idea that 40 mbps is enough no matter what is being shot?
  • 0

#5 Dan Goulder

Dan Goulder
  • Sustaining Members
  • 1259 posts
  • Cinematographer

Posted 19 April 2006 - 09:58 AM

Seems weird to me that the mbps would be the same for a wide shot versus a close up shot.

You're moving the exact same number of pixels at a fixed rate of compression, regardless of content.
  • 0

#6 Mark Allen

Mark Allen
  • Basic Members
  • PipPipPipPip
  • 591 posts
  • Director
  • Los Angeles

Posted 19 April 2006 - 03:50 PM

DVCPRO-HD is 100 Mb/sec for 720/60P. So if you are shooting 720/24P, it's only like 40 Mb/sec.


so... at 1080... I thought it was not able to do 60p at 10080 does that mean it would do 30p at 100 Mb/sec? and therefore 80 Mb/sec at 24p?
  • 0

#7 Stephen Williams

Stephen Williams
  • Sustaining Members
  • 4708 posts
  • Cinematographer
  • Europe

Posted 19 April 2006 - 04:04 PM

so... at 1080...


Hi,

I don't think you will get better pictures at 1080 with an HVX, there is a limit with its ccd's !

Stephen
  • 0

#8 Alessandro Machi

Alessandro Machi
  • Sustaining Members
  • 3318 posts
  • Other
  • California

Posted 19 April 2006 - 04:36 PM

You're moving the exact same number of pixels at a fixed rate of compression, regardless of content.


The camera may be moving the same number of pixels but that doesn't mean it's not a flawed concept.
  • 0

#9 Matt Irwin

Matt Irwin
  • Basic Members
  • PipPipPipPip
  • 389 posts
  • Cinematographer
  • Los Angeles

Posted 19 April 2006 - 05:27 PM

so... at 1080... I thought it was not able to do 60p at 10080 does that mean it would do 30p at 100 Mb/sec? and therefore 80 Mb/sec at 24p?

In the 1080 modes, there is no "native" function. 1080/ 60i, 30p, and 24p are recorded at 60 fps and then use pulldown to extract the desired frame rate. So the datarate would be 100Mbps for all frame rates in 1080.
  • 0

#10 Phil Rhodes

Phil Rhodes
  • Sustaining Members
  • 11937 posts
  • Other

Posted 19 April 2006 - 05:43 PM

Hi,

The data space required for a complex frame is the same as that required for a simple frame - because the complex frame is compressed more. To be technical, the complex frame produces longer discrete cosine transform coefficients which are truncated to the same length no matter what, so you end up losing more information.

This all comes from the need of a tape deck to receive a data stream at a constant rate - while I'm sure it might be possible to build a VTR servomechanism to record at a variable rate, as far as I know it's never been done and it'd be far more complicated. Even IBP MPEG compressors like HDV must try to make the GOP a certain size. Perversely, this makes the compression a lot harder to do - you can't predict what size frame you're going to get out of given picture data by applying a given coefficient truncation, and decent DV codecs try two or three times with different numbers to get to a size that makes best use of the available data space.

Sharp-witted individuals will notice that this means things like DVCAM are in fact not 25mbps because it's impossible to tailor the compression mathematics to create a frame of exactly a given size; what they actually mean is "up to 25mbps", or in fact "as near to 25mbps as our mathematics can get," which of course depends on the codec. It's not too difficult to imagine that the more tries you want to make, the faster your codec silicon has to be, and the more expensive it becomes - which is, at the end of the day, why expensive DV codecs perform better than cheap ones.

Phil
  • 0

#11 Alessandro Machi

Alessandro Machi
  • Sustaining Members
  • 3318 posts
  • Other
  • California

Posted 19 April 2006 - 06:02 PM

Hi,

The data space required for a complex frame is the same as that required for a simple frame - because the complex frame is compressed more. To be technical, the complex frame produces longer discrete cosine transform coefficients which are truncated to the same length no matter what, so you end up losing more information.

This all comes from the need of a tape deck to receive a data stream at a constant rate - while I'm sure it might be possible to build a VTR servomechanism to record at a variable rate, as far as I know it's never been done and it'd be far more complicated. Even IBP MPEG compressors like HDV must try to make the GOP a certain size. Perversely, this makes the compression a lot harder to do - you can't predict what size frame you're going to get out of given picture data by applying a given coefficient truncation, and decent DV codecs try two or three times with different numbers to get to a size that makes best use of the available data space.

Sharp-witted individuals will notice that this means things like DVCAM are in fact not 25mbps because it's impossible to tailor the compression mathematics to create a frame of exactly a given size; what they actually mean is "up to 25mbps", or in fact "as near to 25mbps as our mathematics can get," which of course depends on the codec. It's not too difficult to imagine that the more tries you want to make, the faster your codec silicon has to be, and the more expensive it becomes - which is, at the end of the day, why expensive DV codecs perform better than cheap ones.

Phil



Why not just relate discrete cosine transform coefficients to the lens setting and contrast, meaning the camera itself generates bigger and smaller pixels depending on the lens setting but the actual output to tape is steady, (as it should be).
  • 0


Ritter Battery

FJS International, LLC

Broadcast Solutions Inc

Media Blackout - Custom Cables and AKS

Abel Cine

CineLab

Willys Widgets

Wooden Camera

CineTape

Technodolly

Visual Products

rebotnix Technologies

Metropolis Post

Glidecam

Rig Wheels Passport

Gamma Ray Digital Inc

Paralinx LLC

Tai Audio

The Slider

Opal

Aerial Filmworks

Aerial Filmworks

Abel Cine

Visual Products

Metropolis Post

Glidecam

Tai Audio

Wooden Camera

rebotnix Technologies

Technodolly

Broadcast Solutions Inc

CineTape

FJS International, LLC

Media Blackout - Custom Cables and AKS

Rig Wheels Passport

CineLab

The Slider

Ritter Battery

Willys Widgets

Paralinx LLC

Opal

Gamma Ray Digital Inc