View Full Version : What P2 format needs is MPEG compression


Radek Svoboda
June 21st, 2005, 01:52 AM
With mild MPEG2 or 4 compression quality would stay high but media cost would go down significantly. I vote for both DVCPROHD and MPEG compression modes. With P2 there will be not potential dropout problem that Sony tape based HDV camera have.

Radek

Radek Svoboda
June 21st, 2005, 02:07 AM
I'd rather have 50 Mbps MPEG2 than 100 Mbps DVCPROHD 1080i60, with 2x the media cost. 50 Mbps would still be significantly better than 25 Mbps HDV.

I'd rather have 40 Mbps MPEG2 than 40 Mbps DVCPROHD 720p24 because MPEG2 at same bit rate would have superior picture quality.

So DVCPROHD and MPEG mode switch make sense and should be included. If Panasonic listens to complaints here about high media cost, they should include such switch. That give poeople different options. If you don't like MPEG, don't use it.

Radek


P.S.

There is new development. I just posted about it in News section. Samsung, after making SD cards makers believe they can use Samsung's NAND memory in cards now wants not to allow that. SD Cards were developed by Panasonic. Samsung working closely with Sony. Who knows what's going on behind scenes. IMHO Samsung may be in for lawsuit by Panasonic.

Since this is latest news, IMHO Panasonic may decide change storage options on 200 camera. Hard drive and/or MPEG compression option would be nice.

R

Rob Lohman
June 21st, 2005, 04:59 AM
I highly disagree. In my opinion it is a very bad idea to use an intra-frame
compression based algorithm like MPEG-2 (I think MPEG-4 is intra-frame as
well, not sure on that though) for your source footage that still needs to
be edited, especially at HD resolution.

I'd much rather see a high bitrate inter-frame compression algorithm like
MJPEG (which a lot of professional analog capture boards use).

Radek Svoboda
June 21st, 2005, 05:38 AM
Rob, naturally, but I'm talking about reducing P2 media cost, which is very expensive. It would be just extra mode on the camera, inter frame still be available.

Sony uses MPEG4 in most expensive CineAlta SR, MPEG2 in very expensive XDCAM, Panasonic mentioned possible MPEG in conjunction with their high end D5. In ideal world where money and size is no object, inter-frame is best, in practical world movement is more and more towards intra-frame.

Radek

Chris Hurd
June 21st, 2005, 06:32 AM
Actually the MXF files which are written to the P2 cards during recording *already are* based on MPEG-4, so it pays to do your research... see http://www.pro-mpeg.org/ for more info.

Steven White
June 21st, 2005, 08:35 AM
This argument is quite simply:

What do you think is better? DV or HDV? Given the choice have having either with identical lattitude and sensitivity, which would you shoot? This is the question that faces HDV owners since our cameras do both... and the answer is almost always HDV. Using inter-frame MPEG-2 at the same resolution as I-frame only compression will always yield superior results. In fact, the very worst you can do for a given data rate with a properly implmented inter-frame compressor is I-frame only compression.

In short, Radek's got this one right. Actually, in a direct analogy to HDV, using the exact same process you could acheive 2560x2430 HDV quality using a 100 Mbps stream.

The way I see it, the industry is going to go one of two ways:
Either we're going to get to 4:4:4 uncompressed, or we're always going to try and do better by upping the resolution and using compression. I'm not really sure which I'd rather. Realizing with 100 Mbps we can't even do uncompressed SD, it seems like we're not going that direction.

People bemoan compression, interlace and colour sampling schemes as archaic and counterintuitive... but when you look carefully at the thought that went into them, you realize the sheer genius of every one of them. With the 1.2 Gbps necessary for uncompressed 4:4:4 1080p24, you could get a hell of a lot more REAL information using compression algorithms.

-Steve

Simon Wyndham
June 21st, 2005, 09:50 AM
Hmm. From what I can tell, all this stuff about getting Panasonic to add this that or the other feature to the HVX is in vain. The HVX is not being aimed at the DVX100 level market. To shoot HD there's a huge difference in price. To be fully setup for using the HVX200 as well as edit using it, you will have spent the same as a lower end 2/3" CCD SD camera such as the DSR500!

The HVX200 will have what it has. The design will have been finalised ages ago. Now it is just a case of figuring out how you will use it and work around its limitations.

I would be more worried about how much any speck of dust will be visible on the lens in HD mode given that it will most likely have an absolutely huge depth of field just like the DVX.

I would also be worried about how, given that the camera will be finalised by now with no chance of MPEG etc, I would shoot live events. The limitation of P2 sizes very much limits the camera to short news segments and indie filmmaking. Documentary production on P2 with this camera will be troublesome.

In fact with regard to P2 it is odd that I have never really read any first hand accounts from people who have used the system in the field, and how they coped with the running time of the cards and having to offload them regularly.

Rob Lohman
June 22nd, 2005, 03:41 AM
Steven: I agree that the advanced compression algorithms out there today
are sheer genius. We can't (currently) live without them. Heck, I've twice
designed some compression algorithms of my own (I am a programmer) and
that is very hard to do.

However, with that being said I still believe at least MPEG-2 is just the wrong
format for editing. Yes it will reduce file sizes (but there are much better
formats for that already out, mainly MPEG-4), however it is really problematic
for the EDITING SOFTWARE BUILDERS to work with that format. Which in turn
means slower performance for us. Just look at it. Almost everybody who is
doing HDV editing is converting to some other codec (Aspect HD for example)
to do the actual editing work, heck, a system like this is deliverd WITH the
latest Adobe Premiere & Sony Vegas out of the box to speed up editing.

In my opinion that is just the wrong approach. There are better ways out there
to handle that problem, no doubt we really do need compression (especially
for HD sources) to store our data and try to reduce prices on P2 media.
We all agree on that.

Anyway, as Chris already pointed out they are going to use some system
based on MPEG-4. I will read up on this to see what they are doing exactly.
I'm not judging how this new camera handles stuff, merely that I think
MPEG-2 would be wrong (which they didn't choose anyway, hopefully they
made some sane choices with MPEG-4).

R Geoff Baker
June 22nd, 2005, 07:04 AM
I highly disagree. In my opinion it is a very bad idea to use an intra-frame
compression based algorithm like MPEG-2 (

Note that intra-frame is exactly what the DV family is, and what MPEG2 (except for I-frame only) is not -- MPEG2 is inter-frame, not intra-frame.

Given the plunging costs of storage media of all types, and the developed if not bleeding-edge status of the DV family of codecs .... lets stay with 100Mb/s 4xDV 4:2:2.

Those that are unwilling to swallow the cost of storage media, and must have results today instead of later when they find the costs managable, can stream from the camcorder to some external compressor of their choice, to a storage media they can afford.

JMHO

GB

Steven White
June 22nd, 2005, 08:14 AM
MPEG-2 is just the wrong format for editing

I wholly agree.

However, I think the time that your acquisition format and your editing format must be the same is at an end. We are not particularly limited right now by storage - we can get a lot of that for cheap. What we are limited by is bandwidth. We can't write data to portable storage fast enough - if we could, we'd all be doing uncompressed work without thinking about it.

I say an acquisition format should be able to cram absolutely the most information into the given bandwidth in the best way possible. Obviously, there are different interpretations for this, but I maintain that for a given bandwidth, the worst you can possibly do is use an I-frame only approach. If you come up with a good acquisition format that is also good for editing - power to ya. But post-production shouldn't be the primary motivator for acquisition.

As for editing, if we want to get serious about high-resolution editing and effects work, we're going to have to get familiar with the notion of proxy and digital intermediate workflows.

-Steve

R Geoff Baker
June 22nd, 2005, 09:05 AM
Actually the MXF files which are written to the P2 cards during recording *already are* based on MPEG-4 ...

Chris, I don't doubt that this fact can found at the site, but can you direct me to it? I found comments like this: "(including but not limited to MPEG based coding)," but that seems rather different than 'based on MPEG-4'.

Cheers,
GB

Barry Green
June 22nd, 2005, 04:38 PM
MXF is a "Media eXchange Format", but doesn't require MPEG, does it? Nothing on the P2 card will be MPEG format (unless you're talking about low-res proxy files on the SPX800). But the files will still be stored using MXF. I believe that MXF is a universal audio/video file interchange format, basically an alternative to .mov or .avi, but doesn't specify what the contents of that file should be.

Chris Hurd
June 22nd, 2005, 09:15 PM
Oy. Geoff, I am indeed forced to back-pedal on this and must re-state myself.

MXF is a derivative of AAF (the Advanced Authoring Format), not MPEG. However, MXF was indeed developed by the Pro-MPEG Forum. MXF has been described as "the digital equivalent of videotape." Part of the specification for the MXF file structure allows for a Generic Container, which can encapsulate DV or MPEG. So I guess I should revise my comments to state that the MXF file wrapper can certainly accomodate MPEG-4.

And in fact, Panasonic already has an MPEG-4 proxy encoder card for its high-end P2 camcorders. The model number for this particular P2 card is AJ-YAX800. It is not yet available as far as I know nor is it clear whether this MPEG-4 encoder card will be compatible with the HVX200. But I am right in the sense that Panasonic has already thought of direct MPEG-4 encoding via P2 at least for proxy purposes. It may not be a done deal just yet, but it's obviously very much in the works.

AAF (Advanced Authoring Format): http://www.aafassociation.org/

MXF (Material Exchange Format): http://www.anoca.org/smpte/essence/mxf.html

Chief Architects of MXF: http://www.pro-mpeg.org/index.html

How MXF relates to AAF: http://www.snellwilcox.com/knowledgecenter/mxf_aaf.html

MPEG-4 into MXF file structure via SMPTE standard
#381M Television – Material Exchange Format (MXF)
Mapping MPEG Streams into the MXF Generic Container:
http://www.smpte.org/smpte_store/standards/pdf/numindex.pdf

About the Panasonic AJ-YAX800G Proxy Encode Card, from
https://eww.pavc.panasonic.co.jp/pro-av/sales_o/p2/p2cam/index.html:

"Mount an AJ-YAX800G Proxy Encode Card, and the AJ-SPX800 records MPEG4 proxy (low-resolution) data — useful for news flash or other studio news system use — onto the card along with the full-resolution data. The three levels of proxy video are available: 1.5Mbps, 768kbps or 196kbps. Proxy data can also be recorded onto an SD Memory Card mounted in the slot provided, for easy viewing on a laptop PC. The encode card, available as an option, lets you upgrade as future image encode systems evolve. Proxy data is AV data with low-resolution MPEG4 video and audio containing time code, metadata, and other control information."

See also this partial quote from http://www.blue-order.com/pdfs_news/22-09-2004_NBC_Olympics_and_Blue_Order.pdf:

"The Media Archive system at Athens 2004 also was the first MAM system to leverage the capabilities of Sony’s new generation of network-enabled A/V devices: fifty Sony e-VTRs were used as ingest devices, recording incoming camera feeds onto videotape and at the same time creating MXF MPEG-2 files at 50 Mbit/s and MPEG-4 files at 2 Mbit/s, with the MPEG-4 files being time-coded,frame-accurate low resolution browse proxies of the MPEG-2 content. As part of the development effort for the project, Blue Order enhanced its Multiformat Browse Architecture (MBA), an integralpart of the Media Archive Enterprise Media Management Platform, to support this MPEG-4 browseproxy format, eliminating the need for additional browse encoders and for the first time using the upcoming MPEG-4 format for browsing, EDL creation and keyframe extraction."

Let's all please remember that P2 is not a recording format; it's just hardware. The actual file structure on a P2 card is MXF, which can be used for just about any sort of digital video recording, including DV, DVCPRO50, DVCPROHD, and yes, even MPEG-2 and MPEG-4. Hope this helps,

R Geoff Baker
June 23rd, 2005, 04:36 AM
Thanks Chris, that's a good summary.

Cheers,
GB