Back       

Summary Notes on Jim Kutzner’s Talk

While derived from audio tapes and audience notes, the following generally represents a paraphrased condensation of main points, not a transcription. 

As a late entry, Jim was asked to present a talk limited to eight minutes.

“This conference is all about lossless compression, unfortunately we at PBS are all about lossy compression” for business reasons.

Jim introduced both what PBS and the National Digital Information Infrastructure and Preservation Program (NDIIPP) are doing.  Other NDIIPP participants present were noted: WGBH’s Dave MacCarn and Library of Congress’ Carl Fleischhauer.

Goals of the Public Broadcasting Service

PBS is a membership organization with 170 licensees.  It needs to provide a balance between quality and cost - unfortunately which are at odds.  For example, “it’s prohibitive to exchange data with lossless compression.”    [For why, see Post-meeting Clarifications below.]

PBS distributes 10,000 hours of unique programming per year to its member licensees.  The licensees in turn deliver content to the public via over-the-air broadcasts.  PBS Online also delivers content directly to the public, in concert with licensees.

The PBS Workflow

Business reasons are forcing a complete change of processes, driven by the need to replace traffic and scheduling systems, and relocate PBS this fall/winter.  Construction, new system development, and staff training are underway.  After all this is done developments elsewhere (such as at meetings like this) may make it necessary to go back and refine and retool the new systems.

In the old days, production, distribution, and preservation all were based on the same format: video tape. Now there are several formats for origination, production, distribution, emission, and display.  PBS is engaged a little bit in production and post-production, but mostly distribution, and fully cognizant of emission formats.  PBS has adopted a short hierarchy of formats.  It is recognized that upstream, stations may operate in lossless or quasi-lossless formats, for preservation reasons.

Right now, the latest SD workflow is:

·        Ingest video tape, since almost everything in-coming is videotape, mainly Digital BetaCam for SD (or HDCam for HD);

·        Save as IMX-50 aka D-10, within MXF.  D-10 uses MPEG-2 compression, specifically 422P@ML, I-frames only.  This is 50 Mbps plus associated audio;

·        Convert to a reduced bitrate for SD delivery via satellite to stations, namely MPEG-2  at 8 Mbps plus associated audio;

·        Starting next year, PBS will also deliver files at faster (or slower) than real-time at that same bitrate;

·        Stations then broadcast somewhere in the 2-8 Mbps range, generally 3-4Mbps.

This hierarchy targets what the viewer will see, but is affordable to capitalize and operate.

For the HD workflow there are similar “landing points.”  HD tape is ingested in uncompressed form, compressed for production, further compressed for distribution, and then emitted at 18 Mbps (or less).

The NDIIPP Project

When talking about preservation, NDIIPP is looking back upstream. Questions being asked are:

·        How do we create and save information for posterity, keeping in mind that we’re in business?  We are exploring formats.

·        What is the selection process?  Certainly finished program must be preserved, but what about other material “in the can”, such as scripts and graphics.  It would be nice to save everything, but it’s too costly.

·        Who will save it?  Until now, PBS has been the depository keeper of master video tapes for affiliates, under pseudo-climate control in the D.C. area.   There are now 140,000 titles, going back to the early 1970s and 2” or 1” videotape formats.  Some stations like WGBH also have there own archives.  In the future, who will save masters and for what purpose?

·        How long will this material be saved?  As discussed earlier in this meeting, is it for 50-100 years, or just for 5 years [to the next migration time]?

·        Why don’t we just jump to the final format now?  Because we’re at various conferences like this one, trying to answer these questions.

No Questions – Lunch Time

Post-meeting Clarifications

[from an exchange between Glenn and Jim]

Satellite delivery remains by far the cheapest form of a one-to-many, highly asymmetrical distribution system.  An 8 Mbps video can be easily delivered by one satellite transponder.  A transponder with guaranteed backup costs a couple of million dollars annually, a value that is decreasing quite slowly, unlike digital storage costs.

Delivery of virtually-lossless 50 Mbps D-10 files would need 3-4 transponders, with a proportional increase in annual costs.  (Similarly for any mathematically-lossless format with 4:2:2 chroma subsampling, likely to be of similar bitrate.)  And no one at home would ever see the difference, since the system bandwidths are reduced as one moves through the complete broadcast chain from producer to home.  Analog television at home can never reproduce the quality of videotape, much less the quality of D-10 or higher.  HD will provide a tremendous improvement but will be compressed. 

The economics get even more prohibitive for lossless compression at higher bitrates than D-10 (e.g., unsubsampled 4:4:4).  Thus, lossless compression has its use in high-end production and in preservation, but not as a means of “wholesale” (from PBS) or “retail” (from stations) distribution.