Story of IMF:  Annie Chang's Recollection

I am probably one of about 3 people that knows how IMF began.  In 2004, videotapes were still the norm for video masters.  After having worked with Pixar Animation on The Incredibles home video mastering when I was at THX Ltd., I thought it was silly for them to create everything in a file-based manner, only to make a longplay videotape which was then encoded to make a file again (at the time for DVD).  In 2005, I joined Walt Disney Studios and told the VP of Production Technology, Howard Lukk, that we needed a file-based master format that could handle all the different language versions instead of using videotape.  Howard and I visited Pixar to discuss this idea, and Rick Sayre, the Supervising Technical Director, said, "Why don't you do something like Digital Cinema?" That’s when a light bulb went off for Howard and I.  At the time the Digital Cinema standards were quite nascent, but the format was flexible with the Composition Playlist (CPL) and was able to create different versions with only the uniquely different pieces of audio and video.

After doing a little internal discussion at Disney, we determined in 2006 that having a Disney-proprietary file format would not be as useful since we sometimes had to exchange masters with other studios.  We decided that we should work with other studios and Jeff Miller, the President of Worldwide Postproduction and Operations at the time, helped organize the University of Southern California Entertainment Technology Center (USC ETC) to provide a safe, anti-trust space for studios to talk through the requirements of this new file-based format.

Many of the same people that directly worked on the requirements document are still involved with the current IMF effort. We had Arjun Ramamurthy from 20th Century Fox, Brian Vessa from Sony Pictures, Mike Trainotti from Paramount Studios, Brad Collar from Warner Bros., and me from The Walt Disney Studios.  Universal Pictures joined later with Harvey Landy.  We also had a few consultants helping out including John Hurst from CineCert, Mike Smith who was the JPEG2000 expert and Pierre-Anthony Lemieux from Sandflow.  There was an executive sponsor layer as well that included executives from each of the Studios to oversee the work. 

The USC ETC working group hotly debated and discussed the requirements of what we called the Digital Video Package (DVP) from 2006 through 2010.  At some point Chris Carey from Paramount suggested that we change the name to something like the Interoperable Master Format. We stuck with it and finally completed the USC ETC document and submitted it to SMPTE in December 2010 as an input document for new work.

SMPTE decided to accept the work and Howard encouraged me to chair the IMF working group, but I was hesitant since I had never done any standards work before.  Little did I know that it was going to be quite the learning experience!

Everyone at SMPTE was very welcoming and appreciated that I brought a real-world problem that I understood well to them.  Pat Griffis, then future SMPTE President, took me under his wing and helped me navigate the SMPTE Standards Organization. Before the IMF work started, SMPTE HQ let me know that they were hoping that the effort would showcase the "newer, faster SMPTE.”  I wanted to fast track IMF as soon as possible because my studio (and others) needed a file-based format right away.

We broke into several working groups, one for each document which we felt would follow in the Digital Cinema standards format with a suite of documents.  We started off with working groups on some of the basic needs - the MXF wrapper and the CPL. As a working group chair, you end up having a lot of side meetings to figure out how to break through logjams, like the “Dreaded CODEC Conflict of 2011.” 

The studios wanted JPEG2000 Part 1 because it was license-free, but there were other companies that wanted MPEG-4 Simple Studio Profile (SStP) as the CODEC for the IMF standard.  We were at a standstill during an in-person April 2011 IMF meeting. We started saying that we were "IMF-ed" and dead in the water!  We brokered a compromise by adding the idea of Applications which allowed for specific use cases and constrained parameters for those use cases.  We ended up with IMF Application 2 which was the JPEG2000 version that the Studios wanted and IMF Application 3 which was the MPEG-4 SStP version that other companies wanted.  

The Application idea ended up being a game changer since it allows for other Applications to co-exist in the IMF eco-system.  Now that the blockers were out of the way, everyone went on their merry way developing their Application drafts. 

Towards the end of 2012 with almost 1.5 years of drafting work behind, it became a race on what would happen first - our first published IMF standards document or my first-born son.  Unfortunately for IMF, my son decided to come 5 weeks early; however, IMF still proved that SMPTE could publish standards quicker! The first IMF SMPTE standards document published was ST2067-5 Essence Component, which defined the MXF wrapping for IMF.  

Soon after, the next published IMF standard was ST2067-3 Composition Playlist.  Once these dependencies were completed, the other documents in the suite could be drafted and work was done on audio, the core constraints, and the output profile list.

While this work was on-going, I continued to do presentations to evangelize IMF. I created the first SMPTE webinar video on IMF to explain how it worked to non-technical people.  Mike Krause, from Disney, and I went to Netflix to introduce IMF to Chris Fetner, Steven Kang and several others who were there at the time to talk about the benefits of the IMF format.  Netflix was the first to adopt IMF.  I felt that they understood the complexity of IMF and knew that they needed to develop systems to be able to manipulate the various components of the Standard.  Arjun Ramamurthy and Mitch Jacobs at 20th Century Fox also did an amazing job at adopting the JPEG2000 version of IMF Application 2 which eventually ended up getting Disney to adopt IMF after they acquired Fox. 

As part of the strategy of creating an ecosystem for the Standard, several people and I encouraged implementers like DVS to create IMFs because we needed good implementations in the field that worked with each other.  We started plugfests to help with interoperability of IMF in the field since it was a completely new way of working for many of the companies that implemented the format. I am proud to see where IMF is in the world today, and I have to thank all of the people that worked on IMF with me and after me who carried on the passion for a truly ubiquitous mastering file format.  I'm truly honored to be the "mother of IMF."