The present and future of post production business and technology | Philip Hodgetts

CAT | Interesting Technology

Because I am researching my journey through my earlier writings on metadata and interactive story telling I came across my ‘review’ of NAB 1998 thanks to the Wayback Machine. This was the year everyone was coming to terms with ATSC – digital broadcast – and how it was to be implemented. From my review it seems my attention was on interactivity and QuickTime 3, neither of which is surprising.


A few years ago, we considered supporting transcripts in Lumberjack System. At the time our goal was to quickly prepare for an edit, and transcriptions took days and cost serious money.

Two years ago we supported the alignment of time-stamped transcripts to Final Cut Pro X Clips and a year ago, introduced “magic” keywords, derived by a cognitive service. Since Lumberjack doesn’t (yet, I might emphasize) support a speech to text service internally, what are the options and what do they tell us about the state of play for transcription in April 2017?




In Just 10 Years

While projecting the changes that Artificial Intelligence (AI) and Machine Learning (ML) might bring about in the future, it was interesting to look back and see just what didn’t exist 10 years ago. Keep in mind that the Internet itself is only just over 30 years old.


One of the powerful way Artificial Intelligence ‘learns’ is by using neural networks. Neural Networks are trained with a large number of examples where the result is known. The Neural Network adjusts until it gives the same result as the human ‘teacher’.

However, there’s a trap. If that source material contains biases – such as modeling Police ‘stop and frisk’ – then whatever biases are in the learning material will be contained in the subsequent AI modeling. This is the subject of an article in Nature: There is a blind spot in AI research  and also the praise of Cathy O’Neil’s book Weapons of Math Destruction that not only brings up that issue, but the problem of “proxies”.

Proxies, in this context, are data sources that are used in AI programs that are not the actual data, but rather something that approximates the data: like using zip code as a proxy for income or ethnicity.

Based on O’Neil’s book, I’d say the authors of the Nature article are too late. There are already institutionalized biases in very commonly used algorithms in finance, housing, policing and criminal policy.

The extensive article by Steven Levy – The iBrain is Here – is a fascinating read on how Apple are using Machine Learning, neural networks and Artificial Intelligences across product lines. It’s well worth the time to read through, but this quote from Phil Schiller stood out:

“We use these techniques to do the things we have always wanted to do, better than we’ve been able to do,” says Schiller. “And on new things we haven’t be able to do. It’s a technique that will ultimately be a very Apple way of doing things as it evolves inside Apple and in the ways we make products.”

The ways this could all be aligned with editing? Speech-to-text; keyword extraction (just like Magic Keywords in Lumberjack System); sentiment extraction; image recognition; facial detection and recognition; speech controlled editing (if anyone really wants that), and the list goes on.

I’d like to believe the Pro Apps Team are working on this.

UPDATE: Ruslan Salakhutdinov is Apple’s first Director of AI.

Most of the thinking – the little that’s done – around the affect of Artificial Intelligence and Robotics replacing jobs, is somewhat negative, so it was almost a relief to read John Hagel’s perspective that we could use this transition as an opportunity to rethink the nature of work.


Maybe I’m pushing this subject a bit hard, but I really believe we are on the cusp of a wide range of human activities being taken over by smart algorithms, also known as Machine Learning. As well as the examples I’ve already mentioned, I found an article on how an “AI” saved a woman’s life, and how it’s being used to create legal documents for homeless (or about to be homeless) in the UK.




The Problem with Machine Learning

I’ve been talking about machine learning and smart APIs recently, where I think there is great potential to make pre-editing tasks much easier. But they are not without their downside. They are built on sample data sets to ‘train’ the algorithm. If that training set is not truly representative of the whole data set, then the results will go horribly wrong.

Cory Doctorow at Boing Boing uses the Trump campaign as an example of how this can play out in ‘the real world’.

A couple of recent articles have pointed to Artificial Intelligence writing, or contributing to, a screenplay. A narrative script. I find this fascinating, even though my own area of interest in applied AI is in non-scripted.

There is no doubt that computer algorithms – up to true AI – will be involved in productions future. Smart people will work out how to master it.




Automating Final Cut Pro X

Since starting work on the Lunch with Philip and Greg I’ve battled a little with the multicam. Largely because I’m using it in an atypical way, although I suspect setups like mine will become more common in the future.

My solution was Automator actions, triggered by Function keys and activating an AppleScript, so that the mode is first switched to Video Only (for angles 1 or 2) or Audio only (3, 4 and 5) before switching to the angle.  It reduces a lot of repetitive strain injury potential!

The tutorial is over at, but here’s a little background.


Older posts >>

June 2017
« May