Eric Escobar speculated about where or not what has happened in the world of print will happen in the world of video post-production, specifically to the job of the editor. Automation has completely changed entire industries, to the point that right now, software can piece together a story in a writer article and identify pictures. How long before they can use those same algorithms to assemble an edit?
To discuss what that means to editors today and to the profession as a whole, Eric talks with Scott Simmons, Steve Hullfish and Christian Glawe. In the podcast the guys discuss the kind of nuance that only a human being can put into an edit, whether “good enough” will be good enough for some people, whether or not now is a good time to get involved in post-production, and plenty more.
Scott: …you bring up a good extension of the article. Automation is one part of what can happen to media in general, but outsourcing and finding the cheapest of the cheap to do some of the legwork and less creative stuff is definitely a real issue. Now Christian, you’re out in LA where more content is edited than most places. Are you seeing people trying to outsource, even if it’s just shipping a drive to some really cheap person up in Montana somewhere. Is automation or outsourcing that affecting your market or you yourself?
Christian: It doesn’t really, and I tend to agree with Steve on a lot of this. Storytelling is such an abstract discipline, and the example that comes to my mind is the movie Gladiator. The movie opens with this wonderful shot of Russell Crowe’s hand over the wheat field, and that decision happened in editing, it was never scripted that way. That was a decision made in an inspirational moment in the edit room, and that stayed in the film. I have a difficult time seeing a computer being able to do that.
Scott: Let’s talk for a minute about that whole idea of outsourcing. Are we seeing a world where the Internet becomes so fast and the cloud so grand that you could upload all your dailies, send them over to someone overseas and have them do a rough cut from your script, or are we talking more outsourcing to young, cheap people?
Steve: It could be either one. I get projects in Chicago that normally would have to be done in New York and then LA. All of the my biggest clients are New York, LA, Nashville and Dallas.
Scott: Now Steve, you are a creative editor, a craft editor, so you’re not just an online editor or colorist. So if we’re talking about automation or outsourcing, it seems to me that most creative part of the edit will be the hardest thing to outsource or automate. Is that a true statement?
Steve: I think Christian is right. How do you make a decision like the opening short of Gladiator? I just cut a feature film that’s coming out at the end of this month, and even after the whole thing was edited, the decisions of trying to cut it down to where it feels right so it’s not too long and there aren’t dead spots couldn’t be done by the computer.
Scott: That’s true, but I think we should divorce this conversation from the feature film talk. If you think about all the editors and all the people editing in the world, those cutting feature films are such a tiny minority, especially in the media world we have today. I think the talk of features should be off the table because so many people are making a living in post-production that aren’t working on features, so I’m thinking of this other 99%. I don’t cut features, so is it us that have to worry about our jobs going overseas or being automated?
Eric: I absolutely agree. I wouldn’t be so quick as to decide what is the edge case, and that’s not something software can play a role in. It’s not like computers are being taught to edit video. What happening is that there’s deep dives into databases of edited material. Models are being assembled. Are they going to be original? No. They’re going to be copycat, they’ll be in the style of, or they’ll look like something. Nothing is ever going to replace that person’s mind that can do that, but what you’re going to get is an assembly of the millions of minds of people who have been editing for the last hundred years coming up with sophisticated pieces of software that can say, “you’ve got this shot of this hand and this shot of this person, here are a bunch of interesting ways this has been done before, perhaps this could be useful to you,” in that sort of helpful way that Amazon can suggest the perfect present for your Dad based on what you and your Dad browse. It doesn’t feel like this software is coming in and doing it, but it’s offloading a lot of the skill to referencing something no human being could ever do, which is analyzing million of films and edits to offer up suggestions. And to Scott’s point, we can look at the edge case, but we back down from that and look at the vast majority of content being created, and moving forward the vast majority of content that’s being created is not feature films or TV shows. It’s this stuff that’s the bread and butter work for these small companies and one-person shops. Those are the places that are going to feel the pain first and most. It just becomes an option for a client to never enlist the services of a video editor because there’s a set of software tools available to them that take their video content and arrange it in interesting ways that tells some kind of story or is on-brand for what they’re doing that looks a lot like the material that’s already out there and has been done before. That’s the sort of material editors use to cut their teeth and how people work their ways up the ranks.
Scott: Well, you know I’m going to break my own rule and mention feature films, because with what you just said I can almost see an automated type of thing where you could get the most basic rough cut of a film that’s done by automation, but then the craft editor, who we all know and love, has to come in and sort of finesse that and make it better. As a craft editor myself, that feels so painful to me. The thought of even having that happen three times a year on a job where someone used software to get the project to a certain point and then says, “okay take this and make it better,” is awful. Do I still go in and have to watch the dailies and make notes, or do I just take what it’s got there and try to trim? I don’t know the answer to that.
Eric: I think the answer is pretty clear, because it’s not going to be driven by you. It’s going to be driven by the producers who have figured out this way to reduce the amount of labor involved in putting together that first assembly. They’re going to say they’re on a tight schedule and they need to get this thing done. That’s what all this stuff does. It not only devalues the work, it reduces costs in places where you really shouldn’t be reducing costs because it looks like it’s removing labor in some way. The think that gives me pause is what’s becoming more and more available, which is online editors. That is, software as a service for editors. You go on, you upload your footage and you’ve got this cool tool that you can use to cut this stuff together. I think YouTube might have one, someone else might have one, but these NLE’s that exist in the cloud as software as a service are freaky because as more and more people start using those things, it’s not just the final edited project that becomes a source of information for the database. It’s the decision making process of the editors. Because you’re right. There are all these decisions that editors make to decide when to put that hand in front of the wheat field kind of thing. There are all these decisions that go on. There’s fifty or a hundred different versions of that edit that they tried. There’s all the things that you throw out which, to a smart piece of software, is interesting information. All the decisions that they didn’t make. There are all the ideas they looked at, tried and discarded. There are patterns and things that can be gleaned and replicated and then rolled into interesting and useful tools.
Steve: In some ways I’d almost welcome some of that. To eliminate some of the tedium of what you do on a daily basis. It kind of reminds me of the simplicity of what editing can be. One of my first editing jobs, editing a news story when I was a youngster, I had somebody watching over my shoulder while I’m trying to find the perfect shot, and the guy goes, “what are you doing,” and I go, “I’m trying to find the right shot.” And then he said, “do you want to edit or do you want to screw around?” So that was basically him saying that I needed to put something in that spot without worrying about it too much. That’s what I feel the computer is going to do. I’m not trying to find the exact right moment. Think about a long shot. Someone shoots a couple of minutes of some action and you find just the right three seconds of it to cut into a spot. I can’t imagine software can do that, but if it can that could be a good thing, but it does take you down a slippery slope.
Christian: I think that where we’re going to see this type of automation is with things like what Facebook did about a year ago, which was all the rage. You could ask Facebook to take all of the pictures that had been posted to your account and it would put them into a montage and publish it. So for certain montagy things, I could see it happening, or even using Steve’s local news analogy. Maybe. Perhaps. But where I think we’re going to see automation really making waves and where it already has is in what was traditionally the assistant editor role, in terms of logging. I believe FCPX has facial recognition to aid in logging already. And also in maybe things like rotoing. I think those are the areas where we’ll see automation making waves. On the other hand, computers still don’t even have speech-to-text down, unless you feed it a pre-written script. I think Adobe basically threw up their hands and gave up on putting documentary style or non-fiction reality footage into its’ algorithm and getting an accurate transcription of it. We’re not even there yet with computers, but someday we may be.
Scott: They had speech recognition, but I believe the last version took it out. I actually had good luck with that on really good recorded interviews in terms of getting something readable and usable out of it, but I know it’s a very difficult thing to deal with. But I guess that’s the discussion. If these computers get that much smarter and faster then at some point even my fast-talking can be recognized. That’s going to happen, and when it does maybe this automation thing goes even further, because then it can analyze the actual dialogue, instead of just the picture.
Eric: And to Steve’s point, that it’s not one or the other…
That’s just a preview of what Scott, Eric, Steve and Christian discussed during the podcast. Hear the rest about what the guys have to say around how automation and outsourcing will impact professionals across the industry and plenty more via the links below.