Archive of TideArt content.
Fri, Jul 22 2011 14:20:21 UTC
Interview about Novacut
Back to index
Today we have an interview with Jason Gerard DeRose, Lead Developer for Novacut
, in which he describes this new project for us and the Kickstarter program they are currently running.
Could you briefly describe yourself and your background?I'm a software developer. Professionally I've tended to work on web backends, usually database heavy stuff. My last gig was at Red Hat where I had the pleasure of working with the great folks on the FreeIPA team. But my personal passion has always been multimedia apps.
What is Novacut?Novacut is our 2 pronged plan to help artists be profitable telling the story they want to tell.
First, we're building a direct-to-fan distribution platform and a player that tightly couples the viewing experience with the opportunity to support the artists who worked so hard make what you just watched. We want to gather all the great direct-to-fan productions like Vampire Mob and Pioneer One in one place, get them a bigger audience, build a better user experience, and make them more money. This is also where Novacut Inc will make money: when artists make money through our platform, we take a cut, which we'll use to cover infrastructure costs, pay designers and developers, and do things like revenue sharing with hardware manufactures to get the player installed by default on tablets, TVs, etc. But artists retain ownership and we don't want artists to be exclusive... we want them to make money anywhere they can. So when artists make money elsewhere, obviously we're not trying to take a cut then.
Second, we're building an open-source pro video editor designed for storytellers. The most exciting thing about our editor is it's collaborative, meaning you can work with other artists in a geographically distributed team. Our editor will even allow real-time collaborative editing similar to Google Docs. We're initially focusing on making it as good as possible for HDSLR cameras like the Canon 5D Mark II, because that's what the majority of the direct-to-fan productions are using. And I would say the most important thing about our editor is that it's being designed through constant, open dialog with artists. We're not making assumptions about what storytellers need. We're crazy OCD about design. We start with research, which mostly consists of talking to artists, and then with each little baby step we take, we vet the design with artists. Because so many artists have been so helpful in guiding our design, we're going to end up with something amazing. Our editor wont be suitable for all pro video, but for storytellers on a tight budget, for whom time savings is so critical, our editor is going to lead by a *wide* margin.
How did you get started with Novacut, what were the initial goals?Jeffrey Ballagh (business lead, co-founder) and I have daydreamed about something vaguely like Novacut for probably a decade, so the idea didn't come out of nowhere. Tara Oldfield (community manager, co-founder) is a photographer and shoots with a 5D Mark II, and starting to play with its video capabilities was the first ''ah-ha'' moment that the market was ripe for something like Novacut.
We're also big Joss Whedon fans and were furious that Dollhose was canceled (which Fox buried alongside Firefly, I suppose). We've always dreamed of doing something to allow artists to tell their story and get it to their fans without interference. Somehow it finally all clicked during a month when I was basically procrastinating on finding a new job.
We knew a collaborative editor would really help artists, and be a big differentiator for us as it would be a world first. But initially the editor appeared to be far too much work, so we planed to start with the distribution platform and player.
Eventually it dawned on me that if we built the editor around CrouchDB, we could basically get the collaborative part of free, not to mention drastically simplify the design. Still, the editor idea was on the back-burner. It was Jeff in the end that decided we had to start with the editor because that was what made Novacut sexy, and he was right. Artists are excited about the player, but the conversation almost always starts because of the editor. The design work on the editor has helped us build close relationships with artists because it forces us to intimately understand their storytelling process.
We're truly in this to help artists, but without the editor, we'd be rather disconnected and naive about what storytellers face day-to-day. So we think the editor and player are a match made in heaven.
Who all is involved in the development of this project?Ian 'IZO' Cylkowski did our stunning brand identity design.
James Raymond is the lead UI designer (and UI implementer, as James is extra awesome like that). James also did the narration in our Kickstarter video.
David Jordon did the Blender animation in our Kickstarter video, and is working on a Google Summer of Code project to add optical flow to GStreamer, the fantastic multimedia backend we're using.
Akshat Jain is a small miracle... his insistent belief in the project kept us going through some very hard times, and his constant work to promote the project and recruit new developers has played a big part in us accomplishing so much so quickly.
David Green has been the most consistent code contributor, although we haven't gotten to know him too well yet as he doesn't tend to hang out in the Novacut IRC channel.
Martin Owens wrote a nice chunk of backend code for us, and has been a real cornerstone in our community.
Rockstar helped us get our projects setup on Lauchpad according to best practices, and has been a great friend to Novacut.
Jeff Ballagh is the business lead and has done wonders for us in terms of refining our message and focusing our efforts. He also has an uncanny ability to tie up all the loose ends I'm too scatterbrained to manage. He's currently busy with a super secret skunk works project (hint: not related to editor).
Tara Oldfield basically spends all her time looking for high-quality direct-to-fan productions, and building relationships with the artists who create them. As Tara is an artists herself (dancer and photographer), she has done wonders in translating my geek-speak to human. All our key UX research has grown from conversations Tara has started with storytellers.
And I'm the lead developer, responsible for the overall Novacut architecture and code quality. You could perhaps call me the lead UX designer also, but really the UX design comes from the storytellers who are graciously steering us in the right direction. However, I am *definitely* the lead UX typist (hehe) when it comes to formally documenting our UX designs.
And there are a bunch of other people who provide constant moral support, small code contributions, and overall awesomeness.
Why did you decide to develop a brand new software package instead of a plugin for an existing editing program?All in all, collaborative editing is just a bigger feature than we could do through a plugin.
But there is a deeper, more important issue. We feel that artists absolutely must be able to shape their creative tools, both by being included in the design process and by being able to modify their tools at anytime without permission. Artists deserve open-source tools.
How far along is the development of the program?Our overriding design goal is to save artists time. In our own experience using HDSLR for weddings and timelapse, we knew file management was stressful, distracting, and a huge time suck. We also needed a sane way to get files back and forth between artists when collaboratively editing. And we solved both with the Distributed Media Library (aka dmedia).
dmedia is a bit difficult to explain as there has never been anything like it, but dmedia:
- provides the fastest, safest, least distracting ingest workflow on the planet
- automatically backs up your files onto multiple drives, computers, and the cloud (cloud is opt-in, optional)
- automatically frees space (only when absolutely safe to do so) to make room for active project files, new ingest
- downloads assets from local computers and the cloud as needed
- creates smaller proxy (offline) versions of full video, which helps with upload bandwidth issue for collaborative editing
- works equally well for content consumption (so well that we'll use it as a backend for our player too)
- provides rich tools for tagging, rating, and organizing your media.
In FCP terms, dmedia would basically replace Final Cut Server, although in comparison dmedia is a sleek piece of alien technology. dmedia is almost production ready. This is a huge piece of work, and gets us easily 75% to collaborative editing. A lot of functionality that otherwise would be in our editor has already been implemented in dmedia.
We spent the last year doing extensive UX research, talking to artists, and we deliberately didn't let ourselves start working on the editor user interface till we had a really solid picture of where we were going, and how we were going to measure success. But now we're off and running with the UI.
Our first UI task is insanely fast frame-accurate cutting and sequencing. We're still prototyping and the design isn't final, but one innovation we're bringing to the table is what we call the storyboard view. We noticed that during the narrative *sculpting* (where editors spend a lot of hours), you really don't care about the duration of the slices, and that a tradition timeline makes it difficult to see your story. So our storyboard view does away with time, and instead clearly shows you the starting and ending frame of each slice. You'll be able to flip back and forth between storyboard and timeline view, work in whatever is fastest for your current task.
I encourage people to watch this 1 minute demo video to see just how quick and easy it is to make frame accurate adjustments from within our storyboard view.
One other thing I'll mention is that we're building on very high-level open-source components, so the editor isn't nearly as much work as it might seem. It's certainly a difficult and ambitious project, but we're not chasing windmills. Importantly, we're not touching the actual video processing (by far the most work) because we're using GStreamer.
To be on the safe side, I'd say we're 3-4 months away from the editor being usable: limited features, but fully collaborative with wonderfully fast cutting and space-age media automation.
What are the benefits of going with an open source model?For one, risk reduction, something we've been talking about since at least September 2010. Back then, we probably seemed needlessly paranoid, but these days, that message is resonating with everyone from low cost HDSLR productions to Hollywood.
For another, considering how important software has become in the storytelling process, it's ridiculous that artists can't directly shape their creative tools. The idea that you can meet your users needs by designing in secret without talking to your users each step of the way... were that true, it wouldn't be so easy for small fries like us to come along and do better. Honestly, most of our innovations thus far have been picked from low hanging fruit, strong evidence the competition didn't even try.
Now I realize people are rightfully skeptical: many open-source video editors have come before us, and they didn't deliver something superior for professional artists. Heck, they often didn't even deliver something *usable* for professional artists. But I assure you, Novacut is different for three key reasons: design, design, and design!
We're not scratching our own itch. We're carefully designing for our target user, and no one else, not even ourselves.
You currently have a Kickstarter with a $25,000 financian goal, how is it going so far?At the time I'm writing this, we have 211 backers and $11,025. Our campaign ends Friday, July 29, and if we don't reach at least $25,000, our backers are *not* billed, and we don't get any money. Kickstarter is all or nothing.
If you'd like to help us out, we most of all need help getting more press, and the best way to do that is by asking your favorite blogs and news sites to interview us or to write an article. We have a compelling and buzz-worthy story.
Oh, and if you happen to be at Comicon, we'd be forever grateful in you talked to people about Novacut!
How will the money help with the development of Novacut?We were broke when we started Novacut. We tried Kickstarter, and it failed. But we drug ourselves along for nearly another year, in the process burning through every drop of savings, maxing out all our credit cards, and borrowing basically all we can from friends and family.
We've worked so hard and we're so close to our editor being usable, being tangible. We're certain that even if all our editor can do is great cutting, being able to do so collaboratively is going to blow people's minds (plus dmedia is going to blow people's minds in and of itself). And we feel confident if people can play with the technology we've built, we'll be able to generate enough interest to keep moving forward, whether through grants, investment, or even another Kickstarter.
We've risked it all to get this far, and we're asking for a fraction of what our current debt is... just so we get to the point where things start to get *really interesting.*
Can you talk about the upcoming features you intend the software to have and when we can expect them?After we have cutting that's as fast or faster (in person hours) than the competition, we're diving into multicam, something we're extremely passionate about.
Multicam is strategic for low-cost HDSLR productions especially because the biggest cost is people's time on production days. If you have a cast and crew of 20 people, you need to extract as much value as possible from their day on the set. If a shot works from multiple simultaneous camera angles, shoot it that way. Or shoot multiple focal lengths at once from basically the same angle. Or be shooting footage for a ''making of'' documentary. Just try always to be shooting something potentially valuable. You want a production day to yield as much marketable content as possible.
Now the trick to the above is the software needs to have very productive multicam workflow, and needs to automate the mundane stuff whenever possible. No sense in extracting as much value as possible on a production day if there editor burns through so many person hours trying to do multicam edits that it's a net loss.
We have some unique automation opportunities because dmedia gives us full EXIF metadata for videos (extracted from the .THM files), so we know the serial number of the camera a video was shot on. So we know if it came from camera A, camera B, etc.
As far as when to expect any given feature to land, it's hard to say, both because it's difficult to estimate software development time, and because we have a rather unique process.
Our goal is to *measurably* save artists time compared to competing software, and we take this very seriously. So we do our UI work in two stages. First, for a given workflow we'll implement several variations as deliberately throw-away prototypes. This allows us to try bold things, give unexpected solutions a chance. We'll measure how fast the workflow is on each variation, and pick the fastest one to advance onto the second stage, the full implementation on the live editor backend. But a design will only advance onto the second stage if we have something that is measurably as good or better than the competition... tell then, we keep tuning, trying new variations.
By the way, anyone is welcome to join in the development, and the novacut-prototypes project is an especially good place to start. The more workflow variations we build, the higher chance of stumbling across something that really raises the bar.
Here are some more roadmap details.