April 10, 2008

Open source video editing: what we have now and what we need

Author: Rui Lopes

Watching the evolution of open source tools for video editing
and manipulation over the last 10 years has been less than a thrilling
experience. But are things about to change for the better in the near
future? Can even the people most disenchanted with the current state of
affairs feel tempted to regain a spark of hope?

The problem

The everyday user of GNU/Linux (or any of the other Unix-derived
operating systems out there) can presently watch, listen to, and even
produce rudimentary media content quite comfortably, but open source has
not kept up with proprietary software vendors when it comes to the
development of professional and semi-professional video editing tools.

I've always maintained that one of the great strengths of Linux is the
ability to create distributions that are tailored for specific areas of
professional activity. But, while there are some good distros around
that appeal specifically to the video and audio crowd (like dyne:bolic and Ubuntu Studio), their number is
small, and some of them have stagnated or even completely disappeared
over the years.

One of the roots of this problem is that there is currently no
professional level non-linear video editor for Linux. Adding to this,
almost all of the video editors currently available have development
rates that are alarmingly slow. Let's take a look at the best choices
we have right now:

The editors, and what to expect of them

There are video editors that I could never compile, or whose
latest release is more than five years old. Among them are: Vivia, ZS4, and Scilab
Aurora
. I won't look into them. Avidemux can perform basic
editing operations, but that's not its main function.

Let's start with the most basic video editors. First on the list
is Kino. Kino is a very simple
tool, but is the most stable and robust of all the Linux-based video
editors I have tested. It has never crashed on me. For the casual user,
Kino may be the only place they need to look to satisfy their editing
needs. It allows capture from a camcorder, basic editing, adding
transitions and effects, and exporting to other formats. Unfortunately,
it's too limited for experienced video editors. It lacks essential
features such as multiple video and audio tracks, advanced titling, and
compositing. These will probably not be added in the future, as Kino
seems to have a clear idea of where it belongs, and is sticking to its
niche. I currently use Kino to capture raw footage, and often for
exporting video to various formats -- after I edit my video with
another program.

Moving beyond Kino, we have editors that support multiple video
and audio itracks, and seem to have some hope of gaining additional
utility (although not much). First is Kdenlive, which seems to be a promising
project. Or rather, it seemed. The truth is, it's being developed too
slowly, and it's still in a stage where it crashes too frequently on too
many systems. Then we have Pitivi, which is still
too young to be correctly assessed, but which will certainly occupy the
same niche as Kino and Kdenlive, making it almost non-relevant for video
professionals.

I've used Open Movie
Editor
enough times to be able to say it's a solid application with
a nice set of features. These two positive points are enough to make me
keep an eye on its development. Then we have LiVES, which shares these
positive traits as well, and seems to be updated very regularly as of
late, even trying to make the life of developers easier with a plugin
builder
bundled. Both of them seem to be good bets for the more
advanced hobbyist at the present, and perhaps for the video professional
in the future (if they can add compositing and a broader range of
plugins and effects to their feature list).

MainActor is the only commercial application on this list. It is
no longer available. It was no Avid Xpress or Final Cut,
but its mere existence was encouraging. There is currently a petition aimed
at the company that
developed MainActor asking them to release the code (or parts of it) to
the public.

Jahshaka is one of the most
hyped pieces of open source software I've encountered. The vision of the
developers was (and is) an application designed from scratch to be an
editing and compositing powerhouse. The hype promoted on their website
was so effective that for many years I (and many others) eagerly awaited
a stable release with all the industry-standard features that I'd been
promised. At some point, Jahshaka even had corporate sponsors, including
Nvidia. But that turned into a mess which would take too much space to
present here. The good news is that, on January of this year, according
to the developers, Jahshaka was "liberated from its evil benefactors,"
which means that they can now resume their original path. And I have to
admit: I know I shouldn't, but I still have great expectations for
Jahshaka.

I use Cinelerra as my main
application for video editing and compositing. It is professional and
capable in some areas, and barely at alpha stage in others. First
released by an anonymous entity/person called Heroine
Virtual
, for years it remained a paradox: it was the most powerful
FOSS video editor and compositor available, but was also notoriously
unstable, had a hideous interface, and its development pace was glacial.
A community version of Cinelerra
appeared later, which was basically a patched and bug-fixed version of
the original Cinelerra, done by other developers (with the consent of
Heroine Virtual). I'm using the latest community version of Cinelerra,
and I can say it's now stable enough for almost all production purposes.

Cinelerra's learning curve may be a little steep for beginners,
but it's worth it (and the community provides good documentation). It has some
features that you can only otherwise find in commercial applications,
including video and audio multitracking, some decent bundled effects
(among them LADSPA audio plugins),
a good compositor, three-point editing, and motion tracking. Cinelerra
also supports renderfarming
natively
, which puts it on a class of its own. With Cinelerra it's
dead easy to set up a render farm with five or six nodes, and watch your
render times take a significant decrease.

Complaints include: lack of more pleasant themes, lack of
advanced titling, lack of more video effects (and poor documentation for
the ones included by default), and better capture and render functionality.

Finally, we come to Blender. Outside of pure video
editing applications, I would like to point it out as an example to
follow. It is feature-rich, powerful, has lots of manpower working on
it, and tutorials and videos about it abound on the Internet. And,
surprisingly, it allows quite reliable video
editing and compositing
. However, as good an all-rounder Blender may
be, we still need -- and I cannot stress this enough -- good standalone
non-linear video editors.

A look ahead

To say this to a video professional would have seemed like a bad joke
ten years ago, but Linux has turned into a good platform for multimedia
creation and manipulation. What's lacking is the tools. We need more
powerful, more feature-rich tools. Even Hollywood seems to be using
Linux these days (and not just for rendering tasks), but the studios
use their own in-house applications. I'm not the only video professional
using FOSS who is disgruntled with the state of things.

Let's face it: most video professionals who use Linux in conjunction
with FOSS tools as their main platform for video editing do it because of:

  • Cost
  • Ability to customize raw performance
  • Affection/Ethics

Note that neither richness of features nor the presence of industry
standard features are on that list. There are also two important
external factors that I want to point out and that should invite
reflection:

  • the apathy of video software companies towards FOSS
  • the lack of developers for FOSS video editing and
    similar applications

I want to finish by mentioning compositing and special effects
again. These are areas that have been sorely overlooked. I'm not asking
for Adobe After Effects for Linux, rather something on a smaller and
more practical scale with FXhome being
the perfect example of the type of applications we need to fill the gap.
Regarding compositing, Cinelerra is the only FOSS application that
currently allows me to produce decent results (excluding Blender).

My main hopes for the near future are on Jahshaka and Cinelerra,
thanks to the recent news I already mentioned. The key, I think, is
speed of development, and decent scheduling of major releases. Do not be
mistaken -- it will take years to develop high-end open source video
software, and the video professional cannot wait years. However, this
area finally seems to be heading in a better direction than it was even
a few years ago, and that may allow for a bit of optimism on our part.

Categories:

  • Commentary
  • Linux
  • Graphics & Multimedia
Click Here!