I’ve been seeing a lot of questions/debate on a couple of the boards and Twitter about which compositing tools are best to use, most effective, help build strong bodies twelve ways, etc. There is always the question of what tools are the best for a particular type of work, and, for compositors looking to work in the industry, what film and video companies are using and what skillset they are likely to hire for.
The answers are many and diverse, and offer even more questions …
Are you looking to do real-time compositing, or will batch rendering do it for you?
Do you want to work in feature film or in video (this is becoming less relevant as the distinction between the two gets blurred.)
I thought I’d do a roundup of some of the choices available in the comp tools area, with some thoughts about features, price, and which I’d use for a particular set of circumstances. But here’s the deal: As I mention in the title, people get wildly fanatical about the compositing software of their choice. Nothing I’m saying here is being put forth as definitive, or authoritative, just my POV, ‘kay? I’d like it nothing more if other VFX peeps chimed in with their point/counterpoints to get a real dialog going.
Before the round-up, let me state my own experience/biases, as the way you come to compositing has a lot to do with your preferences later on.
The first comp package I used was Eyeon’s Fusion, back around v2.0. I got a student rate on it when I was at NYU’s CADA (Center for Advanced Digital Applications) studying VFX back in ’99. It was a node-based tool that we were using for doing multi-pass composites from AOVs (different lighting and component passes) generated in Maya at the time. From there I went on to a Flint/Flame class and picked up a Combustion license to try an leverage some of those skills. I never really loved the Discreet “Artist Interface” (and just about nobody I know in the business does), and really took to Shake when I started working at Warner Bros., for a number of reasons including the intuitive, node-based, interface.
That said, I’ve been working, almost exclusively, with Shake for the last five years, and have recently started transitioning over to Nuke. That makes sense for me, as I live in the world of high-end, feature film, VFX. There are other choices which make equally as much sense for compositors doing different types of work.
Adobe After Effects
First off, After Effects, or AE, is the one tool I’m writing about here that I’ve never used and don’t know. Again, this has more to do with just what I started out with and momentum than anything else.
After Effects is probably the most widely-used Compositing and Motion Graphics package around. It provides Swiss Army Knife-like capabilities to address a whole host of tools to address just about every situation. There are a wide variety of 3rd-party plugins available and, best of all, AE is, by far, the least expensive of any professional compositing package.
AE’s strength is probably also its weakness, in that any tool that is as broad in scope is liable to suffer in depth in some respects but, again, that’s where plugins come in.
If I were starting down the compositing road, looking to get the most bang for my buck, and learn a tool that is used in a lot of boutique VFX shops, particularly for TV and Ad work, I’d pick up a copy of AE. In fact, I am going to pick up a copy of AE as it is the major hole in my arsenal in terms of compositing tool knowledge.
For some reason software companies seem to gravitate toward cataclysmic-type names for comp software … Discreet, now Auto Desk, named their line after progressively scaled fire terminology. Combustion/Flint/Flame/Inferno/Smoke, etc. Most of the higher-level tools, i.e. Flame, Inferno and Smoke (which is a non-linear editor with massive compositing abilities) are designed for real-time work and require very pricey, dedicated hardware. The software originally only ran on SGI computing hardware, but over the years Autodesk has migrated towards more generic Linux-based platforms which have become the Industry standard.
You see Inferno and Smoke used for a lot of high-end TV work, as well as feature films, but the entry point (somewhere around the mid 5-figures) is pretty much out of the ballpark for most home studios (not to mention the tech support required, unless you’re also an -IX-type system administrator in your spare time.) This is where Combustion comes in.
Combustion started out somewhere around 2000 as a Windows-based, down scaled, version of the Flint/Flame/Inferno line. It’s interface is very similar, although not identical to its big brothers in the Autodesk product line. It was really touch-and-go for a number of years as to whether or not Combustion would survive as a product … It started out being a $3,500/seat license (and I was an early adopter … Ouch!) and dropped in price to under $1,000 a few years later. Now, in its latest incarnation, Combustion 2008, it is feature-rich and offers much of the functionality of its high-end siblings, albeit in a much more batch-processing oriented way.
Just a rundown of the product specs can give you an idea of the power of Combustion. It’s always had the great paint tools of the Discreet line, but the addition of the Diamond Keyer and the Mesh Warper are big gains. The particle library can be very useful, and it tends to hold up better, and be easier to work with, than trying to work with Motion’s particle engine.
My biggest concerns about the product at this point is the user interface (which, face it, is what it is and not likely to change at this point) and the way Combustion handles playback in memory. While the other Autodesk products use a hardware-based frame store for high-speed playback, combustion has always been kind of clunky, and it’s render- to RAM preview functionality a bit of a pain to use. It also seems to still be a strictly 32bit application which, like Shake, limits you to addressing only 2Gb of memory. A limit that seems more archaic by the day. Also, there aren’t a whole lot of 3rd-party plugins for Combustion, which may pose a problem for some artists who have their favorite set of tools they can’t live without.
I stopped upgrading at Combustion v3 and have been using Shake for most of my projects, including my recent work on “The Guild“. Combustion 2008 is on my list for software upgrades, and I’m seriously thinking about using it, at the very least, for paint work on my next show.
As far as use goes, a good number of boutique TV and film FX houses use Combustion, but it’s still a niche market. Aside from learning it to use, having a good understanding of Combustion may help you leverage yourself into a place, job-wise, where you might be able to learn Inferno. It wouldn’t be my first choice to learn strictly on the basis of having a skillset that was employer-friendly.
Not a fan.
There’s a lot of promise to Motion 3. Promise that it never delivers upon. Small, motion graphics, projects aside, once you get started on a project, you get to a point where you’re generally committed before realizing that Motion just isn’t going to get it done for you.
I recently used Motion for smoke particle sims that needed to be animated along a motion path (Something I would ostensibly think a package called “Motion” would be able to do.) It was an exercise in futility that lead me to start learning Houdini for 3D-CG and FX simulation work … The tool is infinitely more complex than Motion, but also infinitely more capable.
Apple did take the core optical engine from Shake and build it into Motion … If you want to do things like motion-ramping or frame blending in your Final Cut projects, taking them into Motion will result in much better image output.
I’m waiting to see what Apple has in store for NAB 2009 before taking a serious dump on their ProApps strategy. All of their tools are due for a major overhaul and they have not yet delivered on a promise for a Shake replacement, Motion-based or otherwise. At this point I have serious doubts about Apple’s commitment to the Video/Film/Entertainment industry and am loath to adopt, or make a serious investment of time, in their newer tools until something happens to convince me that they are going to remain a vital, viable force for digital visual professionals.
Shake has pretty much been the industry standard for the last 5-7 years. It has a history that started in the ’90s as a command-line tool from a company called Nothing Real that was later bought out by Apple. In its early days as a commercial product, Shake came with a low 5-figure price tag that settled down to about $3,500/seat when Apple took them over. Most of the major effects movies you’ve seen over the last five years have been, and many continue to be, comped in Shake.
Shake has a node-based interface, similar to Fusion, and an optical engine that renders amazing images. It’s pretty full-featured in that it gives you tools for Comp, paint, roto, tracking and color-correction and some pretty nice warping tools. Shake 4 also added a 3D mode that’s useful for layering images in perspective.
Shake comes with a very robust set of nodes, but there are also some great 3rd-party add-ins available, such as Furnace from The Foundry (very pricey, but great stuff. Their De-Noise node is a major butt-saver when working in HD) and Sapphire from GenArts.
One of the things that is insanely great about Shake, if you’re working in a facility, is the command-line functionality. Owing to its roots as a command-line tool, Shake comes with a full set of syntax that makes it very easy to integrate into an automated workflow using languages like Python or PERL. This has come in real handy in situations where I needed to set up rendering, dailies, or quicktime pipelines quickly and on a budget. It’s also a great way to impress your colleagues by typing a bunch of incantations into the command line and having a quick, annotated, comp pop up out of nowhere (comes under the heading of “Stupid Geek Tricks” 🙂 .)
Shake is getting a bit long in the tooth, however.
Apple decided, in their infinite wisdom, to end-of-life Shake in 2006; Dropping support for the product and lowering the price to a very reasonable $499/copy. (I immediately bought one.) Many companies opted to buy the source code from Apple to continue developing the product for in-house use. One of these companies, Double Negative, is using an internally-developed 64bit version of the product. For most of us using it at home studios or smaller shops, that just means that 4.1 is what we’ve got to work with and that means no feature enhancements, a 32bit memory limitation of 2Gb of addressable memory, and the knowledge that, at some point, a version of OS X will come along that 4.1 will no longer live on (and probably that a lot of us will have a separate machine just for running Shake as the future starts to pass us by.)
Right now, though, if you’re interested in doing feature-type film work, there is still a strong demand for Shake compositors in the market place. Moreover, an investment in learning Shake is also a good investment for future employment prospects, as you will be able to leverage your skills in learning the next big thing in compositing packages … Nuke.
Nuke has been around for a few years, but is really picking up momentum. Developed by the Foundry, a company owned by Digital Domain, a big player in the VFX business in Venice, CA., Nuke has started to replace Shake in the toolboxes of many large-scale VFX facilities.
To a large extent, Nuke has taken the core features and functionality of Shake and brought them to next-generation levels. Any compositor who has worked in Shake will find that they can adapt to Nuke very quickly, as the user interface and major modes mirror Shake to a significant degree.
Nuke also has a programmable interface, lending it to large-scale pipeline work, and a list of features too long to go into.
If you’re looking to do feature film work, Nuke is the tool you want to learn. That said, it’s got a price tage of between $3,500 and $12,000 depending on the features/add-in you choose and is probably a bit out of range for the home studio. Again, starting with Shake is a good route if you want to get here.
I started out this article mentioning Fusion, and should just touch on it briefly. I see Fusion demos every year at NAB and SIGGRAPH, and the feature set continues to develop. You do see help-wanted ads for people with Fusion skills every once in a while, but I wouldn’t call this a mainstream tool.
Beyond that, I haven’t used it since version 2.0, and would welcome comments from readers who are more familiar with current versions of the product.