A pair of pupils with good control on two axes, a decent blink and maybe a brow move or two are, in many situations, the most expedient way of getting a character to appear that it is a thinking, self-aware being. Guests tend to forgive quite a bit when eyes behave the way they are expected to.

It's a bit of a cheap trick, playing on the fact that we focus more of our attention on what the eyes are doing than anything else, but a trick that works is valuable, no matter how cheap.

The advantage of naturalistic anatomy is that it, by definition, has to follow physics. Human eye muscles actually don't offer that much variation in shape-change: Lids mostly work without lots of angular stretching; brows can do a few things but they are fairly constrained as well; eyes dart, but only on two axes and are always spherical (well, primarily spherical). However, we put so much time into watching faces and developing sensitivity to very subtle variations and combinations that the moment accuracy isn't spot-on, including minor changes in adjacent muscles and tiny micro-expressions, lots of negative cues get thrown.

Fundamental skin and collision/friction issues that directly impact reliability quickly create disadvantages that feel like they totally outweigh the advantages. That is, until you try and work with cartoon eyes...

It doesn't take much scrutiny for the weaknesses of an animatronic figure to become apparent. C3PO is fortunate to be a character defined by a human pretending to be a stiff robot with limited range-of-motion. Still, his control panel is in full view of Guests and he can't really turn to it or actually touch. The animator is left to just bat C3P0's left hand near it once in a while. By himself, C3P0 can't really stay convincing for very long at all, but fortunately there is a lot going on around him.

He works pretty well in the overall scene, except for the minute or so Guests are right in front of him. If making him a stronger presence was ever needed, his staging and function choices would have to be modified and there are many different ways that could be approached and still stay true to the IP.

This same sort of calculus is done for every figure as its staging and performance are worked through. How long does the character need to stay alive and what are the options for making that happen?

The designers who laid out the Star Tours pre-show did a good job of creating a space where there is a lot to look at. Quite a bit of storytelling goes on, timing carefully controlled so that the scene works as a whole. The elements aren't particularly complex, with C3PO as probably the biggest operational and presentational risk. Guests move through the space without thinking about that too much because there is so much eye-candy to keep them engaged.

The choreography of multiple show elements in different areas, well-sourced audio/video, use of different visual planes and strong thematic decisions all combine to give this attraction an amazing feeling of immersion.
[Note: I've animated many of the figures in various versions of Star Tours but played no part in the overall show design.]

Trying to get a presentation edge - add spark - to a new attraction often means working to implement new tech. In some cases that means tailoring the story to fit the technology. For Alien Encounter, the stiffness and physical constraint of the seats were elements required to make the binaural sound work. Adding in the thematic pastiche of an uncaring galactic corporation helped provide cover for the discomort of Guests of being held firmly in place.

There are so many constraints involved with animatronic design that you really want the biggest bag of tricks you can get. For example, I appreciate the value of electric actuation, but when that tech is specified and hydraulics aren't on the table there are fewer design options, often limiting some aspects of performance. When you run out of tricks on the technical side, the only remaining option in most cases is to modify creative intent - change the character design, performance and/or story to find a workable solution.

Some technical advancements can be made during design and production, but there's a natural limit to how many new miracles can be pulled off - it's always good to have a wide variety of proven solutions to pull from.

Over the last 50 years, Guests have naturally developed an understanding of what animatronics are and aren't: legs bolted to the ground, stiff lower-bodies, fair medium and large movement but limited in both range and speed, clacking eyes, poor resemblance to the character as portrayed in other media, etc. - often appearing broken. For any animator, it's a difficult list to accept, but it also can't be wished away. Sometimes it's assumed that those of us who have been working with this medium just can't see those things. I can assure you that isn't so. No one is more aware of or wants to break out of the animatronic vocabulary more than an animatronics animator.

The trick is that those elements occur for very real, if unfortunate, reasons. They can be overcome, but only with both an understanding of why they occur and a lot of focused problem-solving. Since we are working with machines that are required to perform dependably and safely for 5000+ hours a year, progress in overcoming these obstacles tends to be painfully slow.

In the decades I've been working in this field there have only been a few truly big leaps, and those tend to carry significant caveats and counter-issues:
- Electric Actuation: This has cleaned up animatronics by reducing reliance on the messy use of hydraulics and increases potential speed and range-of-motion in many situations. Unfortunately, using electric motors also dramatically reduces power-per-inch which impacts many aspects of design and has other drawbacks (see Tidbit #8 for more).
- Skin Systems: Like actuation systems, the advances in skins are pretty moderate, and depending on which skin experts you talk to, the best methods can vary dramatically, but I can attest to the fact that we have more robust skin systems available than we did, say, 20 years ago.
- Simulation & Digital Design: Although digital design also applies to control system advancements, it is now so integral to mechanical engineering that it deserves to be listed twice. If we are careful, we have the ability to dramatically increase our confidence that our figure design decisions will support the specific target performance - represented in a CG rig and animation file.

Determining what Guests will see, and maybe more importantly, how they will perceive it is one of the baseline bits of information needed to make good design decisions. We have to make many choices in the design process and some inevitably will be in the form of concessions to balance various requirements. Doing so effectively requires knowing where the sweet-spots and boundaries are from the Guests' POV.

Cultural factors matter too. I don't spec or animate figures quite the same way for Japanese or Chinese audiences as for Americans or Western Europeans. Some Western gestural patterns are off-putting to other cultures and can weaken how a performance is received depending on the context of the story.

When they're being jostled around in a ride vehicle at rollercoaster speeds, Guests' ability to process visual information is impaired, so while everything still needs to look good, it has to 'read' fast - the storytelling has to be clear and very efficient.

It's a good idea, in general, to break down performance design in terms of specific beats, defined by timing and content, and that is a particularly critical task in a fast-moving ride where you might only be able to put over a couple beats before Guests careen toward the next scene.

Quick example: If a script calls for a monster to do a roar and a swipe as the vehicle flies by, it'll be good to be sure that there is time for both. A swipe - usually a large, multi-axis arm motion - may involve high complexity/cost. If the roar is most important, can that carry the scene? And if the swipe is most important, maybe the roar is heard more than seen, or perhaps the ride needs to slow down a bit more to sell the whole scene. Those are things the team needs to work out way before steel gets cut.

Marc was keenly aware of the limitations of animatronic figures, which were even more restrictive in the 60's and 70's. He often tried to increase audience engagement by giving the impression that something seriously dynamic was happening or was likely to happen in the coming moments - drunk pirates balancing precariously on kegs of dynamite is one example. Once 'precariousness' was established, it only took a little bit of well-placed motion to keep selling the idea and keep the scene alive. Working through these types of solutions can help with both pacing and budget management - saving the big, expensive moments for highpoints of both story and Guest experience.

If you can communicate a page of treatment text into a single image, you're doing pretty good. The image shown here was for a game demo, not a theme park attraction, but concievably it could work for an attraction. It gets a fairly complicated concept across quickly and visually. This is also a good example of using non-physical elements to aid performance. A well-placed special effect can be worth its weight in gold.

The big magic trick required of a modern animatronic pipeline is maintaining a performance sample - a CG rig and animation - that accurately reflects creative intent at the outset, continuously represents the evolving state of figure design and performance capability throughout design and production, and then smoothly becomes the show data that drives the practical figure through installation and turnover. For those who don't actually have to make all that happen it might seem like a straight-forward process, but if the messy undercurrents of physics, gravity and figure finishing aren't respected and properly managed, unpleasant surprises are guaranteed.

Having an integrated editing toolset helps make it possible for an animatronics animator to keep Creative and Engineering on the same page, even through the transition from the theoretical to the real-world.

These days the design process for high-end figures includes significant mechanical simulation rigging and data analysis tools. This is a critical advance, allowing us to do things we couldn't have considered just a few years ago, enabling us to take full advantage of electric servos and other emerging technologies, and to better assess risk early in the process.

A common misconception, however, is that the simulation will accurately translate directly to a precise replication of the CG performance on the physical figure by just converting the data into a format usable by the control system. There are a handful of real-world factors that get in the way, including:

- Motor Response: Response curves vary based on the size of the motor and they often require data timing adjustments. Most systems don't account for the timing issues this creates, which can be significant.

- Mechanical Design Anomalies: Human and animal figure designs get complex, and sometimes unexpected things happen, even with the best of digital design and analysis efforts. These can introduce abberant or unexpected motion - little tics - that have to be managed by the animator.

- Figure Finishing: The costume, skin and shell elements introduce variables that can affect motor performance, range of motion and secondary issues that affect visual performance. These cannot be effectively modeled in the actuator-space rig, or at least not within the scope of most projects.

- Animator Input: Pushing the animation requires that the animator assess the specific, detailed strengths and weaknesses of the figure. Making subtle, or sometimes major, structural changes to the performance based on a nitty-gritty evaluation of what works best on the physical figure is often necessary.

- Field Discoveries: Issues tend to be discovered during Installation that weren't anticipated in Concept and have to be resolved. And there are also often performance opportunities that weren't anticipate that can improve show quality.

To get the full level of performance value out of a figure, an animator still needs to be actively involved with figure setup and then tailor the performance data to find the right balance of show quality versus maintainability in cooperation with the engineering team.

Electric actuation has recently come into its own as the darling of theme park animatronic design, with its clean operation and some major advancements in control and, for some solutions, in packaging. It's a very valuable tool in the toolbox, providing new performance opportunities in terms of range of motion, positional accuracy, and in some cases, speed - as well as the potential for getting figures closer and more responsive to Guests.

There's no denying its advantages, but new challenges have come along with the shift from hydraulics to electric motors:

- Power vs. Packaging: Getting actuators to fit into very specific organic shapes is a huge design issue, and hydraulic actuators has roughly a fourfold advantage in many situations. Some areas, like hands, tend to get impacted more than others.

- Weight Budget: As power requirements increase as more weight is added to the figure, larger motors are needed. When electric motors are overtaxed, they tend to overheat and fail. In addition to hydraulic actuators being able to handle more load in scale, they tend to not move rather than break under pressure. This creates a situation where there is greater design dependency between assemblies and less design flexibility as issues arise.

- Setup & Control: Setting up electic servos takes significant control engineering skill and tends to push the artist out of the process. Getting the look and feel of each function right and all functions to work together aesthetically requires artist input and can require adjustments be made throughout the final performance editing process. Doing effective setup on electrics has essentially changed from being a cooperative task between artist and technician to more of a tennis match. When schedules and budgets get tight, as they always do at the end of the project cycle, this can impact show quality.

One of the things that sets animatronic editing apart from CG animation editing is the thing the animator is analyzing is not on the screen. The animator has to view the figure, scene and attraction the way the Guest will, and adjust movement (bigger, smaller, faster, slower... different) based on what Guests will actually see, making the UI workflow relationships notably different from what is common for film and game production. Hopefully the digital design effort will have maintained the performance to be fairly close and the adjustments will be minor, but I can tell you from first-hand experience, it's not heathly to bank on that. Things change, and other things just don't come out the way they were expected to, from figure finishing and skin tears/wrinkles to sightline, ride, lighting, FX surprises, to decisions that some functions have to be disabled and locked off. The final push to deliver a show demands skill, flexibility and efficiency.

One of the first tasks I had when with WDI in the early 90's was to update the UI for the editing software/hardware in conjunction with others in the group. Over the following 5 years I and the other animators and show programmers worked with those tools daily and came to understand its strenghts and weaknesses. In the mid-90's we had the chance to re-design and upgrade the entire system and implemented features to address deficiencies, such as including curve editor-based keyframe animation. I've now had roughly the same professional time spent with Maya editing and those proprietary real-time tools, and see value in integrating both editing environments for pipelines I help build in the future.

- Pre-vis Editing & Digital Design: Performance editing done prior to assembly of the figure.

- File Transfer Figure Editing: Converting data generated on a CG rig into a format that can be imported into the figure control system. This process implies that changes made will only have visual feedback from the CG rig. Notes from animatronic playback figure have to be video'd or otherwise observed without a direct editing loop. Any significant differences that exist between the CG and animatronic playback will result in a very iterative, slow and painstaking process.

- Real-time Playback Linked to Editing: 'Real-time' is not a great term, as it means different things in different contexts, particularly when used in relation to control system design. The biggest efficiency, most needed by the animator, is to be able to play a specific segment of a performance (not necessarily from beginning to end) quickly and easily. Then, just as quickly and easily, edit the performance data and immediately replay it. The faster this basic editing loop is, the more efficient the animator will be, as they are likely to do thousands of edits in any given session, many of which will be undone or written over as solutions to problems are experimented with, thrown away, or pursued. If the animator can achieve and maintain deep-focus for long periods of time, with an efficient workflow, high production speeds are possible.

- Real-time Input Editing: This might be a waldo replicating an arm & hand, and/or a few rotary knobs/sliders and/or some for of mocap wearable for generating and capturing compound motion on multiple axis. There are many ways to approach this class of editing, but the salient criteria for me is that it permit the animator to work with a more performance-based workflow, generating organic motion with their own body rather than through mouse drags. Having worked with Henson's Muppet team I recognize the qualitative value of solving performance problems with less deconstruction and more spontaniety. The trick is to make that available without slowing down the rest of the animation process - it has to be quick and easy to accomplish and something the animator can do alone.

- Figure Setup and the Show Environment: The editing won't be of any use if the response of the figure, or speed of the vehicles, or placement of set pieces or lighting changes dramatically. And the more those elements deviate from pre-vis expectation, the more work there will be to do. As a consequence, the animator also often ends up being part coordinator to ensure that all of the pieces are in place so the work they do makes progress toward turnover.

We're creating an illusion, but instead of mirrors and light we're working with human behavior and human communication norms - in many cases, much like a street performer does. Here are just a few concepts I like to keep in mind:

Hide, Don't Show, Technology - We want the audience to be focused on the character and the story, not how cool our toys are. It's better that the audience not notice we are doing something extraordinary than to be thinking about it. If you tell Guests they are in an "interactive experience," their one and only focus will be to break it. Simulating natural responsiveness is a form of magic that is most successful when we sneak up rather than hit them over the head.

Have Escape Strategies to Get "Back to Simple" - It's easy to get caught up in trying to chase the complexity of human communication. Avoid that chase. Your show can be successful even if it's simple, as long as it's solid. Have a way to break focus and reset when things seem likely to go south. Street performers get really good at this.

Guests Seen vs. Heard - Certain types of automated orientation and eye-contact can be pretty effective and aren't very complicated to achieve, but once you attempt to do more than just look at the audience and/or use "one-way" throw-away comments, there's really no substitute for a human brain - an operator/performer. A number of things jump dramatically in design complexity when that step is taken, like UI efficiency, exponential content needed, smaller segments and more complex transition strategies - things get messy fast at that point so...

Playtest Early - What kills most interactive experiences, in addition the fallout from not following some of the notes above, is underestimating the importance of timing/pacing of the illusion. Every millisecond it takes for a figure to respond beyond the time it would take a real person is 10% off of your tip for good interactive design service. It's gotta feel fluid to feel responsive. How do you do that when you have segments that are mostly 5-10 seconds long and are triggered by an operator who has been pressing buttons for 6 hours using a multi-page touchscreen? That's another level of design. Give me a call.

Most of the interactive shows done in the last couple decades have focused mainly on scripting - selecting a large number of dialog lines and connecting them together with fairly simple branching logic. Between any two sentences there is - always - a common blank "space". However, a figure has to be in a pose, and to use a common pose for everything is painfully restrictive - and problematic for timing/pacing. Doing so requires that there be enough "blank space" between lines of dialog to get to the common pose and then back to the next first move, which also tends to look awful. This is often realized way late in the process, sometimes patched with some additional transition complexity and compromises to show quality.

Ultimately, responsiveness drives segments to either be extremely short and/or have a level of arbitrary (or semi-arbitrary) 'jump-out' capability and a robust transition strategy that can handle getting from any out point to any in point with presentational grace and appropriate speed.

More effective control structures are certainly possible, but have been historically hobbled by issues surrounding the limitations of the data structure. I did some initial work in this arena to add transition-specific information to the standard performance data format a couple decades ago wit some success. To make big strides forward, however, I believe animation needs a new data type - one that robustly defines transitions and embeds performance meaning in the structure itself - I'm working on that presently.

Animatronics for film only have to work long enough to get one good take, but for theme parks everything has to work reliably all day, every day, for many years. And we have to keep everything safe - Guests, employees and the equipment. This creates heavy-duty design requirements that dramatically change the game.

Engineers will do the math to work out what that means, but the Creative team still has to set an achievable target. Navigating through evolving theme park operational requirements and knowing what has worked (and what hasn't) in the past is a key part of the process.

I've always winced a bit at Marty Sklar's famous "blank sheet of paper" paradigm. It's not usually advisable to just toss idea, no matter how amazing, over the wall to Design & Production. It's an inspirational approach, but functionally the animatronic design process is just messier than that, even in Blue-Sky.

Mainly it's "the wall" that I take exception to. If you set a target that is way out of the realm of possibility, a production team will often accept the challenge and do what they can to achieve it, but the difference between expectation and delivery will be significant, and usually disappointing. Basically, if you aim too high, you tend to fall much farther than if you set a near-achievable target.

That doesn't mean that a design target shouldn't be a challenge, just that feasibility is also an important part of the process and it's vital that the Creative and Technical teams work cooperatively throughout the project cycle - beginning, middle and end - to have some confidence that you can deliver on promises to stakeholders.