Company Post

Collectif Scale's Astounding "Ballet of Robots"

Meet Collectif Scale, a Paris-based group of artists and creatives from various callings who joined forces in 2010 with a shared passion for creating state-of-the-art installations and performances that have been quite aptly described as the ‘ballets of robots’. We talk to the collective about these futuristic and elegant kinetic light and sound compositions featuring the hypnotic, inconceivably tightly-coordinated choreography of – in the case of Coda 20 robot arms – and how TouchDesigner has optimized their ability to create adaptive and generative precision mechanical choreographies and control them.
Derivative: Please tell us about Collectif Scale - how you would best describe your collective and how you came together, what excites and drives you, and how you chose your projects and work together?

Collectif Scale: In 2021 we will be celebrating our 10th anniversary. Scale is the name of an artistic director behind whom hides several people from very different but complementary professional and artistic backgrounds. Today the Scale team consists of 10 members based around Paris. But Scale is above all the association of a group of passionate friends who just wanted to get out of their usual paths by combining their know-how after hours.

We are not artists with a capital "A", we do not define ourselves individually as artists, although it is claimed that Scale produces art installations. Indeed, none of the members of Scale come from a renown art school. Our educational and professional paths are more technical. With us, the culture of cables and soldering iron dominates. We have always loved video games and rides as much as Bill Viola or Vasarely. But it is by coming together under a single name that we can now claim to defend an intimate artistic language. One of our specificities is that all the members of Scale have their favorite specialty but that all also have at least a second specialty. This allows us to be very complementary and also to be able to cover all the costs of the complete production of an installation without subcontracting.

Whether it is motion design, research, development, mapping, led design, interactivity, robotics, programming, all these technological masteries are therefore at the service of an artistic proposal.  Over the years and creations for stages or for muséo-graphiques environments, we have developed our own language to create sensitive, interactive, playful works, accessible to the greatest number and with a strong attachment to never cease to innovate. We question ourselves on the links that unite music and the visual, light and architectural design, entertainment and contemporary art, nature and the future, man and machine and it is by searching that we try to provide answers synthesized through our installations mixing several mediums.

Those who see our work speak of digital art, contemporary works, or digital scenographies. For our part, we speak more simply of augmented installations, that is to say the ability to formalize an idea so that it becomes a living, dynamic and often tangible visual object.

Derivative: How did you meet TouchDesigner and why did you decide to adopt it?

Collectif Scale: We first heard about TouchDesigner through concerts by the band Nine Inch Nails. We are first and foremost Mac OS users and when TouchDesigner came out for Mac OS we decided to re-adapt our current Ammonite installation which was previously controlled by VDMX (for image generation on a shader basis) and Madmapper (for the led mapping). We wanted to be able to develop a more personalized, versatile and adaptive generative controller. Also, the majority of our installations are controlled by hand by ourselves, especially when we do concerts and live events. It is therefor because we were able to build optimized interfaces completely dedicated to our projects that our interest in TouchDesigner was immediately significant.

Derivative: How do you use TouchDesigner and Madmapper respectively?

Collectif Scale: In general, our recent installations for about 2 years have either been driven by TD only, or by Madmapper only or by TouchDesigner and Madmapper. When we do LED/light mapping, Madmapper is always our Artnet mapping interface. Then our image streams are either generated by TD (this is the case of Flux for example) or by Madmapper, depending on the installations and the principle of live activation. When we do motorization control (CODA, FLUX, Hulahoop and Quercus soon), it is always with TouchDesigner.


Derivative: Could you tell us a bit about the projects where TouchDesigner is heavily featured in terms of your workflow and the management of motor movements. I also understand that Bullet Solver has been very useful?

Collectif Scale: Behind the TouchDesigner development of our robotic, mechanized projects such as Coda, Flux and Quercus the main idea is to be able to work as much in an “offline” manner without the physical installation as in an “online” manner. For this we have created a Master Container in TouchDesigner containing both a 3D preview of each motor module/LED bar/support, and a part for sending data over the network to the Raspberry/Arduino that control our motors. Then via the Replicator Comp (and as many times as necessary) we create instances of this container in 3D space to preview the installation. As in each container the 3D part and the data-sending part are closely linked, we are sure that the 3D preview will correspond to reality.

For Coda, we had to implement additional functions. In fact, Coda is composed of 20 robots with 5 axes of rotation arranged in a grid and a 1.5m RGB LED bar fixed on the last axis of rotation. We wanted to keep a relatively small spacing between each robot but the risk of collisions (collisions between the LED bars, collisions between each LED bar and the robot support) then becomes very high because the robots do not necessarily execute the same movement at the same time. So we decided to slightly hijack the Bullet Solver. A simplified version of the installation is created in 3D in the Bullet Solver Comp thanks to the CHOP Bullet Solver which returns us the information of potential collisions. By pre-testing the “offline” movements, we can therefore know whether there will be collisions in reality or not.


Derivative: The “time delay system” you’ve developed that lets you delay or advance the sending of data is very interesting. Can you explain how this lets you artistically manage and convey interesting physical and organic movements?

Collectif Scale: To begin with we often talk about “gradient of positions” and this was initiated with our Ammonite installation which was however static. We wanted a ripple effect where the position in space of each LED bar to be a slightly modified repetition of the position of the previous LED bar. For Ammonite this allows us to obtain a very balanced overall architectural object, as if a logical and mathematical movement were propagated to describe and imprint in space the positions of each LED bar. We often see in this controlled repetition a fairly organic structure.

This concept of "positions gradient" is therefore present in almost all the installations that have followed Ammonite. We have tried to apply it to kinetic and motorized installations by very simply using this principle of temporal delay. Thus, that we have managed to give the impression that a movement (therefore positions that are constantly evolving) is multiplying. Each child container precisely follows the master container but with a slight time lag. We have therefore developed what is called a "delay generator" which allows us to draw and define the characteristics of the propagation of movement from one element to another. So, from a fairly simple main movement we get to organic and complex shapes quite easily.


Derivative: Have you encountered any situations where there were some surprises in how you solved a problem or came to any kind of “new” realization.

Collectif Scale: With TouchDesigner in view of the range of possibilities we are in constant learning. So yes, we are finding new ways to solve problems all the time. For example, in our first projects everything was almost exclusively based on CHOPs and TOPs. It worked all the same but it was not very optimized. Then we got more and more interested in Python in TD, and there the realm of possibilities opened up to us! What we love about TouchDesigner is that there are a thousand ways to approach a problem, but you can always dig deeper and find a smarter way of doing things.

Derivative: User interface design and function is also very important and you have some excellent examples. Could expand on how you build these with the two-part thinking – the “edit” and “play” aspects? And as a side note, are you always on hand to run your shows? I would imagine so given the assets involved but imagine also with the UIs you’ve designed that it would be possible for an operator who does not know TouchDesigner to effectively run things?

Collectif Scale: Our projects are often divided into two distinct parts, an "EDIT" part and a "PLAY" part. They are developed in a relatively similar way but with a few differences:

  • The "EDIT" part has a few additional modules allowing us to program our movements, write each of the positions that make it up, manage associations of motors, assign respective delays, etc.
  • The "EDIT" part does not include the management of sending data to the hardware because it is intended to be used exclusively "offline".
  • The "PLAY" part integrates these data transmissions in addition to a "preset/cue" system which allows us to trigger and organize movements according to pages.
  • The "PLAY" part also has some additional locks so that the live game with the engines goes correctly. For example, a script prevents data from being sent if the motor has not finished its previous move. Another example, the "PLAY" part blocks us from modifying the movements to avoid generating collisions by mistake.

For the construction and definition of these interfaces we do a lot of work on the user experience (UX) so that the project is as easily as possible controllable. Even if the operation of our installations is very dedicated and appears to be complex, we try to optimize as much as possible so that the use of the systems (EDIT + PLAY) is as intuitive as possible and that the complexity becomes transparent. We hate the idea that an interface because it is too complicated prevents us from being so versatile or responsive. Remember that the majority of our projects are performed and directed live by an operator. It is therefore also necessary that our interfaces can be very easily understood by someone who does not necessarily know how TD works. From our point of view, the more the interface is clear and optimized, the more the artistic intent will be apparent.

Derivative: Are there any new or emerging technologies right now that excite you? And any thoughts on what you might like to explore next?

Collectif Scale: We are very significantly in a period where we want to deepen our exploration of kinetics, physical movement and robotics. For the past ten years and until the Hula Hoop project we envisioned two weeks ago, light has been our main medium. Motorization has always been at the service of light. With Hula Hoop we want to use physical movement as the main and unique medium. We still have a lot to explore in this area.

The current period dictated by the health crisis is complicated for the presentation of projects in public. So today musicians are doing concerts on the networks. These are certainly not real concerts with all the social warmth that it brings, but these streaming concerts allow very distant audiences to listen and see the artists perform. This is a real novelty and maybe this model will stay.

It also allows us to approach things differently. Today for example we would like to produce an installation that is situated somewhere in the world with no audience present at that location but which could be seen and controlled remotely by anyone in the world 24 hours a day. The future may also take us towards less scenic projects.


Follow Collectif Scale and on Instagram