Cork School of Music: the collaboration grows

After the first session in Belfast, we knew what we needed next: more time, more instruments, and a space where we could start thinking like performers.

In June 2025 we met at Cork School of Music and worked in the Stack Theatre over a long weekend. The group had grown:

• Ioana Petcu-Colan (violin)
• Ross Lyness (trombone)
• Alex Petcu (percussion)
• Lina Andonovska (flute)
• Amy Prendergast (choreographer)

This was the weekend the collaboration stopped feeling like a prototype.

Cork School of Music and the Stack Theatre space (June 2025).

Context

Between Belfast and Cork, the drawing system changed shape.

We’d come away from Belfast with two big needs:

• a more legible shared language (so performers can tell what the system is responding to),
• a richer visual vocabulary (so the drawing can change character when the music does).

We arrived in Cork with a rebuilt setup. The original app had been split into a chain of separate modules, more like a synthesiser than one big program, and we started calling it MarkSynth.

It could run in two broad modes:

• audio reactive,
• audio and video reactive (using a camera feed).

It still drew continuously at high resolution, and we could save snapshots as we worked, which keeps the live performance connected to the physical work.

The session

Day one: testing the system

We spent the first day settling into the space and pushing the system to see what it could do. We started with “everything on”, lots of layers piled up, because that shows the range quickly, even though it’s not where we want to end up.

We also set up basic documentation:

• audio recorded from a good mic,
• video recorded from the camera feed the system was “seeing”,
• snapshots saved whenever something interesting appeared on the projector.

Two things stood out.

First: a wall-sized projection is powerful, but it can also be distracting. When the drawing is large and bright, it competes with listening and with the subtle communication performers are used to relying on.

Second: the screen only shows part of what’s being drawn. When we zoomed into the snapshots, there was a lot of detail we weren’t showing. The drawing rewards close looking, but we hadn’t found a way to show that live yet.

Day two: simplify, then build an arc

With the full group in the room, Amy made a real difference. She kept nudging us from “trying things” towards “making choices”, and we started treating the weekend as the start of designing a performance.

We stopped running everything at once and broke MarkSynth into simpler chains, so we could see cause and effect.

That simplification helped in a few ways:

• performers could more easily associate a musical action with a visual change,
• we could imagine a longer piece where the drawing evolves in chapters,
• we could spot what felt strong and what was just visual noise.

We also removed the bold black foreground lines from the live projection. In physical work those lines (often picked out later in silver leaf) are useful because they hold “the now” against slower layers underneath. In motion they were too dominant, and made the live drawing more agitated than it needed to be.

By the end of the day, the instruments, extended techniques, movement and drawing together started to feel like they could hold a longer piece without falling into chaos.

Overnight we sketched a loose sequence of five configurations, “visual chapters” we could move through to give the piece contrast over time.

Day three: a first long-form run

On the final half day, Amy suggested a simple prompt to give shape to one continuous improvisation: the river.

She outlined five movements (springs, gathering, flow, ecology/texture, sea/return), each with a musical emphasis (percussion, sustained tone, melody, texture, and a final return).

That prompt lined up neatly with our “chapter” idea from the night before, so we combined them and tried a full run with each section lasting a few minutes.

At one point the wireless camera feed failed and we were stuck with the length of a cable. It wasn’t a disaster, but we’ll need a backup plan as the work gets more public.

Even so, the run held together, and it was long enough to feel like a piece. When everything locked in, you could feel it in the room. At one point the drawing visibly responded to Amy’s movement and she blurted out “I exist!”, which is exactly the kind of loop we’re after.

What changed

Shared language

Breaking the system into simpler chains made it easier to read straight away. It didn’t solve everything, but it made it easier to talk in concrete terms about:

• what a performer did,
• what the system registered,
• what the drawing did in response.

The “chapter” approach also helped. Even without a strict score, moving through a planned sequence gave everyone a shared sense of where we were in the arc.

The range of marks

Cork showed us two things:

• the audio-reactive system already had a range, but it benefited from restraint and clearer hierarchy,
• video responsiveness opened up another layer of interaction (especially with movement), but it needs reliable capture and careful tuning so it doesn’t overwhelm everything else.

Removing the live foreground lines made space for other behaviours to read. The open question is how to keep what those lines mean (immediacy) without letting them dominate the projection.

Stability vs responsiveness

The weekend confirmed what we’d felt in Belfast: responding is only useful if the system can also hold steady.

In practice, that meant:

• fewer layers at once,
• slower changes when the music is stable,
• more obvious shifts when performers make clear changes.

What we brought back

We left with a lot of material: audio recordings, video, and a large set of snapshots.

Some of the most useful material wasn’t pretty: just two snapshots side by side, showing that different playing really does make different traces. That’s the whole idea of Performance Fingerprints, and Cork made us more confident it works.

Next steps

• Build a reliable way to run a sequence of configurations (chapters) during performance.
• Find a better live strategy for “immediacy” marks (foreground lines): tone them down, gate them, or make them conditional.
• Improve how we reveal detail in performance (zooming, cropping, or moving through the high-res drawing space deliberately).
• Tighten the video workflow (wireless reliability, backup options, and calibration).
• Assemble a short process/documentary video from Cork material.

Videos