Choose what the image should reveal
An audiovisual patch becomes interesting when the image makes a musical relationship visible. A repeated shape can reveal a rhythm, a slow change in scale can expose a pressure swell, and a scene transition can articulate a section that the sound only implies. Begin with one of those jobs. If every parameter reacts to the same amplitude signal, the display may be lively without telling the audience much about the performance.
Magic Music Visuals is a modular desktop environment for live and recorded audio-reactive imagery. Its current site lists version 2.5. The user guide distinguishes audio input from MIDI and OSC, with MIDI and OSC features in the Performer edition. Check the edition before planning a controller-driven set. A microphone-reactive demo and a MIDI-addressed performance system do not necessarily require the same product features.
Build a two-layer relationship
Follow the guide’s quick-start example, then reduce the scene to one simple form and one transformation. Connect audio level to size over a restrained range. Play the quietest and loudest material you expect to use, then adjust the response so both remain visually useful. The mapping should retain some motion at the quiet end without letting every peak fill the screen. Test this before adding feedback, textures or a second layer.
Add a second visual layer with a different time scale. It might respond to a slower musical envelope or remain nearly still while the first layer moves. Give it a separate purpose: framing the active sound, marking a sustained condition or preserving a trace of an earlier event. Visual contrast can come from different rates of change as much as from colour. Leave enough stable information for the moving element to be perceptible.
Map a performed gesture
In the Performer edition, select the intended MIDI source and use the documented MIDI/OSC Learn workflow for one parameter. Begin with a conventional controller message whose value you can vary slowly. Confirm which channel is selected. Then decide whether the same movement should change the sound and image together, or whether the image should answer with a delay or a different scale of movement.
For MPE-derived visuals, first establish what the visual patch actually receives. Independent note expression involves multiple channels; a single channel selection or a single global value is not equivalent to displaying every voice independently. Start by following one voice or by deliberately reducing several gestures to one visual control through a routing tool you understand. Make that reduction a compositional choice rather than an unnoticed loss of performance data.
Rehearse the screen as part of the piece
Record or rehearse a complete passage and watch it once without listening. Does the image have a readable shape in time? Then listen without watching. Finally, bring them together and ask what each contributes. Exact duplication is one possibility, but an image can also prepare an arrival, hold a memory or resist a busy musical moment. Decide where that difference helps the piece rather than treating synchronisation as the only goal.
Use the real output resolution and display arrangement during rehearsal. A scene that behaves comfortably in a small preview may ask more of the system at performance size. Keep a simple fallback scene ready and make scene changes easy to reach. Save the project with its media accounted for. The most useful visual system is one you can perform and recover confidently while continuing to listen, not merely one that produces a striking screenshot.
Try this.
Build one Magic scene with two shapes. Let audio level affect the first and one deliberate MIDI gesture affect the second. Use a restrained range for both. Perform a one-minute phrase containing a quiet passage, a sustained change and a clear ending.
Listen back
- What musical relationship does the image make easier to perceive?
- Are you following a single MPE voice or deliberately combining voices?
- Can the scene remain expressive during a quiet passage?