pixelsncodes/lumen

Lumen is a music synthesizer with an unusual trick: it can turn a photo into a playable sound. It ships as a VST3 plugin for music software and as a standalone Windows app, built with C++ and the JUCE audio framework. When you open it, you land on a default pad sound with four macro knobs labeled Tone, Motion, Space, and Texture, a scrolling display of the sound's frequencies, a spot to drop an image, and a keyboard you can play with your mouse. The core feature is called Lens. You drop any photo onto the window, and Lumen reads the image's pixel data and converts it into a wavetable, which is the raw waveform shape the oscillators play. One mode reads the image row by row as a waveform, so playing a note morphs its way down through the image. Another mode treats the image more like a spectrogram, building the sound's frequency content directly from the picture's brightness pattern. There is also a setting called COLORS that additionally reads the image's colors and blends them into the filter, envelope, drive, and other settings on top of whatever patch you already had loaded, rather than replacing it outright. The same photo always produces the same sound on any machine, and nothing is uploaded anywhere. Underneath the image tool sits a fairly standard subtractive synth engine: two oscillators plus a sub oscillator and noise source feed a filter with drive, three envelopes, three LFOs, and the four macro knobs, all of which can be dragged onto any knob to modulate it. The signal then passes through chorus, delay, reverb, and a limiter before the output. It supports polyphonic, monophonic, and legato playing modes, with adjustable glide between notes. The plugin includes 32 factory presets across bass, leads, pads, keys, and textures, and you can save your own presets, which store the complete patch including any loaded image and its resulting wavetable, so a preset still recalls exactly even if the original photo is deleted. The README documents the underlying math for how images are converted to waveforms in detail for readers curious about the signal processing, but using the plugin itself does not require understanding that math.