note.init — ALTÉRATON

intent#

Altératon is a research-creation device that questions the conditions under which an image and a sound are produced through code. The project draws an explicit parallel with experimental cinema: it takes up its fundamental gestures — delegation, alteration, materiality — and extends them through algorithmic operations.

The title condenses the program: altération (alteration) and Altératon share the same root. Cinématon (Gérard Courant, 1978–) is the inaugural reference — a serial protocol founded on the delegation of the filmed portrait. Altératon proposes a reconfiguration: the gaze is delegated, but the image is no longer returned as is. It is traversed by code, rewritten, transformed.

The displacement is precise. Where Courant maintains a stable, protocol-bound recording device, Altératon moves the constraint elsewhere: it is no longer the filmed subject that follows a protocol, it is the delegation of the gesture itself. The author is no longer the one who captures, but the one who sets up the conditions of capture and of its re-reading.

the device#

The protocol unfolds in three stages:

1. Delegation. A Super 8 camera is handed to a participant. No aesthetic instruction is given. The participant films freely — gestures, contexts, their own intentions. They become the temporary operator of a project whose structure and destination they do not control. This dissociation between individual capture and overall structure is constitutive of the device. Each reel keeps a record of its provenance and is covered by a written authorization.

2. Digitization. The exposed film is developed — partly autonomously, through the resources of L’Amorce, the long-term goal being a personal lab — then digitized frame by frame. Grain and defects (scratches, jumps, fogging, burns) are not corrected: they are the signature of the medium. The move from analog to digital is not a mere conversion: it marks the entry into a computable regime. The image leaves the domain of the imprint to become an array of values.

3. Re-reading. Digitized images are not processed: they are played. No editing software, no graphic post-production tool is used. Code accesses the data directly and operates on it. This second, algorithmic writing is the core of the project. Code is not an auxiliary tool: it is the medium of the work.

the protocol has already been tested#

A first film corpus — close to five thousand Super 16 frames, scanned on a professional scanner — served as a test bench. It was filmed by Lothaire Charuel and Benjamin Sebbagh, with their consent, under the delegation protocol: no intervention on framing or subject.

This first trial validated the full chain, from scan to projection, and confirmed the essential point: delegated material resists the device, it does not let itself be reduced to a decorative motif.

Super 8 will carry the delegated capture to come. It is the only format compatible with the protocol — a light camera passed from hand to hand, film I can develop myself, an economy that makes a series possible.

documenting the real, documenting the medium#

The project is documentary on two inseparable counts.

It documents the real first: each reel is an archive of true gestures, real light, fragments of life caught by hands that are not mine. Film preserves the indexical trace of the world — grain, instability, a trembling hand — that no synthetic render could fabricate.

It then documents its own medium. The second life of these images — development, digitization, re-reading by an engine written for the occasion — is an investigation into the very matter of the image, photochemical then digital. The project makes visible the technical chain that usually stays invisible: what a frame becomes when it is read as a signal, what the machine keeps and what it loses.

That is the tension that interests me — between the indexical fidelity of film and algorithmic betrayal. Between what the real inscribed, and what code makes of it.

a bank of viewpoints#

By multiplying the reels handed to different hands, Altératon gradually builds a bank of images and viewpoints: an archive of real, situated, irreducibly human gazes.

This ambition takes on a particular meaning in an age of AI-generated imagery, where images proliferate without referent, synthesized from aggregated data, with nobody behind the gaze. Altératon takes the opposite stance: here, not a single pixel is invented. Everything comes from a gesture filmed on film, chemically inscribed, undeniably having happened. Code transforms this material, deforms it, makes it drift — but it never fabricates the real.

Building and preserving this bank of authentic gazes then becomes, in itself, a documentary act — and, modestly, a gesture of resistance: reasserting the value of the lived image against the generated one.

an instrument, not a piece of software#

Excluding every graphic application (no GUI, no post-production tools) is not a pose of austerity. It is the condition for code to genuinely become a medium of writing rather than a mere technical intermediary.

In its place: a real-time engine written for the project, in C++ and GLSL, which displays the digitized film and deforms it at sixty frames per second while I play it. It holds the whole chain: streamed playback of the corpus, full film transport, thirty-four stackable transformations, a persistence buffer, a hot-reloading shader editor.

The displacement is essential. Editing software asks you to decide in advance, then wait for the render. The instrument answers immediately: I type a character, the image changes before my eyes, and I correct. The relation to the material becomes manual, almost physical — closer to instrumental playing than to parameter tweaking. The thirty-four transformations are not “filters”: they are the keys of a keyboard — stackable, dosable, something you learn to play.

This constraint extends the practices of experimental cinema where tool and work are inseparable — McLaren scratching and painting directly on film, Brakhage gluing moth wings onto celluloid, the structural filmmakers whose apparatus determines the form.

The instrument can be seen running on the project page — full take, uncut.

algorithm and number as writing#

This stance has a decisive consequence: it is not only the effect produced that counts, but the method that produces it. Choosing an algorithm and a mathematical structure is itself a gesture of writing.

The instrument is not driven with a mouse. It is commanded by a score written live — a grid of characters where each sign triggers a transformation and its value sets the intensity. The same language drives the film’s own motion: speed, direction, in and out points, freezes, back-and-forth. There is therefore no editing in the usual sense: there is a writing of the traversal, played over the corpus, replayable differently every time.

This is where number becomes material. Entrusting the temporality of the work to structures foreign to it: slow motion paced by the sequence of prime numbers, repetitions set by recursive sequences, polyrhythms where image and sound follow offset cadences that gradually drift apart. Editing is no longer dictated by taste but generated by number — a temporality that never “lands on the beat”, irreducible to the mechanical rhythm of projection.

This is how the project goes beyond a mere glitch effect: program and number are not in the service of a look, they are the writing.

the loop: the image that re-reads itself#

At the heart of the instrument sits a mechanism with the exact shape of Lucier’s piece. At each frame, the engine does not start from the film alone: it re-reads the frame it has just produced, offsets it, flips it, attenuates it, and blends it into the next. The material feeds back into its own reading, sixty times per second.

Where Lucier recopied his voice from tape to tape until the room absorbed it, here it is the frame that dissolves into its own re-reading — into trails, spirals, afterimages that never quite fade. With each pass, something of the initial real is erased and something else arrives. The document becomes erosion, and that erosion is played: I can let it run, hold it back, restart it.

the sonic dimension — what works#

Sound is already there, and it is not added to the image: it shares its origin.

The grid of characters that drives the deformations drives, at the same time, sixteen synths, an FX chain and a sample bank written for the project. One and the same keystroke produces what is seen and what is heard. Not by illustration, not by after-the-fact synchronization, but by common origin. There is not a music and some images; there is one writing, and two outputs.

the sonic dimension — what remains to be written#

Both ends of the circle remain to be closed. This is the main open work the project faces today.

Super 8 never had an optical soundtrack (unlike 35 mm, whose lateral strip encodes sound as a light pattern read by a photocell). Altératon aims to invent one for it through code.

Reading (image → sound). Pixels from each frame read as an audio signal. Luminance converted into amplitude, the scan path (raster, column, spiral, Hilbert curve) determining timbre. The film becomes its own score. Sound is not added: it is extracted from the image. Every transformation applied to the image would then propagate automatically to the sound.

Inscription (sound → image). The inverse operation: an audio signal inscribed into the lateral strip of the image, in the manner of 35 mm optical soundtracks. This gesture reproduces McLaren’s (Pen Point Percussion, 1951), Fischinger’s (Tönende Ornamente, 1932) and Sholpo’s (the Variophone, 1930s) — but the source is no longer a drawing: it is sound captured by a participant.

The round-trip. The chain sound → image → sound sends the signal through the visual medium. Super 8 grain, compression, digitization artifacts become a physical filter on sound. Each pass is an alteration. Iterated in a loop, the process produces successive generations where the sound degrades like a photocopy of a photocopy — a principle akin to I Am Sitting in a Room, except that here the sound travels through visual space, not acoustic space.

Sound as control. Turning pitch, intensity and timbre into a full-fledged source of control over the image — continuous values and named parameters, where the score today offers only thirty-six steps. Sound would then deform the film directly, and no longer only the score that writes both.

Once these ends are joined, Lucier’s loop will no longer pass through the image alone, but from one medium to another.

under study — sonic delegation#

The delegation protocol can be extended to sound: an audio recorder (a C5 cassette, whose 2 minutes 30 per side matches exactly one silent Super 8 reel) lent to a participant. As with the camera, no instruction is given.

The captured sound would then be inscribed into a digital image — a synthetic film strip generated by code, with perforations and a variable-area soundtrack, scrolling on screen in sync with the audio. The participant produces sound, but the result is an image. The heard becomes the seen.

Crossing the two protocols opens a further possibility: one participant’s film (image) carrying another participant’s sound (audio). Authorship is doubly fragmented; nobody controls the outcome.

Open avenue, dependent on closing the image ↔ sound loop described above.

positioning#

Altératon sits at the intersection of experimental cinema, algorithmic art, glitch, live coding and delegation practices. It follows the lineage of graphical sound synthesis (McLaren, Fischinger, Oram) and of open serial practices (Courant).

The project does not seek to produce a film in the classical sense. It seeks to produce a cinema without a center — a form whose origin is disseminated (multiple participants), whose transformation is computational (exclusively through code), and whose process is ongoing (an open, potentially infinite archive).

This is not a figure of speech: technically, there is no final file. The work only exists while it is being played, and each reading of the corpus yields a different version. What is preserved is not an edit — it is the bank of gazes, and the instrument that knows how to read it.

The film becomes a structure, a flow, a system.

forms of presentation#

  • video installation (continuous playback, never identical — a visitor returning an hour later does not see the same thing)
  • open archive, interactive variant under study (handing the visitor part of the instrument’s controls)
  • serial projection (contributions scrolling one after another)
  • live played session (the score written in view of the audience)
  • synthesized optical soundtrack (upcoming)
  • scrolling synthetic film strip, image generated from a participant’s audio (under study)

state of play#

instrument .............. operational
test corpus ............. proven (Super 16, ~5000 frames)
delegated capture ....... upcoming (Super 8)
development lab ......... to be built
image ↔ sound loop ...... open work
exhibitable form ........ in progress

The instrument is a tool, not the work.

references

  • Gérard Courant, Cinématon (1978–)
  • Norman McLaren, Pen Point Percussion (1951), Synchromy (1971)
  • Oskar Fischinger, Tönende Ornamente (1932)
  • Evgeny Sholpo, the Variophone (1930s)
  • Daphne Oram, Oramics (1957–69)
  • Alvin Lucier, I Am Sitting in a Room (1969)
  • Stan Brakhage
  • Structural film (Sharits, Snow, Frampton)