720 Minutes

The work and its context

720 Minutes lets a clock become difficult to recognise, then makes its function visible again. Clouds of coloured particles fold into bright edges, open into dark pockets and scatter across the field. A straight line can suddenly emerge through this activity, directing the eye toward a central meeting point. The clock does not sit behind the animation as a separate mechanism. Its directions are made from the same unstable material as the surrounding image. ¹²³⁴

Alexis André released the 720-work series through Art Blocks Curated on 5 March 2021. Each token is assigned a minute in a twelve-hour cycle, with a special activation associated with that time. The edition size consequently describes the structure of the work as well as its supply. Together the tokens distribute attention across a complete cycle, while each individual clock retains its own visual character. ⁵

The Alexis André dossier places the project within his wider practice. Here the central problem is perceptual: how much can a clock change while continuing to tell time? The inspected outputs answer through degrees of legibility. Some expose a conspicuous spoke; others make the viewer search among folds and particle trails. Looking for the time becomes part of looking at the image. ⁶¹²³⁴

Conception and development

In the release interview, André traces the idea to a conversation with Erick Calderon about a pulsing daily animation and the possibility of a work that did something with time. The concept led to a number of editions corresponding to minutes, rather than an arbitrary cap on variation. His emphasis is on designing a coherent system whose combinations matter, not selecting a handful of fortunate outputs. ⁷

The same interview connects development to EoN, his generative audiovisual collaboration with Jean-Michel Jarre. André identifies its 120-beat-per-minute music as an influence and says he listened to it while developing the clocks. That is evidence about the work's making. It does not establish that the token renderer contains music or requires EoN to run. ⁷

The artist's project page treats the recurring special minute as an occasion to reflect on change: a clock's hands return to the same positions while its viewer has lived through another interval. The page also documents a later Sony Vision Shibuya installation with motion tracking, music and a numeric display. Those additions belong to the extended installation and should not be transferred automatically to the collected browser work. ⁸

This conception makes the act of returning consequential. An ordinary still image can be revisited without the work itself selecting a different temporal condition. Here the viewing moment is an input. Returning changes the hour and minute directions and may encounter a special branch of the program. The promise is not simply more pictures from one seed; it is an organised relationship between a particular token and recurring time. ⁶

How the work was constructed

720 Minutes gives a token two different relationships to its input. A fixed-seed procedure first shuffles the 720 positions in a twelve-hour cycle. The token identifier selects a position in that shared permutation, assigning its special minute. A separate sequence seeded with the token hash supplies visual parameters. The special time is consequently not a straightforward reading of the edition number as minutes after midnight.⁶

The renderer also needs information that does not come from the token. It reads the current system time and applies the local timezone offset captured when the program starts. Hours and minutes become directions in the image; seconds supply a recurring pulse and help organise changes in colour and particle placement. The artwork is therefore more accurately described as a process that combines token identity with a viewing moment than as a picture recoverable from the hash alone.⁶

Its moving material consists of 30,000 particles held in typed arrays. A native WebGL renderer draws them through instancing, reusing a small, roughly circular shape at many positions and sizes. Colour comes from compact palette data and interpolation between neighbouring entries. The program chooses among different arrangements and flow behaviours, so a single underlying renderer can sustain substantial differences without replacing its basic particle vocabulary.⁶

Clock hands are formed through forces on those particles. The program measures their angular distance from the hour and minute directions, then adjusts nearby positions and sizes. There is no separate conventional hand drawn over an unrelated animation. Time participates in the same movement that gives the surrounding field its appearance. This mechanism explains why a clock can remain legible while its material is continually being dispersed and rearranged.⁶

The impression of persistence depends on more than the present particle positions. Two textures and framebuffers pass the previous image through weighted blur and transformation before the new particles are drawn. Scaling, rotation and blending choices affect how old marks survive. The image thus has a short visual memory: it contains traces of movement as well as the latest arrangement. These retained traces differ from a shader that simply evaluates several time offsets afresh for each frame.⁶

Each second can replace many of the particles' starting positions. Between these changes, noise-based motion and other forces reshape the field. Pointer interaction introduces an additional influence: holding and moving the pointer supplies a force location, and a press can change its sign. These interactions also consume values from the program's pseudorandom sequence. Reproducing an observed state would therefore involve its interaction history and timing, not just its token hash.⁶

The full special-minute condition compares both the assigned minute and the hour within a twelve-hour cycle. It changes colour behaviour, feedback and forces, including a pair of opposing attraction locations. A separate condition operates when only the minute number matches during other hours. Those branches should not be conflated: a repeated minute-number effect is not necessarily the token's complete twelve-hour activation.⁶

The script handles window resizing by rebuilding its rendering surfaces. It does not use p5.js, despite the artist's earlier engagement with Processing. Its browser requirements include WebGL and the instancing extension used by the renderer. Retaining the JavaScript and token input preserves essential instructions, but a future presentation must also account for compatible graphics behaviour, local time and interaction. The artist's description of continuity is an artistic proposition, not proof that every future browser will reproduce every frame identically.⁶⁸

The mint

The collection page gives 5 March 2021 as the release date. The indexed first token is recorded that day at 02:21:53, activation at 21:30:26, the second mint and scheduled start at 22:59:44, and completion at 23:05:34. Each field describes a different event or setting. Keeping them separate avoids converting the earliest token into an unsupported public-sale start. ⁵¹²

André's own project page reports that the collection sold out in less than six minutes. The interval between the indexed second mint and completion is consistent with that rounded account. It does not independently classify every transaction as public, artist or reserved. The artist's report can be retained as testimony without turning those index fields into a complete reconstruction of the sale. ⁸⁵

The current index reports 720 invocations, a maximum of 720 and a price setting of 0.15 ETH. This is a configuration observation, not independent proof of the original advertised price or every recipient's payment. Historical pricing and allocation details require additional launch-era or transaction evidence. ⁵

Copyright and licensing

The current artist-set artwork label is CC BY-NC 4.0. The inspected metadata supplies no terms link; the corresponding Creative Commons deed is linked separately. It identifies attribution, indication of changes and a noncommercial condition among the terms for sharing and adaptation. A complete history of the artwork's permissions has not been reconstructed. ⁵⁹

The artist's explanatory website carries its own copyright notice. That notice should not silently replace the token's recorded artwork label or be expanded into a verified licence for every component of the renderer. No separate artist-script reuse grant was established in this research. The indexed program contains helper routines as well as artwork-specific logic; their upstream provenance and permissions need their own evidence. The source is analysed here, not redistributed as part of the essay. ⁶⁸

Close reading of four outputs

The comparison uses editions 0, 1, 10 and 20, selected before viewing as an early-edition group. Their full identifiers are 27000000, 27000001, 27000010 and 27000020 on Ethereum contract 0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270. Each canonical generator was inspected in successive live states. These observations were made during a limited session, not a complete twelve-hour cycle. They do not establish the full special-minute appearance of each token or the distribution of features across the edition. ¹²³⁴

Edition #0 spreads blue, pink, peach and cream particles across a black field. Broad curling sheets are made from individual beads, with soft trails projecting into the darkness. A particularly clear straight spoke extends from the upper left toward the central convergence. During later viewing, the folds change position and the spoke's colour changes, but its forceful direction keeps interrupting the softer surrounding flow. The image alternates between a particulate cloud and a structure whose lines can be followed. ¹

The contrast within #0 makes the clock legible through a difference of behaviour. Its spoke is not necessarily the brightest object at every moment, yet its straightness distinguishes it from the curling sheets. Looking more closely reveals that both are assembled from points and traces. The apparent separation between useful instrument and expressive image is therefore repeatedly undone by their shared material. A pointer drag was also tried, but that brief before-and-after observation does not isolate its effect from the animation's continuing movement. ⁶¹

Edition #1 appears cooler and more restrained in the observed views. Pale blue, white and yellow particles form bright crests around extensive black gaps. The crests can look like folds illuminated along their edges, although the source does not simulate cloth. Later, several luminous curves gather around the centre while other fragments remain near the margins. Compared with #0, the interior darkness is more assertive: it separates the moving fragments rather than allowing colour to fill much of the field. ²

Edition #10 concentrates the activity into a smaller, largely grey-white formation. Rows of distinct points bend into closely spaced arcs, exposing a repeated structure that is less immediately obvious in #0's colourful cloud. In one view the bright material stretches vertically; later it folds into a more compact arrangement with a narrow spoke reaching toward the upper left. The black surrounding field makes the changing silhouette conspicuous. The work can suggest a volume turned in space, but the inspected rendering evidence supports a moving particle field and image feedback rather than a modelled solid. ³⁶

Edition #20 uses pink, white and pale blue accents on black. The earlier observed state contains a broad curling band around a dark pocket, with smaller bright fragments separated from it. Later, the shape divides into sharper folds, a long diagonal becomes more conspicuous and successive ridges are visible within some coloured sections. Where #10 exposes rows of points as a principal texture, #20 frequently merges them into a brilliant edge. Its shifting balance between continuous ribbon and particulate surface is central to the comparison. ⁴

These views show several ways to make time perceptible without a conventional dial. A direction can remain readable while particles continually renew it. Elsewhere, the field can approach abstraction closely enough that identifying the clock becomes a deliberate act. The differences concern scale, density, contrast and the survival of previous marks, as well as palette. Reducing the series to colour variants would miss how differently the same task of telling time can occupy the screen. ⁶¹²³⁴

Significance and limitations

720 Minutes makes time both its subject and one of its operating conditions. The token supplies a set of visual possibilities, the clock supplies a changing temporal situation, and interaction can alter the course of the particles. That combination gives the work a particular kind of continuity: identity persists through movement, rather than requiring the same composition to be displayed each time. ⁶

The four inspected outputs reward comparison because their clock function is neither equally prominent nor aesthetically separate from the rest of the image. A bright spoke, a dark interval or a densely repeated arc can redirect attention between reading time and watching a form develop. The work's strongest argument is made at that point of exchange. An instrument associated with measurement becomes an occasion to notice the unstable appearance of the present. ¹²³⁴

A preservation record must keep the limits of this inspection visible. Ordinary motion was observed, but a complete activation cycle was not. The source identifies special conditions more precisely than these viewing sessions demonstrate them. Future comparison should document clock settings, elapsed duration, display dimensions and interaction history alongside the token, so that changes in rendering are not confused with intended differences in temporal state. ⁶

Unresolved research questions:

• Recover launch-era price announcements and classify public, artist and reserved mint allocations.
• Establish separate artist-code permissions and the provenance of embedded helper routines.
• Compare the indexed script against a retrieval pinned to a chain block.
• Observe and document full assigned-minute activations, with explicit clock and interaction conditions.
• Test rendering and resizing across supported browsers and compare a broader range of editions.