Cells

Work and context

Cells #71 stretches a blue field into a lattice of pointed limbs. Magenta bodies sit behind a dark blue foreground body near the lower edge, and thin white threads span the intervals between their arms. Circular openings reveal more crossings underneath. The apparent depth comes from overlap and fading outlines rather than a navigable three-dimensional space. This original renderer was inspected directly; the token’s current indexed owner matches one of the three wallets linked to the freshly checked AB5D profile.¹²

Hevey’s Cells is Ethereum project 122 on Art Blocks’ original V1 core contract. Art Blocks identifies it as a Factory release dated 3 August 2021, with 1,024 artworks. Those identifiers matter because the atlas number and a token’s edition number are different forms of identification. #71 belongs to the project but is not its project number.³

The series turns a recognisable organic silhouette into an instrument for dividing a square. A central body, a hole and several branching extensions remain legible across striking changes of colour and scale. In the inspected works, branches often continue beyond the image boundary. They imply that the visible square cuts through a larger tangle, even though the program actually constructs a bounded canvas. That tension between a complete digital object and an apparently incomplete view is one of Cells’ most effective devices.⁴⁵⁶⁷

Conception

Hevey names the particular shape of neurons as the initial inspiration. The artist’s own collection page explains that the investigation then expanded through size, branching, form, colour, positioning, additional elements and effects. This is a statement about a starting point and subsequent visual exploration; it does not claim scientific accuracy or a functioning neural model.⁸

The distinction is visible in the work. The arms are frequently broad, straight ribbons joined by rounded shoulders. Colours separate overlapping figures, while central holes make each body resemble a cut shape. A biological association can coexist with impressions of punched sheet material, straps or stretched membranes. These are readings of the inspected outputs, not additional sources of influence attributed to Hevey.⁴⁶

The code supports the artist’s account of a family expanded through interacting decisions. It includes multiple styles, palette tables, optional outlines, offset fills, shadows, interior markings, exterior connecting lines and additional small figures. It also changes branch count, orientation and spacing. Variation is therefore distributed across several levels: the silhouette of an individual cell, the relation between neighbouring cells, and the visual treatment of the entire composition.⁹

There is no development diary in the inspected sources explaining which effects arrived first, which alternatives were discarded, or how Hevey selected the final palette weights. Reading the finished program reveals its operational decisions, but cannot recreate that sequence of artistic judgement. The essay treats the public intention statement and the final executable structure as complementary evidence, with different limits.⁹⁸

Construction

The inspected artist program contains 11,899 UTF-8 bytes and uses p5.js 1.0.0. Its compact variable names make the broad design easier to understand through the relationships between routines than through names alone. A helper extracts hexadecimal segments of the token hash and optionally reduces them modulo a chosen number. These values select styles and parameters. Placement also uses p5’s seeded random function, with a string-hashing helper turning token-derived numerical expressions into seeds. p5 documents that a constant seed repeats the random sequence; this supports repeatability, without establishing identical pixels in every browser.⁹¹⁰

The drawing starts from a square whose side is the smaller window dimension. A nominal 2,000-unit coordinate scale is converted to canvas dimensions. Some low-count arrangements use predetermined centre positions; others attempt seeded placements that avoid overlapping central bodies, retrying up to a limit. This matters visually because the script can keep the bodies distinct while their much longer extensions still cross. Separation at the centre does not produce isolation across the image.⁹

Each cell begins with points distributed around its centre. Depending on the selected conditions, their angles and radii can be regular or perturbed. Adjacent points form pairs whose movement is calculated from neighbouring directions. The program advances these pairs iteratively, adjusting direction and magnitude according to distances and other parameters, and stores the resulting point sequences. Joining those sequences produces the extended contours. The branching image arises from this geometric construction, rather than from simulating electrical signals or biological growth.⁹

Several controls change the character of those contours together. The iteration settings affect the number and spacing of construction steps as well as scale-related parameters. Other choices alter the number of arms and their orientation. The familiar body-with-extensions is therefore stable as a visual grammar while its proportions remain variable. This helps explain how #1 can present shorter, visibly terminated arms while #10 sends slender extensions across nearly the full field.⁹⁵⁶

The rendering order is equally important. Before drawing each cell, the code lays a translucent square of the background colour over what is already present. Earlier shapes consequently recede. A shadow or offset-colour pass can precede the final contour, and optional circular contours cut openings through the body. White or black strokes then clarify edges. The result resembles a stack whose lower layers have lost contrast, although the drawing is made with two-dimensional shape operations.⁹

Exterior lines can bridge portions of neighbouring arms, creating the taut, fanlike webs visible in #71. Other branches add interior lines or change filled bodies into line-based treatments. These options are conditional: the five originals inspected here cannot establish the frequency of every possible style. Their role is to test concrete relationships between visible results and the program, not to stand in for an edition-wide census.⁹¹

The runtime also deserves precision. Most conditions call noLoop, while eligible moving elements use a requested 24 frames per second. The animated branches alter interior circular forms or dangling elements; clicking toggles looping only when animation is available. The code does not regenerate a new token on a click. No audio routine or external image loading was found. The selected hero has eight cells and falls outside the ordinary fewer-than-eight animation condition; its inspected image is presented as static. Animated variants were not empirically audited here, and no universal loop duration is asserted.⁹¹

Mint and release

The current index records 1,024 invocations and a maximum of 1,024. It gives the first mint as 28 July 2021 at 20:57:47 UTC, activation as 1 August at 18:08:09, and the second mint and indexed start as 3 August at 18:00:29. Completion is recorded at 19:15:30 that day. The interval from the second mint to completion is one hour, fifteen minutes and one second. The earlier first mint should not be mistaken for the public release date; its exact purpose remains unresolved.³

A contemporary Art Blocks newsletter supplies pricing evidence that the current index cannot. Druid’s report, published on 4 August, describes a Dutch auction beginning at 1 ETH, decreasing by 0.1 ETH every ten minutes toward a 0.2 ETH resting price. It says the project sold out in a little over an hour and raised 78.17 ETH for Food for Life. The article’s dollar conversion and projected charitable reach belong to that contemporary report, rather than to an independently audited accounting here.¹¹

The current stored price is 0.3 ETH. That is a present indexed configuration, not evidence that every mint cost 0.3 ETH or that the auction opened there. The reported schedule and the indexed completion interval would place completion before an eighty-minute descent to the stated floor, if the schedule began at the indexed start. This is a consistency inference, not transaction-level proof of the last paid price. The newsletter’s wording about reaching the resting price is therefore not treated as a verified completed price step. Reservations, individual payments and donation transfers require further historical evidence.³¹¹

Rights and licensing

Art Blocks’ current project record reports CC BY-NC 4.0, with no separate terms URL populated. The corresponding Creative Commons legal text permits noncommercial sharing and adaptation subject to its conditions, including attribution, a licence reference and identification of modifications. It prohibits additional restrictions that prevent recipients exercising those licensed rights. The label is not a transfer of copyright to the token holder, nor permission for unrestricted commercial use.³¹²

For this essay, the display basis is an attributed, unmodified original rendering in a noncommercial editorial context, linked to the token and the licence. The current label has been checked, but its historical changes have not been reconstructed. No independent licence for redistributing Hevey’s artist script was found. Availability for inspection is recorded separately from permission to republish the code.⁹¹²

The dependency has its own terms. p5’s official copyright page identifies the library’s GNU Lesser General Public License 2.1. Those library permissions do not automatically license the artist’s program or the resulting artwork. No raw artist code is included in this essay package, and no commercial reproduction permission is inferred from ownership.¹³

Four outputs

Cells #0 is densely layered against deep purple. A large white body near the top sends broad diagonals downward, while a green body anchors the lower right. Burgundy and grey forms cross behind them. The circular holes are unusually useful viewing points: they reveal additional strips and intersections instead of simply returning an empty ground. Indexed features report nine cells, long branches and shadows. The observation shows how a greater count can become a crowded hierarchy rather than nine equally prominent objects.⁴

Cells #1 uses a bright red ground and several red-brown layers, with a charcoal foreground body at the lower left. Some arms terminate within the square, making their flat ends conspicuous. The white outline gives the darkest body particular clarity against neighbouring red. Its six cells, medium branches and offset backgrounds distinguish it from #0’s longer, more entangled construction. The close colours compress depth, while the contour restores separation.⁵

Cells #10 is much lighter. Cream surrounds yellow, pale orange and a magenta cell entering from the upper left. Long black-edged extensions cross the open centre, and grey offset shadows make several strips appear lifted from the ground. Its five-cell configuration gives the image more breathing room, but a diagonal magenta arm still cuts forcefully through the lower-right region. Negative space becomes a major shape rather than merely the residue between crowded bodies.⁶

Cells #20 returns to a warm ground, this time coral. An orange body occupies the upper region, with muted purple forms receding behind a dark navy cell positioned toward the lower right. One nearly horizontal navy arm traverses most of the square. Fine white contours and slightly displaced colour edges produce a layered graphic effect without #10’s pronounced grey shadows. The seven-cell arrangement feels directional: the foreground junction gathers crossings into one off-centre knot.⁷

These early editions were selected for reproducible comparison, not rarity ranking. Together they demonstrate how count, branch extent, tonal contrast and offset treatment change the reading of a shared shape. The held #71 adds another meaningful difference: exterior connecting lines turn the gaps between arms into active drawn structures.¹

Significance and open questions

Cells makes the distinction between motif and system unusually visible. The neuronal starting point remains recognisable, yet the effect of any one output depends on choices that have little to do with anatomical description: which body is drawn last, how earlier layers are veiled, whether a border is white or black, and whether the background is pale enough to make shadows assertive. Its subject is partly the cell-like figure and partly the field of relations created when several such figures compete for a square.⁹⁴⁶

The historical release also places this visual experiment within a specific form of distribution. A finite generative edition met a declining-price sale and a reported charitable contribution. Keeping those facts distinct from present database settings preserves their meaning. The art can be studied as an executable composition without turning a stored price into a fictional launch history, or a projected charitable benefit into a verified outcome.³¹¹

Unresolved research questions:

• What development tests led Hevey to the final style and palette weights?
• What was the purpose of the July first mint, and were any editions reserved?
• Which auction price changes actually executed, and what did individual buyers pay?
• Which transactions reconcile the reported Food for Life contribution?
• Was a separate artist-code licence issued, and has the artwork licence changed?
• Which animated variants survive long runs, resizing and different browsers with equivalent behaviour?