Unknown Signals

Work and context

Unknown Signals sets a restless geometric cloud against a field of waves. The background appears to record an extended, continuous disturbance, while a circle or square isolates a smaller region of activity. A tiny emblem elsewhere in the image acts as a counterweight. The relationship resembles an instrument and its uncertain reading, although the program is not receiving radio transmissions or displaying measured scientific data.¹

k0ch's project statement invokes deep-space signals and hidden radio stations as reminders of waves outside ordinary vision. Art Blocks lists the work as a Factory collection of 1,000 outputs released on 23 August 2021. The statement offers two controls: clicking to stop or start the animation, and pressing b to remove the foreground.²

The header shows AB5D's #377, verified against the current linked-wallet set. In the coordinator's original-renderer observations, red and violet wave lines cross a black ground. A translucent magenta square encloses a bright, dense triangular mesh, and a small diamond with a central dot sits at lower left. Two views showed the mesh changing while the background and enclosure stayed fixed. This is an artwork whose principal layers operate on different temporal terms.³⁴

Conception

The title's uncertainty belongs to the imagery rather than to a hidden technical data source. The inspected program calculates its wave field and mesh from seeded values. Its suggestion of receiving something from elsewhere is therefore a constructed experience: the viewer encounters a visual language associated with signals without being given a message to decode.¹

A later primary source extends that distinction. The artist's Bandcamp album UNKNOWN SIGNALS, released on 24 April 2026 under the name SPIRITFLESH, explicitly continues the 2021 visual project in sound. Its account identifies Emiliano Canedo, formerly k0ch, and describes recordings made between August 2025 and March 2026 in Asturias. It discusses noise and analogue imperfection as part of the work's direction. This is retrospective evidence of an ongoing concern, not a contemporaneous diary of how the 2021 algorithm was developed.⁵

The album should not be mistaken for a soundtrack embedded in the tokens. No audio implementation appears in the inspected visual-art program, and the album was not auditioned for this essay. The connection is the artist's stated continuation across media. That later continuation is useful because it suggests that signal, noise and imperfect reception form a sustained subject rather than a title applied to a generic moving mesh.¹⁵

The visual construction also supports an interpretation in which communication remains incomplete. The mesh can look concentrated and purposeful, while its continual movement prevents any single configuration from settling into a stable sign. The smaller emblem is comparatively simple and fixed. The work places apparent complexity beside legibility without explaining a translation between them.¹³

Construction

The full indexed artist program is 9,074 UTF-8 bytes and uses p5.js 1.0.0 with its two-dimensional canvas. It creates a square based on the smaller window dimension. A custom random-number class takes the first eight hexadecimal characters after the hash prefix as its seed and advances through bitwise operations. The same class supplies a custom noise function. No external images or live signal feed are loaded.¹

The noise function initializes a table of 4,096 values from that seeded generator, then combines four levels of cosine-interpolated samples with diminishing weights. This is an authored source of correlated variation. It helps neighbouring positions change together, providing a different texture from independent random heights. The noise is subsequently multiplied by a sine expression raised to a selected power, so wave regularity and irregular modulation are combined rather than presented as competing alternatives.¹

The background is calculated once into an offscreen canvas. The program builds 120 rows, sampling 401 positions across each row. Before drawing a row's segments, it fills a shape down toward the background colour. This covers portions of earlier rows. The apparent ridges consequently emerge from both displaced lines and occlusion, not simply from placing transparent curves on top of one another.¹

Each row is drawn as 400 short segments, with colour related to the sampled displacement and a shadow glow applied through the canvas drawing context. Nine code branches define colour behaviour, including one family with randomized RGB components. This count describes the implementation. It should not be automatically equated with a marketplace's palette vocabulary or a statistically verified count of visible colour schemes.¹

The foreground starts with a cloud of points. One branch generates 100 points in a small region situated in a selected quadrant; another places 200 points around the centre. An inline triangulation routine computes connections between those initial points using circumcircles, edge deduplication and an enclosing supertriangle. The resulting list of vertex indices is stored.¹

The important distinction comes after that initial computation. During animation, the vertices move independently, with sine and cosine offsets governed by seeded amplitudes, multipliers and two frequency values. The triangulation is not recalculated each frame. The same connections therefore stretch, compress and potentially cross as their endpoints move. Describing the work as a freshly optimized Delaunay mesh at every instant would miss the source of its restless folding.¹

The moving triangles can be filled or drawn as wireframes, with optional marks at vertices. A translucent circular or rectangular enclosure sits behind them. The enclosure establishes a stable local field against the much wider wave pattern. A smaller circle or diamond, containing a dot, is placed toward an opposing area. Its simplicity balances the many intersecting edges without behaving as another animated cloud.¹

Each frame copies the precomputed background and then draws the foreground. There is no need to regenerate the wave landscape for the mesh to move. This separation makes the contrast observed in the originals precise: a stable trace and a changing network coexist on one canvas. It also means a screenshot can adequately record the background while preserving only one phase of the mesh.¹

Clicking toggles whether the geometry's time variable updates. The coordinator confirmed that a click held the mesh's geometry steady across two subsequent views of #377. The code still draws frames and chooses a small random variation in shadow blur while paused, so this need not be a pixel-identical freeze. On resuming, time is taken from millis, which means the mesh can jump to the current elapsed-time position rather than continuing an accumulated pause clock.¹³

The b handler toggles the entire foreground branch, leaving the precomputed background. That behaviour is explicit in the code and project instructions, but the coordinator's attempted keypress did not visibly remove the foreground; focus was uncertain. It remains a documented control whose successful runtime use was not established in this pass. No resize handler or audio system was found.¹²

Mint and release

The refreshed index reports 1,000 invocations and a current maximum of 1,000. The first token was minted on 16 August 2021 at 12:19:37 UTC. Activation followed on 18 August at 17:10:51. The second token and indexed start share 23 August at 16:01:02, matching the public release day given on the collection page.²

Completion is indexed at 16:27:49 on 23 August. The interval from the second token and indexed start to completion is 26 minutes and 47 seconds. It is a calculation between named index events, not proof of an independently recovered opening announcement or every allocation condition. The earlier token's purpose remains unknown.²

The current indexed price field is 0.4 ETH. No original sale schedule or dated price history has been recovered, so this field is not evidence that every original purchaser paid that amount. The separation matters particularly where an earlier first token, activation and the main mint day are distinct events.²

Art Blocks' September charity report associates Unknown Signals with approximately $395,157 for GiveWell. The report explains that its figures combine artist-reported donations, funds set aside and calculations from pledges. It documents the contemporary reported charitable contribution, while leaving the exact transfers and their conversion into dollars outside this essay's verification.⁶

Rights and display

The current project record reports NFT License but supplies no licence URL or version. The official NFT License website publishes version 2.0. Its operative terms retain creator ownership and grant conditional rights tied to ownership of a purchased token; third-party website provisions include ownership verification and limits on continued display. It is not a blanket licence for unrestricted public reuse. The exact adoption of that version by this project remains unconfirmed.²⁷

The archive preserves the reported label and links the unmodified original of AB5D's verified holding for this critical account. The link identifies the artwork and its observed behaviour; it does not establish which version of the reported licence applies or grant visitors unrestricted reuse rights.³⁴⁷

No separate artist-code licence was found. The p5.js library is covered by its own LGPL 2.1 terms, which do not license the artist's program. The inline triangulation code also requires a firmer provenance and licence identification. The package therefore provides the script's digest and analysis without distributing it. The later Bandcamp album's rights statement belongs to that separate sound release and does not settle the token artwork's terms.¹⁸⁵

Examples

The research coordinator loaded and viewed the actual originals #0, #1, #10, #20 and #377 at two different moments each. Their observations underpin these close readings and are attributed in the evidence record. Earlier blank captures of #0 and #1 were incomplete loading states and were discarded rather than described as artworks.

Edition #0 has a black ground with fine red and violet parallel waves and a broad, more vertical violet ridge. A red translucent square occupies the lower-right region, enclosing a red wire cloud; a small diamond and dot sit toward the upper left. The diagonal separation between the two foreground elements distributes attention across the wave field. The square supplies a stable boundary, but the mesh silhouette changes between views. Its lines consequently read less like a fixed symbol than a temporary configuration inside an observation window.⁹

Edition #1 reverses much of that tonal and spatial emphasis. A pale white-pink ground supports repeated bands of purple and red wave lines. A larger pink translucent circle sits upper left with a dense magenta mesh, balanced by a small outlined circle and dot lower right. The pale field makes the waves feel like marks drawn across a surface, while #0's dark field gives its lines a more luminous appearance. This is a visual contrast, not evidence of different signal sources. In both originals the changing mesh remains distinct from the stationary background.¹⁰

Edition #10 centres a green translucent circle against black. Green and yellow waves concentrate toward the upper left, with blue in the lower-right region, while a small green outlined circle and dot sit lower right. The centred mesh provides a different balance from the offset arrangements of #0 and #1. Its silhouette changed across the two observed frames without moving the surrounding wave pattern.¹¹

Edition #20 returns to a pale pink-white field with red and purple rows. A large red translucent circle and dense red wire cloud occupy the upper left, while a small red diamond with a central dot sits lower right. Compared with #1, the repeated placement does not produce an identical image: colour intensity and the simpler diamond countermark change the relation between the two areas. Its mesh also changed while the background stayed fixed.¹²

Significance and unresolved questions

The work's animation depends on preserving a relationship while disturbing its participants. Connections established among the initial points remain in place as the points move. That procedural decision makes a network persist through change rather than repeatedly restoring an optimal arrangement. It is a useful technical counterpart to the title: something continues to signal, but its changing shape does not deliver a stable interpretation.¹

The distinction between the two layers also resists treating animation as a general effect applied everywhere. The wave field is already complete. The mesh continues to unfold. A viewer can compare an enduring trace with an event that has not finished, and the pause control temporarily shifts that relationship. This temporal organisation gives the collection a specific character beyond its fluorescent colours or triangulated surfaces.¹³

Unresolved research questions:

• Which exact NFT License version was adopted, and what project-specific display terms apply?
• What are the artist-code and inline triangulation component's applicable licences and provenance?
• What were the original sale price, allocation conditions and purpose of the earlier first token?
• Can the reported GiveWell contribution be reconciled with dated transfers?
• Can the b control be successfully verified with confirmed canvas focus, and how consistent are glow and timing across browsers?