# [Dis]entanglement by onlygenerated ## The work and its context [Dis]entanglement makes two faces out of the ends of many lines. Horizontal bars enter from opposite edges, stop at uneven distances and narrow into threads. The stopping points collectively describe foreheads, noses, lips and chins. Between them, the threads cross, gather and sometimes tighten into conspicuous knots. The figures are therefore connected by the very marks that give them an outline. Recognition begins at the edges; the difficulty of following a connection occupies the middle. [^S05] [^S06] [^S07] [^S08] onlygenerated released the 730-work series through Art Blocks Factory in July 2021. The project statement connects it to separation after a long marriage and describes different stages of attachment and departure. This gives the threads an explicitly interpersonal meaning, although it does not dictate a single story for every output. The [onlygenerated artist dossier](https://ab5d.xyz/artists/onlygenerated/) provides broader context for the practice. This essay concentrates on how the series makes its relationship metaphor visible. [^S01] The images are still compositions. Their movement belongs to the eye, which can attempt to follow a strand from one face to the other and lose it among crossings. A bright palette does not necessarily make that experience uncomplicated. In the inspected examples, saturated violet, electric blue and yellow-green accompany arrangements ranging from open space to dense central tangles. Colour and connection operate together without becoming a simple scale from happiness to distress. [^S02] [^S05] [^S06] [^S07] [^S08] ## Conception and development The artist's published account distinguishes the ordinary difficulty of separation from circumstances in which leaving is unsafe or materially constrained. It names abuse, financial hardship and cultural expectations as possible barriers. The accompanying pledge allocates ten percent of artist proceeds to InterAct of Wake County. These statements establish the work's stated point of departure and social concern. They do not provide a dated sequence of prototypes or identify the real people, if any, behind particular generated profiles. [^S01] InterAct's own website identifies its work with people affected by domestic and sexual violence. That independently supports the relevance of the named organisation to the project's subject. It does not verify a payment from this edition. Keeping those questions separate matters: an artist's declaration, a charity's activities and a completed transfer are three different kinds of evidence. [^S10] The visual decision to withhold a filled face is particularly effective. Each person appears through a boundary shared by numerous separate marks, rather than through a single continuous outline. Individuality is both clear and precarious. Small changes of length distinguish a nose from a mouth, while the middle of the composition can overwhelm those distinctions with connection. This is an interpretation of the inspected images and profile tables, not a recovered statement about the artist's design process. [^S02] [^S05] [^S08] The project's linked website could not be securely retrieved during this research because its certificate had expired. The official statement remains the principal inspected source for conception. No influence, development duration or autobiographical detail beyond that statement is asserted here. The absence of a process archive limits how far a formal reading can be turned into a history of decisions. [^S01] ## How the work was constructed The complete indexed script uses p5.js 1.0.0. A numeric seed is parsed from the opening portion of the token hash. Its random function repeatedly applies bitwise shifts and returns a bounded value, with helper functions converting those values into intervals or integers. A separate weighted-choice function accumulates supplied weights before choosing an option. Consequently, the program's apparent spontaneity comes from a repeatable sequence for a given token input. [^S02] Two collections of numeric arrays supply profile shapes. Each collection contains four alternatives, with one array shared between them. These are explicit shape data in the program, not facial images fetched from another service. A selected table determines how far the bars extend into the canvas at successive positions. Reversing a table can turn a profile upside down. When the two reversal flags differ, the composition also rotates through a right angle. These transformations help explain the vertical arrangement in edition #10. [^S02] [^S07] The canvas is square and sized from the smaller viewing dimension. Bar spacing is a fraction of that size; the program derives a count from the resulting interval. Profile values, margins and a small progressive displacement determine each bar's endpoint. Across the field, a list of row indices is partially shuffled. The shuffle's strength changes how far a connection can travel from its original row. Thus two readable contours can remain stable while their connecting lines become substantially reordered. [^S02] The central threads use Bézier curves, whose intermediate control points bend the line between its endpoints. In one branch, a single curve joins the two sides. In the knot branches, two successive curves pass through a shared intermediate region. Several such regions can be selected, and small offsets prevent every thread from following exactly the same path. The visible knots are therefore constructed concentrations of curves. The code does not calculate physical rope, friction or forces between strands. [^S02] Another condition controls whether a central connection is drawn at a particular row. This can leave many bars facing an open middle while only a smaller group remains joined, as in #1. The endpoints themselves are still present. Absence becomes a designed part of the composition rather than a missing render: the viewer can compare the many potential connections with those actually traced. [^S02] [^S06] Colour is assembled through hue, saturation and brightness, with additional lightness adjustments for the background. Conditional branches produce related or more varied line colours, dark or pale treatments, and reduced colour variation. The background may be a gradient alone or include translucent curves or circles. These are painted before the profiles. They supply a softer field behind the relatively exact horizontal bars and thin central threads. [^S02] Native canvas gradients carry colour along the bars. Small triangular or circular tips articulate their ends, while offset darker marks act as shadows. Additional broad curves can suggest hair at the margins. These features make the contour readable without filling in an entire head. Their depth is an optical effect of layering and displacement, not evidence of a three-dimensional scene or a lighting simulation. [^S02] All drawing is invoked during setup. There is no draw function, animation timeline, keyboard handler or pointer interaction in the inspected artist script. The normal work is static even though its curves suggest tension and motion. This research inspected the indexed script without executing it locally, and compared official previews. The digest retained in the evidence package identifies those script bytes; it does not constitute an independent comparison against a pinned blockchain retrieval. [^S02] ## The mint The collection page gives 11 July 2021 as the release date. The fresh index records the first token on 25 June at 16:29:06, activation on 1 July at 04:30:30, and both the second mint and scheduled start on 11 July at 17:00:25. Its completion field reads 18:05:17 on the latter date. The earlier first token should not replace the public release date, and activation is a third event. Full timestamps and offsets are preserved in the research record. [^S01] [^S05] [^S06] At observation, invocations and maximum invocations both equal 730. The price setting is 0.1 ETH. These are current indexed values, not proof of the originally announced maximum, the opening price or the amount paid by every recipient. Editions #1, #10 and #20 share the same recorded second, but that alone does not identify their buyers or distinguish public purchases from allocations. No exact public sellout duration is inferred from the completion timestamp. [^S01] [^S06] [^S07] [^S08] The donation pledge is documented in the project description, but this package contains no charity receipt or independently reconstructed payment trail. Its fulfilment therefore remains a research question. Similarly, no complete evidence of pauses, configuration changes or technical incidents was recovered. The available chronology can be reported without presenting those missing records as proof that nothing changed. [^S01] ## Copyright and licensing The current artist-set label is NIFTY License Version 2.0, with no terms URL in the project record. The standard NFT License website publishes a version 2.0 text dated November 2018. That text retains creator ownership and supplies conditional permissions for holders, including restricted commercial use. It is relevant comparative documentation, but the missing project link prevents this essay from treating every clause as independently confirmed project-specific terms. The caption should preserve the reported label and identify that limitation. [^S01] [^S03] No separate licence for the artist's program was identified in the inspected script. The p5.js 1.0.0 repository supplies the GNU Lesser General Public License version 2.1 for its dependency. That does not grant reuse rights in onlygenerated's artwork or code. This package redistributes neither the artist script nor the inspected preview files. Its original prose is CC0; ownership of an example does not transfer the artist's copyright. [^S02] [^S04] ## Close reading of four outputs Editions 0, 1, 10 and 20 were selected in advance as an early-edition comparison. Their full token identifiers are 97000000, 97000001, 97000010 and 97000020 on Ethereum contract 0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270. All four official previews were inspected as still images. This is a deliberately small comparison, not a representative sample or a survey of feature frequencies. [^S05] [^S06] [^S07] [^S08] In #0, violet bars press inward from both sides of a lavender field. Thin strands gather into a tall central braid, with a conspicuous pinching point near the middle. Pink and blue hair-like curves crowd the right edge, making the two sides visually unequal even though they share a strong horizontal rhythm. The faces are most legible when the eye follows the changing ends of the bars. Concentrating instead on the threads makes the contours recede. One image supports two different kinds of attention: recognition of a pair and scrutiny of what binds them. [^S05] In #1, blue and violet bars surround a much darker purple interior. Many terminate without a visible connecting strand. The remaining curves sweep toward a relatively small knot left of centre, leaving a broad open area above it. This is not simply a less colourful version of #0. Its sparse connections make the empty middle active. The faces seem held apart by distance while the few surviving lines still claim that distance as shared space. Hair-like bands on the left add density where the central field withholds it. [^S06] In #10, the arrangement turns vertical. Yellow-green bars descend from the top and rise from the bottom, carrying opposed profile contours into an olive field. The connecting curves make a loose horizontal tangle rather than one obvious tightened point. Large translucent circles soften the background. Although its indexed Knots value is zero, the image still contains many crossings. The distinction is useful: the named feature identifies a procedural option, not an assurance that the viewer will see no entanglement. [^S07] [^S02] In #20, electric blue threads and bars stand against a nearly black green field. Two concentrations of curves occupy different heights, making the central connection feel elongated and internally divided. The bright strands remain individually visible around the edges of these masses, then become hard to track where crossings accumulate. Unlike #1, this picture repeatedly pulls the eye back into the middle. The dark background strengthens that pull by making even a thin line appear emphatic. [^S08] The header example, #75, was separately verified in an AB5D wallet and visually inspected. Pink bars and marginal bands surround a pale lilac field. Relatively open curves cross without the dense central clusters of #0 or #20. The softer colour and generous intervals offer another balance of connection and separation, while retaining the same recognisable profile mechanism. [^S09] ## Significance and limitations [Dis]entanglement makes its subject inseparable from its construction. The marks that define each person also reach toward the other, so a relationship is neither an added symbol nor a background story. It is the organisation of the image itself. Across the inspected works, the distance between faces can become a field of slack curves, isolated links or concentrated knots. The system gives those differences a consistent visual grammar without requiring that each output be assigned one emotional verdict. [^S02] [^S05] [^S06] [^S07] [^S08] Unresolved research questions: - Recover dated development material and the unavailable project website. - Establish launch-price and original-maximum evidence, and classify reserved versus public mints. - Locate project-specific operative licence terms and any separate artist-code grant. - Verify the charitable transfers independently. - Compare the indexed script with a pinned chain retrieval and broaden visual inspection beyond the early-edition sample. [^S01]: Art Blocks / onlygenerated. [[Dis]entanglement project record and Art Blocks index](https://www.artblocks.io/collection/disentanglement-by-onlygenerated). Accessed 2026-09-18. [^S02]: onlygenerated. [[Dis]entanglement artist script](https://artblocks-mainnet.hasura.app/v1/graphql). Accessed 2026-09-18. [^S03]: Dapper Labs. [NFT License version 2.0](https://www.nftlicense.org/). Published 2018-11-05. Accessed 2026-09-18. [^S04]: p5.js contributors / Free Software Foundation. [p5.js 1.0.0 licence](https://raw.githubusercontent.com/processing/p5.js/1.0.0/license.txt). Accessed 2026-09-18. [^S05]: onlygenerated. [[Dis]entanglement #0](https://www.artblocks.io/token/1/0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270/97000000). Accessed 2026-09-18. [^S06]: onlygenerated. [[Dis]entanglement #1](https://www.artblocks.io/token/1/0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270/97000001). Accessed 2026-09-18. [^S07]: onlygenerated. [[Dis]entanglement #10](https://www.artblocks.io/token/1/0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270/97000010). Accessed 2026-09-18. [^S08]: onlygenerated. [[Dis]entanglement #20](https://www.artblocks.io/token/1/0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270/97000020). Accessed 2026-09-18. [^S09]: onlygenerated / Art Blocks. [[Dis]entanglement #75, AB5D collection](https://www.artblocks.io/token/1/0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270/97000075). Accessed 2026-09-18. [^S10]: InterAct. [InterAct of Wake County](https://interactofwake.org/). Accessed 2026-09-18.