Color Magic Planets
Bård Ionson · Project 58 · Factory
Edition size: 256 · First mint: 27 April 2021

The work and its context
Color Magic Planets turns a sheet of sixteen circles into a field of chromatic experiments. Each circle contains a different encounter between smaller shapes: a triangle crossing an oval, a rectangle cutting into a disc, an arc opening a wedge of ground. The regular spacing makes comparison almost involuntary. The eye moves between cells, recognising a common vocabulary while noticing how differently its pieces can meet. In the four outputs examined here, the background also changes the apparent identity of the colours, from pale cyan and magenta on cream to intense red and blue on black. 1 2 3 4
Bård Ionson released the 256-work series through Art Blocks Factory in 2021. Its source is unusually intimate: an artwork by his daughter Liza Grace, made using translucent vinyl shapes from the Metropolitan Museum of Art's Color Magic Sticker Play Book. Ionson's account names her work Colorform Multiverse. The project carries that activity of arranging and overlapping pieces into a program whose decisions are determined by a token hash. The artist dossier situates this experiment within his wider practice. 5 6
The planetary title invites imaginary worlds, but these are not illustrations of astronomical objects. The discs have neither modelled terrain nor a calculated atmosphere. Their depth comes from intersections and the ambiguous impression of transparency. A small black region can read as an object in front, a hole, or the darkest part of several overlapping surfaces. Those changes of interpretation give a simple grid more activity than its fixed structure initially promises. 7 1 4
Conception and development
The artist describes childhood play and learning as the starting point. His account connects Liza Grace's composition to cyan, yellow and magenta vinyl pieces, including circles, arcs, triangles and rectangles. The importance of this origin is the operation it supplies: colour becomes something discovered by placing one shape across another. The digital work preserves that invitation to experiment even though a viewer cannot rearrange its pieces. 6
Ionson's explanation sometimes uses token address as shorthand for the input that makes each work distinct. The inspected program is more specific: it reads tokenData.hash. Keeping that distinction helps explain the translation from a handmade arrangement into a generative edition. The artist chooses the vocabulary and its constraints; the hash supplies repeatable differences within them. This is not evidence that the software reproduces his daughter's original composition exactly. 7 6
A later account, published in June 2023, describes a return to physical materials. Ionson proposed buying back tokens and making ten-inch-square framed works, with several backgrounds, using materials from the Met. That historical offer should not be mistaken for current availability. It nevertheless documents a useful reversal: a project derived from physical play could become a plan for physical objects again. The later page calls the source artwork Playform Multiverse, rather than Colorform Multiverse; the earlier account supplies the name used here. 8
How the work was constructed
The complete indexed script contains 2,601 characters. It uses p5.js 1.0.0 and begins with a square canvas fitted to the smaller viewport dimension. A 480-unit design provides the reference scale. Inside it, two nested loops place sixteen groups in four rows and four columns, with a common spacing and a large circle in every group. The program does not discover a planetary arrangement through simulation. It establishes a grid and then differentiates its contents. 5 7
The input is expanded before drawing. Thirty-two hexadecimal byte values are extracted from the token hash. A seed derived from its opening characters drives a bitwise pseudorandom sequence that supplies another 225 values. A moving index steps through the combined array. Because different shape branches consume different numbers of values, an early choice affects where later choices read their inputs. The decisions are connected through this sequence rather than assigned independently to sixteen fixed portraits. 7
Each group uses red, green and blue drawing colours. A threshold selects which pass receives the large circle; other passes draw smaller forms. Those fills do not necessarily appear as red, green and blue in the finished picture. The program uses EXCLUSION blending, which combines incoming and existing colour channels. The library describes this as related to difference blending, with less contrast. Consequently, the background participates in the calculation of the visible colours. Apparent transparent layers arise from this compositing rule, not from an inspected optical model of translucent vinyl. 7 9
Five branches supply the smaller vocabulary: a circle, an oval, an arc, a triangle or a rectangle. Dimensions and coordinates come from the working values. Triangle coordinates receive small adjustments when selected coordinates are too close, while rectangles have a minimum dimension added to their variable size. The shared scale keeps these operations recognisable across the sheet, even when a thin triangle is almost lost beside a broad circular overlap. 7
The arc branch contains an unusual comma expression whose final value is 60. In JavaScript that final value becomes the start-angle argument; an earlier expression still consumes an input value. The endpoint comes from another working value. The script does not change p5's default angle convention. It would therefore be inaccurate to describe this branch as a straightforward menu of conventional semicircles merely because some outputs resemble wedges. This reading follows the retrieved expression without correcting or replacing it. 7
A further conditional can add a compact saucer-like arrangement of ellipses or an arc. Its test compares a shape-selector value with the advancing index, not simply with another byte. The artist associates spacecraft with The Entity, but neither the code inspection nor four examples establishes their edition-wide frequency. Availability in a branch and measured rarity across 256 finished works are different claims. 7 6
The normal output is static: setup calls noLoop, and the inspected script contains no animation or interaction handler. The original generators were opened for the visual comparisons below. This verifies their appearance in the inspected environment, while the code account remains an analysis of the indexed script, not an independently pinned blockchain retrieval. 7 1 2 3 4
The mint
The artist's retrospective account identifies May 1 as the release and July 2 as the sellout. It also reproduces an April 28, 2021 announcement naming a May 1 drop. These statements support the official collection's 2021 release chronology. The present page carrying that account displays March 7, 2022 as its publication date, so its surrounding commentary should not all be treated as a contemporaneous launch report. 5 6
The index records a first token on April 27, followed by activation later that day. The second mint and scheduled-start fields fall on May 1, and completion on July 2. Those fields are retained separately in the research record. The earlier token does not invalidate the public release announcement, but its purpose has not been established here. No artist, reserve or administrative allocation is inferred solely from its position in the sequence. 5
Current invocations and maximum both read 256, consistent with the artist's stated edition. The observed price setting is 0.1 ETH. That configuration is not independently verified as the opening price or as the amount paid by every recipient. The broad sale period is supported by the artist's account; an exact public-sale duration would additionally require transaction classification and launch-era records. 5 6
Copyright and licensing
The current project record names NFT License 2.0 and links niftylicense.org, which redirects to the NFT License site. Its displayed Version 2.0 text reserves intellectual-property ownership to the creator or licensors. It grants specified uses to NFT owners, including personal noncommercial use and conditional merchandise use capped at $100,000 in annual gross revenue. Other provisions govern marketplace and third-party display, modification and transfer. The owner licence ends when the token is transferred. This is an ownership-linked permission structure, not a public-domain dedication or an unrestricted copyright transfer. Earlier terms have not been reconstructed. 5 10
No separate licence for reusing Ionson's artist script was identified in the inspected material. The p5.js 1.0.0 dependency carries its own LGPL 2.1 text, which does not establish permission for the project program. AB5D's CC0 dedication concerns the original essay prose. The header's artwork caption identifies the reported NFT licence separately and links its terms. 7 11
Close reading of four outputs
The comparison uses editions 0, 1, 10 and 20, chosen as a small early-edition group. Their token identifiers are 58000000, 58000001, 58000010 and 58000020 on Ethereum contract 0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270. All four original generators were visually inspected. They demonstrate particular relationships, not the statistical distribution of the series. AB5D's separately inspected #43 supplies the header example. 1 2 3 4 12

In #0, a pale cream-yellow field supports cyan, magenta and yellow discs. The upper row already gives several versions of containment: a wedge crossing a cyan circle, a pointed interruption inside magenta, and smaller concentric forms cut by a rectangle. Near the bottom, narrow ovals and rectangular overlaps make darker centres. The fixed outer circles establish an orderly sheet, but the inner shapes resist becoming a repeated symbol. Some are enclosed like samples; others seem to cut their container open. 1

In #1, yellow becomes more prominent across the warm cream field. Red, green and dark intersections gather inside the circles, making their centres feel busier than the ground around them. A magenta ring at the lower left contains a very thin vertical cyan oval. That slender form is useful beside the broader overlaps elsewhere: a small change of proportion alters the apparent weight of a cell without changing its allotted space. The sheet's rhythm depends on these local inequalities. 2

In #10, cyan dominates. The upper-right circle contains a tiny green centre crossed by a dark horizontal line, leaving most of its interior open. Directly below, a much larger dark disc is surrounded by green and cyan rings, with a blue wedge entering it. The contrast is between almost empty and almost occupied versions of the same circular area. At the lower right, a magenta perimeter survives around broad cream openings and pointed coloured fragments. It seems less like a filled planet than a structure assembled around absences. 3

In #20, the black ground changes the entire sheet's temperature. Large red, blue and green circles appear sharply separated, while white, cyan, yellow and magenta intersections brighten their interiors. A white centre in the second row reads almost as a light source within a red disc. Farther down, a cyan horizontal oval crosses a magenta rectangle within blue, pulling attention sideways against the circular boundary. These are visual readings of a static composition, not claims that the program simulates illumination or moving bodies. 4

The held #43 offers another background comparison. Its gold field surrounds red, green and pale pink circles. Blue centres within magenta rings become conspicuous, while a dark upper shape makes one green cell unusually heavy. At the bottom right, a blue triangle sits inside cyan with a green rim. The example makes the colour rule visible through contrasts with the cream and black works, while remaining a distinct arrangement of shapes rather than a simple recolouring of them. 12
Significance and limitations
Color Magic Planets gives a household act of arranging coloured pieces a durable procedural form. The grid makes the work easy to enter, and the overlap rule keeps it from being exhausted by a quick count of its components. Its strongest relationship is between a stable container and an unstable reading of what happens inside: circle, window, body, opening or layered patch. Those interpretations are the critic's, grounded in the inspected outputs; they are not additional feature names assigned by the artist. 7 1 3 4
The documented passage from a daughter's vinyl composition to a digital edition, and later toward framed material objects, gives the project a history beyond its mint. Further research should recover launch-price evidence and allocation classifications, compare the script with a pinned chain retrieval, investigate earlier licence records, and broaden the visual sample. Those gaps limit historical and distribution claims. They do not prevent close attention to the specific ways sixteen circles can become sixteen different problems of colour. 5 6 8
References
-
Bård Ionson. Color Magic Planets #0. Accessed 2026-09-18. ↩
-
Bård Ionson. Color Magic Planets #1. Accessed 2026-09-18. ↩
-
Bård Ionson. Color Magic Planets #10. Accessed 2026-09-18. ↩
-
Bård Ionson. Color Magic Planets #20. Accessed 2026-09-18. ↩
-
Art Blocks / Bård Ionson. Color Magic Planets project record and Art Blocks index. Accessed 2026-09-18. ↩
-
Bård Ionson. Color Magic Planets. Published 2022-03-07. Accessed 2026-09-18. ↩
-
Bård Ionson. Color Magic Planets artist script. Accessed 2026-09-18. ↩
-
Bård Ionson. Color Magic Planets IRL. Published 2023-06-17. Accessed 2026-09-18. ↩
-
Dapper Labs. NFT License, Version 2.0. Published 2018-11-05. Accessed 2026-09-18. ↩
-
p5.js contributors / Free Software Foundation. p5.js 1.0.0 licence. Accessed 2026-09-18. ↩
-
Bård Ionson. Color Magic Planets #43. Accessed 2026-09-18. ↩
Written by GPT-6 Astra for AB5D. Original prose: CC0. Artwork and code rights are addressed separately above.
Batch usage: 6,037,874 tokens, including cached input, across Color Magic Planets research and drafting, plus preceding final checks and browser verification.
Token usage
6,027,857 input tokens (including 5,957,632 cached) and 10,017 output tokens. Measured shared interval including cached input. Work covered by preceding shared intervals is not counted again here. This is not an isolated per-essay cost. Later deployment excluded.
Artist dossier · Source record · Essay data
Research record
The indexed project script was inspected. Runtime observations and limits are documented in the research record. Historical price, allocation and licensing questions remain as identified in the essay. Publication does not resolve those evidence gaps.
