Chimera

The work and its context

Chimera makes a still life that cannot quite remain still. Flowers, vessels and occasional table objects gather into recognisable arrangements, yet their surfaces behave like loose marks continually renegotiating the objects they describe. A flower can become a dense cup of colour, then thin sufficiently for its stem or a neighbouring bloom to appear. The scene remains identifiable while its local evidence changes. In the five original renderings examined here, this instability is the work's central pleasure. The eye alternates between following individual strokes and accepting the bouquet they temporarily compose. ^{1 2 3 4 5}

mpkoz released the 987-work project through Art Blocks Curated in January 2022. Its own description places still life within a history of media absorbing and extending earlier forms. Chimera takes that proposition literally at the level of viewing: it borrows the familiar address of a floral picture, then allows its arrangement to unfold in time and respond to a moving viewpoint. The artist's wider practice is documented in the mpkoz dossier. ⁶

This is more specific than putting motion into a painting. The apparent brushmarks belong to a spatial construction, while their treatment retains the broken surfaces of a picture. Dragging the collection example changes overlaps without turning the image into the smooth, fully enclosed objects associated with conventional three-dimensional graphics. Neither description, painting nor model, completely settles the encounter. That unresolved relation gives the title an observable formal basis without requiring the viewer to accept a theory of technological progress. ^{7 5}

Conception and development

In the release interview, Michael Kozlowski connects flowers with childhood experience and his mother's gardening. He also describes learning creative coding at the University of Southern California and developing work across digital forms before this release. Those recollections give the floral subject a personal continuity. The decision to use it is not explained solely by the arrival of a new sales platform. ⁸

The interview presents Chimera as an effort to join still life and generative construction. Its account of mathematical flower-making is useful, but should be read alongside the deployed procedure. The code generates geometry and marks rather than importing a finished bouquet. Its material is a set of rules for forming a scene, not a stock photograph filtered into an animated style. ^{8 7}

The distinction between an account of an idea and evidence of its implementation matters here. The interview's shorthand about models and textures should not become a literal inventory of absent technical components. The program does construct three-dimensional geometry. It also synthesises a stroke-mask texture and includes a small background-grain image. These findings do not undermine the procedural achievement. They describe it more accurately: the flowers are built by the work's own procedures, and the painterly surface is itself engineered. ⁷

How the work was constructed

The inspected program uses Three.js r124. A random-number class derives two internal generators from different portions of the token hash, warms them through a combined million calls, and alternates their subsequent values. These streams supply the major choices and variations. The resulting system has a persistent token-specific structure, but it would be too strong to promise that every pixel is identical on every viewing. Separate calls to the browser's unseeded random function occur in generated stroke-mask details and some accessory geometry. Rendering conditions also affect the result. ^{6 7}

Flowers emerge through different trajectory-building routines. Stems, petals, centres and clustered forms are not one universal mesh recoloured for each species. The orchid routine, for example, builds moving trajectories through interacting directional quantities, while other routines organise radial petals or clustered elements. These procedures place many small drawing elements in space. The distinction between flower types therefore affects how marks accumulate and how a contour becomes legible, rather than merely adding a botanical name to the token's metadata. ⁷

The bouquet's arrangement supplies another layer of authorship. Candidate positions are distributed around a rounded volume using angular steps, then altered through scaling, rotation and the selected species. The vase and several accessories use geometry generated from profiles or other simple constructions. Placement checks keep accessories from occupying the same immediate space as the vase or one another. These constraints produce a recognisable still-life situation without requiring a separately hand-arranged model for every edition. ⁷

The program's principal drawing economy is instancing: many marks share a small underlying geometry while carrying their own position, colour, size and timing attributes. A generated canvas supplies a strip of stroke masks. The fragment shader uses this mask to give a mark a broken edge instead of presenting every instance as an obvious rectangular plane. The vertex processing adds noise, life-cycle changes and size variation. These are technical reasons that a mass of repeated elements can look brushed rather than mechanically tiled. ⁷

The camera and the surface treatment pull in different directions. A narrow-angle perspective camera views the spatial arrangement, but parts of the stroke treatment depend on the screen. As the viewpoint changes, the flowers offer depth while the marks retain a pictorial insistence on the viewing surface. The code thus explains an important observation from #765: turning the bouquet reorganises the overlaps, yet does not resolve the image into polished sculpture. Its roughness is not a temporary failure to finish rendering. It is part of the rendering method. ^{7 5}

Time is equally structural. The scene is assembled progressively through queues of instances. After that initial accumulation, marks continue to change through the shader and animation state. In the observed originals, petals partly withdrew and returned, making one apparent flower count or silhouette unreliable as a complete description of the edition. Animation begins without a click. Automatic rotation is a separate option, disabled by default; the documented R control toggles it, while A controls animation. Dragging #765 was tested and changed the viewpoint. The screenshot controls were inspected in code but not exercised during this review. ^{6 7 1 5}

The rendering environment deserves a precise place in the record. The program contains a special headless-browser path that changes normal animation and construction behaviour. It also adapts to low frame rates by reducing aspects of rendering quality. The five works were therefore inspected in an ordinary browser using the official generators, not through a headless screenshot run. No audio mechanism was found in the inspected artist script. Preserving Chimera involves retaining the token input, program and compatible graphics environment, as well as recognising that one preview cannot document every temporal state. ⁷

The mint

The official collection page gives 10 January 2022 as the release date. The index records the first token earlier, on 17 December 2021 at 22:40:29, and activation on 20 December at 21:15:08. Its scheduled start is 10 January at 18:00:00; the second indexed mint follows at 18:00:19. Completion is recorded at 18:13:11 that day. Full timestamps with offsets are retained in the evidence record. The December token is evidence of an earlier mint, not a reason to replace the documented January release date. ⁶

Current supply and maximum both read 987. The sales notes describe a descending Dutch-auction ladder beginning at 3.5 ETH and ending at 0.1 ETH, with intermediate steps including 2.9 and 2.4 ETH. The current price field reads 2.4 ETH. Those observations identify the recorded sale structure and configuration; they do not establish what each recipient paid or prove that every listed auction step was reached. An account of proceeds would require transaction-level payments and allocation classifications. ⁶

Similarly, subtracting scheduled start from completion yields an interval between index fields, not an independently reconstructed public sellout time. The first token, public launch, configuration and final indexed mint answer different questions. Keeping them distinct makes the chronology useful without giving it an accuracy the available evidence cannot support. ⁶

Copyright and licensing

The current artwork label is CC BY-NC 4.0. The metadata does not provide a terms link, so the named Creative Commons deed is linked separately. That licence allows sharing and adaptation subject to its conditions, including attribution and the noncommercial restriction. This review has not reconstructed the history of earlier artwork labels. Ownership of the displayed token and the terms for reusing its imagery are separate matters. ^{6 9}

Three.js r124 carries an MIT licence. The shader's credited noise implementation also has an upstream MIT notice. Those dependency permissions do not establish a comprehensive licence for mpkoz's entire artist program. No separate grant covering that complete program was established in the inspected material. The code was read for analysis and its retrieval digest retained; it is not republished wholesale with this essay. AB5D's CC0 dedication applies to the original prose. ^{7 10 11}

Close reading of five outputs

The comparison uses early editions #0, #1, #10 and #20, chosen before visual inspection, together with #765, whose owner was verified against all three linked AB5D wallets. Each official generator was viewed at more than one moment. This is a close-reading sample, not a statistical account of the 987 works. The collection example adds a practical link between the archive's holdings and the analysis without being treated as aesthetically representative of every edition. ^{1 2 3 4 5}

In #0, cream and peach strokes sweep across a lavender upper field, while a blue lower plane swirls beneath the vase. The background is insistently active before the flowers are considered. Pink upper blooms, a pale flower at left and a dark cobalt mass at right compete against those surrounding currents. The mauve vase gathers the stems without providing an entirely still base. Its own marks belong to the same unsettled vocabulary. Figure and setting remain distinguishable, but their shared directional energy prevents the bouquet from becoming a cut-out object pasted onto a neutral ground. ¹

A later moment makes the distinction between a moving surface and a changing visible structure clear. The large blue bloom has partly dissolved, exposing green stems and a neighbouring pink flower. The upper pink mass is narrower. The composition has not simply rotated intact: what can be seen through it has altered. This makes looking cumulative. An earlier view supplies a memory of fullness against which a later, more open bouquet is read. The flowers can appear fragile without the program having to simulate botanical decay. That fragility is an interpretation of their changing visibility. ¹

In #1, the paint becomes paler and more diffuse. Two dominant pink-white flower heads occupy unequal heights, one high at left and one lower at right. Dark rose edges define their folds, while a light yellow-green vase and green stem give them a softer supporting axis. A pink fruit at left and a shallow pink bowl to the right extend the composition laterally. The dusty pink ground carries broad pale loops and scattered blue-grey marks, which keep crossing the eye's attempt to separate object from atmosphere. ²

The thin blue border in #1 is unusually consequential beside those dissolving edges. It supplies a firm outer boundary to a field in which nearly everything else seems permeable. Later observations retained the high-left and low-right arrangement while the petal contours and intervening marks shifted. Compared with #0, the sense of motion is less a struggle among emphatic masses than a gentle uncertainty about where a surface ends. The work's Watercolor feature label is consistent with that appearance, but the reading comes from the original rendering rather than the label alone. ²

In #10, strong blue orchid forms stand above a broad brown vase striped by thick cream passages. Dark recesses give the petals a folded, almost cavernous volume. A blue bottle on the left and low fruit-like forms spread the weight across the foreground. Pale blue and white strokes occupy the upper field; pink and blue movement below suggests a supporting plane without making it inert. In a later frame, the lower-left floral extent recedes and more space opens beside the vase. Here the system produces a heavier still life whose volumes remain legible through the broken strokes. ³

In #20, a squat dark blue vase anchors yellow and olive flowers against a broad peach-and-cream halo. A pale horizontal strip, yellow at its edges, supplies a table-like base. The arrangement is more frontal and concentrated than #10. Two dominant heads were readily separable in the observed moments, although the metadata records three flowers; the image should not be forced into a count that occlusion and animation make uncertain. Between frames the visible petal masses and vase edges changed, while the dark vessel continued to hold the composition together. ⁴

AB5D's #765 removes much of the busy setting. A pale blue-green field and thin rust-coloured frame leave the bouquet relatively isolated. Cream curled forms meet orange-brown blooms, with a yellow accent at left and a red-pink vase below. The separation makes gaps among the flowers particularly important: when a pale mass thins, green stems and distant petals become newly legible. Dragging the image changes those relations again. Its quiet ground allows the viewer to attend to the paradox at the centre of Chimera, a spatial arrangement whose surfaces continually threaten to become independent marks. ⁵

Significance and limitations

Chimera gives a familiar subject enough stability to make procedural change perceptible. A vase, flowers and supporting plane are easy to recognise, so the viewer can notice how their relationships are unsettled by duration and viewpoint. The work's achievement lies in making those two kinds of attention coexist. One can enjoy a floral composition and also see the rules that prevent it from closing into a single final picture. The five inspected outputs sustain that relation through markedly different densities, palettes and settings. ^{7 1 2 3 4 5}

Further research should compare the indexed program with a pinned chain retrieval, test a wider range of graphics environments, and reconstruct the auction from classified mint transactions. A broader visual sample could investigate unusual flower and accessory combinations without turning the present close readings into a rarity survey. The evidence here establishes a specific relationship among procedural geometry, painterly marks and observable motion. It does not establish identical pixels across machines, a complete sales history, or a comprehensive history of rights changes. ^{6 7}