Event Horizon Sunset (Series C)

The work and its context

Event Horizon Sunset (Series C) makes a surface appear where the program has drawn only curves. Against black, thin coloured passages gather into something like folded film: partly transparent, edged with light, and difficult to place in depth. The four examples examined here suggest sails, loops and pockets, but none settles into an identifiable object. Their strongest common feature is the way a narrow bright edge can give a dim, almost disappearing region an apparent physical presence. ^{1 2 3 4}

Jake Brukhman released the 500-work series through Art Blocks Factory in 2021. The official collection dates its release to April 27. Its description connects shape and colour to Perlin noise and identifies two earlier Sunset series containing five and nine works. The Jake Brukhman dossier provides broader context, while this essay concentrates on the particular drawing procedure and its visible consequences. ⁵

The title invites an astronomical reading, yet the inspected program contains no model of gravity, a star or a black hole. Even the sunset is an association rather than a depicted horizon dividing land and sky. The picture's depth is produced within a flat field, through the accumulation and overlap of curved strokes. That distinction does not weaken the title. It leaves room for a viewer to experience the forms as atmospheric without mistaking them for scientific visualisations. ^{6 1 4}

Conception and development

The available project statement gives a concise developmental sequence: Series C follows two smaller groups and introduces the possibility of close proximity. The earlier groups have not been independently compared here, so the statement supports a history of successive series, not a reconstructed account of every change between their programs. No additional studio chronology is inferred from the lettering of the title. ⁵

The statement also proposes that a machine-learning classifier could select preferred results from the large space of possible sunsets. This is phrased as a possibility. The complete inspected Series C script contains no classifier, training process or external model call. Its visible output is generated directly from the token input. The difference matters because a remark about potential selection should not become a claim that these 500 works were chosen by artificial intelligence. ^{5 6}

What can be established directly is an economical formal idea. A curve changes a little, is drawn again, and changes again. The repeated lines become dense enough to suggest a continuous surface, while their colour changes slowly along the sequence. This is a development within the image itself, recoverable from the program even where biographical evidence is limited. It supplies a firmer basis for interpretation than inventing a specific astronomical or art-historical influence. ⁶

How the work was constructed

The full indexed script contains 1,770 characters and uses p5.js 1.0.0. It creates a square canvas fitted to the smaller viewport dimension, paints the background black, disables fill, and calls noLoop. The drawing therefore completes as a still. There is no inspected animation, pointer response or keyboard control. The translucent appearance belongs to a static accumulation of lines, rather than to an object moving through a three-dimensional scene. ^{5 6}

A seed comes from the opening characters of the token hash and initializes the noise generator. A separate bitwise function maps the seed to a bounded scalar. Crucially, that function changes a local parameter: calling it again with the same seed returns the same value. It does not advance a shared random state. The program reuses this scalar to set the number of drawing passes, the spacing between noise samples and the proximity branch. These aspects of variation are consequently coupled. ⁶

Each pass draws a cubic Bézier curve. Such a curve has two endpoints and two control points; the controls influence how it bends between its ends. Here, eight noise-derived coordinates supply those points. The control-point coordinates remain within a scale based on the canvas dimension, while the endpoint coordinates use an additional offset. The library documentation explains the drawing primitive; the particular arrangement of coordinates is specific to Brukhman's script. ^{6 7}

A parameter begins at zero and increases by 0.001 after every pass. Noise is sampled at this parameter plus several different offsets. Nearby inputs produce related values, so neighbouring curves change gradually rather than jumping to unrelated positions. That continuity makes the stack of strokes read as a coherent sheet. It does not require the program to construct a mesh or calculate a surface normal. The impression of a surface emerges from closely spaced two-dimensional marks. ^{6 8}

The nominal pass bound is 750 plus 2,750 times the derived scalar. An integer loop counter runs while it remains below that bound, so a fractional bound is not itself a fractional stroke. Every pass uses a thin line whose width scales with the canvas. Its opacity is low, allowing overlaps to build visible intensity against the black ground. The red, green and blue channels are also sampled from noise at different offsets. Gradual geometry and gradual colour therefore develop together. ⁶

The normal endpoint offset is proportional to 15 percent of the canvas dimension. In the close-proximity branch it becomes proportional to 150 percent. The broader range can place endpoints beyond the visible field while leaving the controls on their original scale. This changes the curves' relationship to the frame without moving a virtual camera. It is a geometrical route to apparent nearness. The four selected examples are all indexed as faraway, so the branch explanation is code-based rather than a direct comparison with a close-proximity output. ^{6 1 2 3 4}

Because the same scalar governs the pass bound and the proximity test, it would be misleading to treat those properties as independent switches. Equally, a threshold in code is not a measured percentage of the minted edition. Establishing the realised distribution would require examining the complete token set. This essay records the available mechanism and the small sample actually inspected. The script's digest identifies the indexed text used, but it has not been checked against a separately pinned blockchain retrieval. ⁶

The mint

The official release date is April 27, 2021. The index records a first mint at 13:24:36 that day and a second mint at 13:32:47; the scheduled-start field matches the second timestamp. Activation is recorded later, on May 1. The different field names are retained because replacing the April mint evidence with the later activation value would obscure the chronology rather than clarify it. ⁵

Completion is recorded on July 2. Current invocations and maximum both read 500, matching the collection's edition statement. The observed price configuration is 0.04 ETH. That is not independently established here as the opening price, a constant price throughout the sale, or the amount paid by every recipient. Nor do the inspected records distinguish every public purchase from artist or administrative allocation. An exact public-sale duration is therefore not asserted. ⁵

AB5D's header example, #165, was minted on June 25 according to its token record. Its freshly checked owner matches an AB5D wallet. This supports its use as a collection example and places it within the recorded mint period; it does not supply missing launch-era announcements or transaction classifications. ⁹

Copyright and licensing

The current licence field contains a link to niftylicense.org rather than a separately named licence version. The linked site leads to the NFT License text, where Version 2.0 is displayed. That text reserves intellectual-property rights to the creator or licensors and grants specified permissions to owners, including personal noncommercial use and conditional merchandise use. The latter includes an annual gross-revenue limit. Permissions are subject to restrictions and end with transfer of the relevant token. This is not a public-domain dedication. The observation establishes the current linked text, not a complete history of the artist's terms. ^{5 10}

No separate grant for reusing the artist script was identified. The p5.js 1.0.0 library includes LGPL 2.1 terms, which remain distinct from both the artwork permission and the artist program. The essay's original prose is dedicated to CC0; the artwork caption carries its own licence identification and link. ^{6 11}

Close reading of four outputs

The comparison uses editions 0, 1, 10 and 20, with full token identifiers 60000000, 60000001, 60000010 and 60000020 on Ethereum contract 0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270. Their original generators were opened and visually inspected as still works. The early-edition sample was selected in advance and is not representative of the distribution of all 500 outputs. AB5D's #165 was separately inspected for the header. ^{1 2 3 4 9}

In #0, a tall violet passage rises almost vertically, with a narrow edge carrying more light than the broad surface beside it. An ochre form crosses diagonally, making the two appear to pass through one another. The lower violet tip becomes so slender that it seems to dissolve into the black. The picture's tension lies between that fragile edge and the broad, dim regions it makes perceptible. Without the edge, much of the apparent volume would be difficult to locate. ¹

In #1, a smaller blue-violet loop tilts upward to the right. A pale green ribbon crosses its middle, while a large area above remains empty. The form feels suspended partly because it does not occupy the available field evenly. Its translucency also complicates front and back: following the blue contour suggests one ordering, while the green crossing offers another. These are perceptual readings of layered lines, not evidence of an underlying object with fixed physical sides. ²

In #10, the activity becomes more horizontal. Olive, cyan and blue folds converge around a compact central coil. A pointed projection extends to the left, while a greenish lobe drops on the right. Compared with #0's long vertical pull, this arrangement feels gathered and compressed. Several pale intersections briefly expose the line structure inside the apparent film, allowing the eye to move between reading a surface and reading the marks that construct it. ³

In #20, violet folds occupy the upper region above a pale green, rounded pocket. Thin pinkish edges and crossing strands connect the two. The lower form seems to contain space even though no solid boundary is drawn around it. Its apparent volume depends on the relationship between a luminous contour, a translucent interior and the surrounding black. Beside #10, it gives the same limited means a more vertically layered organisation. ⁴

The held #165 is quieter in colour. A slender pale blue projection rises above muted brown and violet folds, with a narrow bright crossing near the lower middle. Large black margins make its delicacy conspicuous. It is a useful collection example because the image does not need an intense palette to establish the series' central effect: a few concentrated edges can make an almost invisible field appear folded in depth. ⁹

Significance and limitations

Event Horizon Sunset (Series C) shows how little machinery is required to make a flat drawing spatially ambiguous. The code's economy is not by itself an artistic achievement; what matters is the relationship it creates between continuity, overlap and disappearance. The inspected works repeatedly make the eye infer more surface than it can securely see. Their black grounds remain active, setting the threshold at which a dim passage becomes visible. ^{6 1 2 9}

The earlier Sunset series, launch-price history, allocation categories and historical rights records need further research. A broader sample should include close-proximity works and compare the indexed program with pinned chain evidence. Those limits constrain the historical account and any distribution claims. Within them, the four readings establish a specific formal argument: closely related curves can sustain images that feel continuous while remaining difficult to resolve into stable objects. ^{5 6}