A lens essay. The voice belongs to a critic with a Cinémathèque pass and a long memory for camera grammar. It reads the archive as if it were a director's reel.
Setting aside the discourse of AI image generation for a moment, look at the archive as you would look at a director's reel. What is the cinematographer doing? What are the lens choices? Where is the eye led?
The archive is, in its dominant register, not working in academy ratio or modern widescreen. The repeated compositions — Bond in costume, the Mad Hatter at center, Satan with his joint — are framed in square or near-square. This is not accidental. Square framing is the format of iconography, not of cinema. It is the format of the Russian icon, the tarot card, the album cover. The artist's compositional instinct is portrait-frontal, full body or three-quarter, no parallax, no rack focus. He is making icons, in the precise sense, not scenes.
When the archive does enter cinematic grammar, two traditions dominate. The first is German Expressionism, c. 1920–1930 — high-contrast black-and-white, severe shadow, distorted urban architecture, figures lit from below. The Mickey Inversion cycle is essentially Caligari with a corporate mascot. The cyberpunk Mickey, the dark Mickey, the Orwell Mickey — these are Murnau and Lang re-skinned. The intuition is correct: the way to make a sanitized figure threatening is to put him under the light Murnau used to make Nosferatu threatening. Thomas has rediscovered this trick by feel, without (as far as I can tell from the prompts) ever invoking the historical reference. The model is doing the citation for him.
The second is synthwave / neon-noir, c. 1981–1985 — the Ridley Scott Blade Runner palette, the Tron grid-of-light, the Michael Mann Thief sodium-vapor street. The Raccoon Bestiary's cartoon racoon moon synthwave run is, in cinematic terms, a Thief poster with a different protagonist. The Bond Sequence's cyberpunk run is Blade Runner costume design with a different lead. This is not the Blade Runner 2049 version — Roger Deakins's pastel desolation — it is the original 1982 version, the rain-and-neon version, the version that is iconographic rather than landscape.
Wong Kar-wai does not appear. This is striking. Wong's compositional vocabulary — narrow corridors, ambient blur, color blocking via wardrobe and architecture, the Chungking Express sense of a city as a series of inhabited interiors — is the closest mainstream cinematic match to the Cubist Skylines. But Thomas's cubist skylines have no figures in them. Wong's grammar requires figures inside the color blocks. The cycle stays at the architectural level and refuses inhabitation. This is a tonal choice with cinematic implications: the cubist cycle is production design without a movie. It is location scouting that never resulted in a shoot.
Tarkovsky does not appear either, but he should. The Evolutionary Diagrams cycle has Tarkovskian ambitions — slow contemplation, ordered nature, ranks of small forms studied in patient light. Stalker and Solaris both contain images that would sit comfortably in the Evolutionary Diagrams cycle. The cycle does not invoke Tarkovsky in its prompts, but the model's output drifts toward him whenever the prompt admits stillness.
Soviet montage does appear, repeatedly, in the propaganda runs. The Mad Hatter Obey posters, the Satan-as-Soviet-worker images, the Bitcoin Capitol seals — all are operating in the visual key of Eisenstein and Rodchenko, the high axial composition with a single dominant figure. This is agitation cinema repurposed as still iconography. The argument the cycles are making is that the axial frame and the single figure are the political grammar that survived the twentieth century — that whoever inherits this grammar inherits its rhetorical force, and that Thomas, drawing in 2024, has every right to claim it for his own causes.
The chase scene does not exist. This is among the archive's quieter properties. The Imagined Battles cycle stages clashes — armies meeting, single-frame engagements — but never a chase. There is no figure pursued by another figure across multiple frames. The archive does not believe in time-extended action. Every image is a moment, never a sequence. This is again the iconographic instinct overriding the cinematic instinct. Thomas is not asking what the figure does over the next minute. He is asking what the figure is, captured in one composition, at one tableau.
The dominant shot type is the medium full — figure visible head to knee, centered, addressed to camera. The archive almost never produces a close-up, almost never produces an extreme wide. Close-up is psychological cinema, asking the viewer into the figure's interior life. Wide is landscape cinema, asking the viewer to see the figure in their environment. Thomas wants neither. He wants the icon shot: the figure at the distance at which we can read their costume and their pose, and at no greater intimacy. This is the shot type of religious art, of saints' cards, of military propaganda. The cinematic equivalent is the publicity still — the standalone image whose job is to advertise the figure rather than to narrate them.
A useful exercise: imagine the archive as a director's portfolio submitted to a production company. The director's strengths would read clearly. Strong iconographic eye, formidable production design, comfort with multiple historical visual idioms, weak narrative instinct, no interest in continuity editing. The hire-or-pass conversation would be: yes, for music videos and title sequences; no, for features. This is a useful diagnostic. Thomas's work is the music video of bitcoin commentary, not the feature. It is meant to be encountered as a single arresting frame, looped on a stream, returned to. It is not meant to be read in narrative sequence.
That is a critical assessment, not a dismissive one. The visual culture of the 2020s is structurally short-form. The single-frame icon is the dominant unit of online attention. Thomas's archive is correctly tuned for its century, even when the visual references it borrows from are a hundred years old. He is making the right thing for the moment, with the tools he has, in the grammar that the moment demands.
The grammar is still cinematic. He just refuses the part of cinema that requires waiting for the next frame.