Vasya the Reformer

«Вася-реформатор» is a NEP-era satire shot in 1926 at the First Film Studio VUFKU (Odesa) — six parts, 1,603–2,030 metres. It premiered on 17 December 1926 in Kyiv. The directors were Favst Lopatynskyi and Oleksandr Dovzhenko; the screenplay was written by Dovzhenko, making this his first film work in cinema. No footage survives — the film is lost.

Vasya, a factory worker’s son and a pioneer, goes looking for adventure. He pulls a drunk out of the water — and the drunkard tries to beat him. Hiding from the scuffle, Vasya climbs into a parked car that turns out to belong to the zavhos (farm manager), who had reported the very same vehicle stolen. The boy exposes crooked priests in front of their congregation; the church is converted into a cinema and the priest ends up as its projectionist. The final deed: Vasya catches a thief at home and hands him over to the militia.

The summary above follows the filmographic description published by VUFKU, which reconstructs the film from its editorial script, montage lists, and period press. The cast included Yurii Shumskyi (Pylyp Sotka, Vasya’s uncle), Stepan Shahaida (the bandit Mytko Kutsyi), Dmytro Kapka (the zavhos), and young Vasia Liudvynskyi in the title role; the camera was handled by Joseph Rona and Danylo Demutskyi, with sets by Ivan Suvorov.

The phrase is genuinely ambiguous, and the ambiguity confuses search results to this day. Per VUFKU, «Вася-реформатор» is Oleksandr Dovzhenko’s first work in cinema: he wrote the screenplay and co-directed with Favst Lopatynskyi. Later in 1926 came “Ягідка кохання” (Love’s Berries) — the first film Dovzhenko directed on his own. So: first film work — «Вася-реформатор»; first solo directorial credit — «Ягідка кохання»; and “Сумка дипкур’єра” (1927) followed as his third feature-length engagement of the silent period.

This page uses the formulation “first film work” for «Вася-реформатор» because it matches the archive’s own wording — and because the film’s disappearance is exactly why the credit is so often garbled.

The surviving filmographic card records a cast that today reads as a cross-section of early Ukrainian cinema: young Vasia Liudvynskyi in the title role, Yurii Shumskyi — later a star of Ukrainian Soviet theatre and film — as the uncle Pylyp Sotka, Stepan Shahaida as the bandit Mytko Kutsyi, Dmytro Kapka as the zavhos.

For the reconstruction this list is a working tool rather than a ceremony: the cast defines the character plates. Shumskyi’s and Shahaida’s faces are known from their later films, so the reference images have something to anchor on — unlike the sets, which have to be imagined from text alone.

No frame of the original film survives. What remains is paper and press: the editorial script and the film treatment (published as scans by VUFKU), montage lists, filmographic descriptions, and stills from period publications. The screenplay is also documented in the Russian state archive (RGALI), whose archival record describes the script storage unit. These textual sources are everything any reconstruction has to work with — including this one.

In practice this means a hierarchy of trust: the editorial script and montage lists are primary (they describe the action scene by scene); the VUFKU filmographic card governs credits, cast, and dates; period press shows how contemporaries read the film. Each gallery frame below points to its storyboard scene number — the exact place in the text it was generated from. The direct link to the archive’s film page is in the source list at the end.

The reconstruction turns text into images in three steps. First, the literary script and the editorial script are broken into a storyboard: 40 scenes across the film’s six parts, each scene with its own page on the project site. Second, the recurring figures — Vasya, Uncle Pylyp, the priest, the zavhos — are fixed as character plates so that faces stay consistent from scene to scene. Third, scene stills are generated frame by frame, following the action described in the script for that scene number.

No AI tool recovers lost film. Every image on the project site is an AI interpretation of textual descriptions — what the camera might have seen if the description is accurate — and it inherits the biases of both the surviving text and the model. The images are explicitly labelled as interpretations; the project site repeats the disclaimer on every page, and so does this gallery.

What the reconstruction can honestly do: give students and researchers a visual scaffold for a film that exists only in documents; make the screenplay’s structure visible (six parts, forty scenes); and show where the text is precise and where it leaves room to imagine. What it cannot do: recover Dovzhenko’s actual framing, acting, or editing. The distinction matters, and keeping it visible is part of the method.

The reconstructed frames are static 4:3 images close to a silent-era frame; they do not imitate camera movement, editing, or pacing, which we do not know. The soundtrack of the film — intertitles and a tapper’s accompaniment — is entirely outside the reconstruction: the textual sources describe the action, not the score.

Where the script is silent about composition, the images register that zone of uncertainty instead of filling it in. That is the project’s working ethic: show the boundaries of what is known, do not blur them.

The public research site vasya.poberailo.name hosts the storyboard, character plates, and the frame gallery; the working material is open in the GitHub repository (project_Vasya). The site is in open development — new scenes land as they are reconstructed. The method grew out of the AI methodologies of my teaching course “Interactive Cinema: Theory, Practice, and AI Methodologies”, where reconstruction-from-text is one of the practical exercises.