वडापाव — मुंबईचा अस्सल बॉम्ब is an eleven-scene scroll film about a ten-rupee snack: a real-time 3D stage with generated footage behind it, a Marathi narration track, and a cut that was reordered twice because the obvious running order buried the people and made the gold stop meaning anything.
01 — The Premise
Vada pav gets written about in two registers: nostalgia, or nutrition. Neither is how anyone actually meets it — which is standing up, in a hurry, usually in the rain, from a stall that has been there longer than you have. The film sets out to give a street snack the grammar normally reserved for a luxury product, without the irony that usually comes attached when someone does that.
It is written in Marathi first — not translated into it. The prologue is a two-line exchange between strangers: someone shouts an order in from the street, the stall shouts back that it is coming. That exchange only works in the language it happens in, and it set the rule for everything after it.
02 — The Cut
Every scene owns a slice of the scroll measured in viewport-heights, and its typography beats are placed on that scene's own local progress — so re-timing a line never touches a line of 3D code. The running order lives in one array, which is what made re-cutting the film a one-line change rather than a six-hundred-line copy-paste.
Click any frame to open it full-size.
Graded cold on purpose, so the food that follows reads as heat.
One photograph, sliced per person and revealed left to right as you scroll.
The burst lands at a frame found by eye, not by scene detection.
A triptych, so all three chutneys are on screen together by the end.
Re-encoded all-keyframe so the crack can be scrubbed frame by frame.
The second cold scene, placed to break a long warm run in the middle.
03 — The Story Edit
Two moves away from the obvious sequence carry most of the film's improvement, and neither is a visual change. This is the part of the work that looks like nothing in a screenshot.
04 — Craft
The crowd scene is one photograph, sliced per figure and revealed left to right. Each person carries an hour and a reason — which is the entire argument of the film compressed into five lines.
वड्याचा आत्मा is the only scene with no footage at all. Cut-outs arrive one at a time against pure black so they read as a diagram rather than competing with a video — and the scroll stops on every reveal, not only on the typography beats.
The explosion scene is cut to three moments read off a contact sheet: the pav sitting whole in its own steam, the burst, and the aftermath with the pieces separated and hanging. The obvious automation does not find it — ffmpeg's scene detection peaks a full second late, on the frame where the most pixels are moving, because a slow bloom of an explosion never out-scores peak debris. And the beat before the bang is empty on purpose: a bang needs a silence in front of it, or it is just one loud moment among loud moments.
05 — Design System
Seven colours carry the whole film, and the discipline is in how little the accents are used. Gold is heat; it appears where there is heat. Chilli red is reserved for the sharp end — the blast, the hot chutney — and never used as a UI colour.
06 — Engineering
The division is deliberate and it holds everywhere: the food, the camera and every interaction are live 3D, and video is only ever scenery — a matte painting fogged in behind the geometry. Nothing the reader can touch depends on a video's playhead.
3,380vh page. Scrubbed scenes map their slice onto a clip's currentTime; the rest play on their own.07 — Outcomes
Almost everything that made this piece better was a reordering, a removal, or a colour taken away — not something added. The strongest version of the film has fewer warm scenes, fewer layer labels, and one fewer explosion than the first cut did.
08 — Key Learnings