Mascot product ad

Make a ~30 s product ad that stars the brand's own mascot: a short cartoon story built on the product's name or promise, narrated, with the product's real substance (its UI, code, integrations, dashboard) composited on top as motion graphics and key-phrase captions. Works from the product's website and mascot art alone; an existing launch video, if there is one, supplies the beats, the real screens and the logos, and is the thing a side-by-side is measured against.

Use it for a SaaS, dev tool or app with a mascot or a drawn logo character, a gift video for a founder, or a new take on a brand's existing launch video. For a presenter talking over rebuilt UI use recipe-product-ui-video; for a faceless type-and-screenshot explainer use recipe-product-explainer-video.

Typical cost: about $10 at 1080p (7 shots: stills about $3, video about $6, voice and music under $0.50), three review rounds.

Inputs

Use the brand's own materials, made for or shared by the brand owner, and keep the result private to them until they agree to it being shown. Logos of partner products appear only the way the brand already shows them (its integrations list, its own launch video).

The product site: headline, subhead, features, integrations, supported languages or platforms.
The mascot: the brand's own art (logo lockup, hero illustration). This is the character reference for every still.
Optional: the brand's existing launch video, as a file the person has. Cut it into beats (ffmpeg-scene-cuts or a 1 fps contact sheet) and keep full-resolution frames of its product screens.

Steps

Break down the source. List the substantive points the product makes (what it does, setup time, languages, integrations, monitoring) and, from a launch video, its beats and lengths. Every one of these points must survive into the ad, either in picture, narration or motion graphics. Dropping them was the first round's main note.
Concept and shot list (gate). One story world that puns on the product name or promise and gives the mascot a job (an usher for a product named after seats, a doorman, a librarian). One shot per point, 3 to 5 s each, about 7 shots for 28 s. For each shot: the visual, the narration line, and which real screen or logo goes over it. The narration names the specifics (languages, stores, integrations), not "whatever you use". Show it with a price before anything is spent.
Voice (gate). guide-cast-a-voice: three auditions reading the whole script, so the person hears the pacing too. Tags in the script, eleven_v4, a fixed seed.
Look (gate). One style frame of a key shot with nano banana Pro edit, the mascot art as reference, 2 takes, 2K, 16:9. Match the mascot's own drawing style and palette exactly; describe the new costume pieces (cap, bow tie) and keep everything else. Expect the model to bake labels into props ("Plugin" on a guitar pedal): check every still and remove them with a one-line edit before it moves.
Stills (gate). Every other shot from the picked style frame plus the mascot art, 2 takes each. Describe the new location strongly: with a busy style frame as reference, about half the takes just copy its set. Drop those before the review. The integrations the brand lists can go in the picture itself where the story allows it: give their logos, as the brand shows them, as references ("a brass sign above each booth shows the logo in the second image") so they are drawn into the world, then also as motion graphics. Remove anything characters could walk through (ropes, rails) unless the action needs it; a video model will walk them straight through.
Motion test (gate). Animate one shot first (h3-max-image-to-video, 1080P, prompt_expansion_mode: disabled, "Static shot, locked on the opening framing", mouths closed). Approve the motion before the batch.
Animate the rest. All at once, wait: false. Direct exits explicitly: "turns and strolls to the right toward the open doors". Check each on a frame sheet: raw first-frame SSIM on grainy cartoon renders reads 0.70 to 0.85 even on a correct opening, so judge by eye.
Narration track. Voice each line as its own file (previous_text for continuity), never one take cut into pieces: a cut inside a take left an audible artifact the person heard on the first delivery. Transcribe each line (whisper-cli -ml 1 -sow) for word times.
Music. One bed from a composition plan whose sections land on the cut: a sparse intro for shot 1, a groove under the middle, a final hit as the end card lands. No vocals in the negative styles; check the take with a transcript.
Base cut. Trim each clip to its slot and hold the last frame (tpad) when the slot is longer, place each narration file at its cue, clip sound effects at about 0.35, the bed ducked under the voice (sidechaincompress). Measure the real cut frames afterwards (select='gt(scene,0.2)',showinfo): a 24 fps concat drifts shot starts by a few frames from the planned times.
Motion graphics (gate: live preview). hn4a-compositor-motion-graphics over the base cut:
- Key-phrase captions only (2 to 4 words per shot), in a heavy display serif in the brand's colours, popped on the spoken word.
- The product's real screens (an activation dialog, a code snippet, a dashboard) as cards, taken from the brand's own launch video frames or site screenshots, popped on the line that mentions them.
- Small logos (languages, integrations) popping in one by one on each spoken name.
- A card that replaces something in the picture (a dashboard on a theater screen) cuts on and off on that shot's measured cut frames with "in": {"fx": "cut"} and "out": {"fx": "cut"}. A pop or a fade lets the original show for a few frames, and the person will see it.
- The URL as the last caption. Show the spec live on a review page with stills at each cue; render only after a yes.
Deliver. Render, check the frame count, transcribe the cut against the script, measure loudness (about -17 LUFS, peaks under -1.5 dBTP; trim the gain rather than re-normalizing), and put the film and a side-by-side with the original (original left, ours right, our audio) on an approve page. Fix notes by re-rendering only the changed span (render --frames A:B --into) on a copy of the last render.

Choices this recipe makes

The story world comes from the product's name or promise, and the mascot gets a job in it. That is what makes it more than a re-skin of the original.
Substance rides on top as motion graphics with the product's real screens, never as generated UI: generated screens get the text wrong.
One narration file per line.
Every look is gated: voice, style frame, stills, one motion test, the live layout.

Dead ends

Keeping a "no on-screen text" rule for a product ad: the first cut was charming but lost the product's substance. Captions and the real screens fixed it.
"Whatever language you ship" in the narration: name them.
A dashboard card that popped in 0.4 s into its shot: the picture underneath flashed.

Sharing it

When the ad is made as a gift for the brand's founder, send it to them privately with the side-by-side, say what it cost and how long it took, and post it only once they say yes.