---
title: "Loading reveal — Meleon Skill"
description: "Make a 7 second vertical loading reveal from one photo. The first 2 seconds hold your \"before\" photo still with a loading bar that fills and a \"loading...\" label. A white flash lands on the beat, then the \"after\" plays in motion for 5 seconds: your pet in a costume, a room redone, a sketch made real or a product unboxed. Nano Banana Pro makes the after still from your photo, a cheap 480p draft tests the motion, Kling 3 pro films the silent final, and ffmpeg draws the loading bar, the freeze and the flash locally, so the label is spelled exactly and no video model draws any interface. The file is silent by design and carries a printed beat time and a Beat chapter, so you add a trending sound in the app and line its drop up with the flash. For creators, pet owners, brands and social teams. People only with your own photo or written consent: no celebrities, no minors."
url: https://meleon.ai/skills/loading-reveal-transition/
author: "Meleon AI"
---

# Loading reveal

> Turn one photo into a 7-second loading reveal: your before frame frozen with a loading bar, a white flash on the beat, then the after in motion.

- Type: Video Skill
- Category: Video & animation
- Version: 1.0.0 (updated 2026-11-13)
- Made by: Meleon AI (official)
- Price: Free to download. Outside costs: About $1.07 per reveal on fal (after still $0.15 on Nano Banana Pro edit, 480p draft $0.36 on Seedance 2.0 mini, 5 s silent final $0.56 on Kling 3 pro), up to about $1.78 with one still and one final re-roll. The loading bar, freeze and flash cost nothing. Billed to your own fal account; quote the total and get a yes before the first paid call.
- Runs in: Claude Code, Codex and Cursor
- Page: https://meleon.ai/skills/loading-reveal-transition/
- Download: https://meleon.ai/downloads/loading-reveal-transition.zip
- Manifest: https://meleon.ai/skills/loading-reveal-transition.json

## What it does

Make a 7 second vertical loading reveal from one photo. The first 2 seconds hold your "before" photo still with a loading bar that fills and a "loading..." label. A white flash lands on the beat, then the "after" plays in motion for 5 seconds: your pet in a costume, a room redone, a sketch made real or a product unboxed. Nano Banana Pro makes the after still from your photo, a cheap 480p draft tests the motion, Kling 3 pro films the silent final, and ffmpeg draws the loading bar, the freeze and the flash locally, so the label is spelled exactly and no video model draws any interface. The file is silent by design and carries a printed beat time and a Beat chapter, so you add a trending sound in the app and line its drop up with the flash. For creators, pet owners, brands and social teams. People only with your own photo or written consent: no celebrities, no minors.

## Made for

- Creators
- Pet owners
- Brands
- Social media teams
- Interior designers

## Before you start

- You give it: Before photo + what the reveal is
- You need: Node.js 18 or newer and ffmpeg with ffprobe for the scripts; your own FAL_KEY only if you run scripts/fal-run.mjs.
- Cost per run: About $1.07 per reveal on fal (after still $0.15 on Nano Banana Pro edit, 480p draft $0.36 on Seedance 2.0 mini, 5 s silent final $0.56 on Kling 3 pro), up to about $1.78 with one still and one final re-roll. The loading bar, freeze and flash cost nothing. Billed to your own fal account; quote the total and get a yes before the first paid call.
- Makes: 1 after still + 1 silent Kling clip, joined locally with a drawn loading bar and flash into one MP4, 9:16, 720p, 7 s

## Install

Run the command for your agent in a terminal. It downloads https://meleon.ai/downloads/loading-reveal-transition.zip and unzips it into the agent's personal Skills folder.

### Claude Code

```sh
mkdir -p ~/.claude/skills && curl -fsSL https://meleon.ai/downloads/loading-reveal-transition.zip -o /tmp/loading-reveal-transition.zip && unzip -o /tmp/loading-reveal-transition.zip -d ~/.claude/skills
```

Then ask Claude Code to use it: `/loading-reveal-transition`.

### Codex

```sh
mkdir -p ~/.agents/skills && curl -fsSL https://meleon.ai/downloads/loading-reveal-transition.zip -o /tmp/loading-reveal-transition.zip && unzip -o /tmp/loading-reveal-transition.zip -d ~/.agents/skills
```

Then ask Codex to use it: `$loading-reveal-transition`.

### Cursor

```sh
mkdir -p ~/.cursor/skills && curl -fsSL https://meleon.ai/downloads/loading-reveal-transition.zip -o /tmp/loading-reveal-transition.zip && unzip -o /tmp/loading-reveal-transition.zip -d ~/.cursor/skills
```

Then ask Cursor to use it: `/loading-reveal-transition`.

## Example request

```text
/loading-reveal-transition my terrier, now in a red hero cape and goggles
```

In Codex, start with `$loading-reveal-transition` instead of `/loading-reveal-transition`.

## Files

Every file in the package, readable before you install:

- [SKILL.md](https://meleon.ai/skills/loading-reveal-transition/files/SKILL.md) (13.1 KB)
- [references/output-spec.md](https://meleon.ai/skills/loading-reveal-transition/files/references/output-spec.md) (4.1 KB)
- [references/prompt-patterns.md](https://meleon.ai/skills/loading-reveal-transition/files/references/prompt-patterns.md) (6.6 KB)
- [references/reveal-grammar.md](https://meleon.ai/skills/loading-reveal-transition/files/references/reveal-grammar.md) (4.9 KB)
- [references/safety-and-likeness.md](https://meleon.ai/skills/loading-reveal-transition/files/references/safety-and-likeness.md) (3.3 KB)
- [assets/example-input.jpg](https://meleon.ai/skills/loading-reveal-transition/files/assets/example-input.jpg/) (245.1 KB)
- [assets/generation-input.json](https://meleon.ai/skills/loading-reveal-transition/files/assets/generation-input.json) (4.5 KB)
- [assets/poster.webp](https://meleon.ai/skills/loading-reveal-transition/files/assets/poster.webp/) (108.2 KB)
- [assets/preview.mp4](https://meleon.ai/skills/loading-reveal-transition/files/assets/preview.mp4) (1.8 MB)
- [assets/prompt-template.md](https://meleon.ai/skills/loading-reveal-transition/files/assets/prompt-template.md) (2.7 KB)
- [assets/reveals.json](https://meleon.ai/skills/loading-reveal-transition/files/assets/reveals.json) (4.2 KB)
- [scripts/assemble-reveal.mjs](https://meleon.ai/skills/loading-reveal-transition/files/scripts/assemble-reveal.mjs) (12.1 KB)
- [scripts/build-plan.mjs](https://meleon.ai/skills/loading-reveal-transition/files/scripts/build-plan.mjs) (9.2 KB)
- [scripts/build-requests.mjs](https://meleon.ai/skills/loading-reveal-transition/files/scripts/build-requests.mjs) (12.0 KB)
- [scripts/check-reveal.mjs](https://meleon.ai/skills/loading-reveal-transition/files/scripts/check-reveal.mjs) (14.6 KB)
- [scripts/fal-run.mjs](https://meleon.ai/skills/loading-reveal-transition/files/scripts/fal-run.mjs) (8.3 KB)
- [scripts/loader-ui.mjs](https://meleon.ai/skills/loading-reveal-transition/files/scripts/loader-ui.mjs) (9.0 KB)
- [scripts/prepare-before.mjs](https://meleon.ai/skills/loading-reveal-transition/files/scripts/prepare-before.mjs) (4.0 KB)

## Versions

- 1.0.0 (2026-11-13)

## Instructions (SKILL.md)

Use this skill when someone wants the loading-screen reveal that works on short video: a photo held still with a loading bar creeping across it, a white flash on the beat, and then the same place in motion with one thing changed. The pet is suddenly in a costume, the room is done up, the sketch is real, the box is open. Only the "after" is made by a model. The freeze, the loading bar, its label and the flash are drawn locally with ffmpeg, so the words are spelled exactly and the cut is always on time.

The file is silent on purpose. The format lives on the sound people pick in the app, and Meleon neither makes nor copies that sound. The reveal comes with a printed beat time, a ready cut point (the first fully white frame, also set as an MP4 chapter) and a short recipe for lining a sound up with it.

### What you need

- **One "before" photo** that the user owns or may use: a pet, a room, a product, or a sketch or drawing. Sharp, the subject whole and easy to see, with room in the middle of the frame for the loading bar. A wide photo is cropped to 9:16 first, so check the crop.
- **What the reveal is**, in the user's own words: "a small red hero cape and round goggles", "the same living room in a warm Scandinavian style", "the real motorbike from this sketch", "the open box with the candle inside". The menu is in `references/reveal-grammar.md`: `costume` (pet, product or person), `room`, `sketch`, `unbox` (product) or `custom`.
- **For an unboxing:** a photo of the product itself, so its shape and colours are kept.
- **Optional:** the label (default "loading", up to 16 characters), the hold time (2 s by default, 1.5 to 3) and the bar colour.

**Limits.** The default subjects are pets, rooms, products and sketches. A people version is possible only with the user's own photo, or an adult's photo with written consent, and never a celebrity, a public figure or anyone under 18 (`references/safety-and-likeness.md`). The people version skips the Seedance draft, because Seedance on fal refuses a clearly visible realistic face, and uses Kling. A pet must look calm and comfortable in its costume, with nothing over its eyes, nose or mouth. Nothing in the reveal may claim to be real footage: post it with the platform's AI label. The after is a new image, so check that the subject still matches the photo: markings, colours, shape and the room's layout. In the sample run the dog's patch, collar and goggles held through all five seconds of Kling; the slow push-in made the dog fill more of the frame by the end, and it blinked once. Do not use a trending sound by name or ask for one to be copied: the user adds their own from the app's licensed library.

### What you make

1. `plan.json`: the lock (what must not change), the reveal and its motion, and the file paths.
2. `before-9x16.jpg`: the user's photo cropped to 9:16, with its metadata removed.
3. The approved after still (JPEG, 9:16, 2K).
4. A silent 480p draft (except for people) and a silent 5 s Kling final.
5. The reveal: H.264 MP4, 720 × 1280, 24 fps, about 7 s (2 s before, 5 s after), no audio track, under 8 MB, with the beat time in its edit list and a "Beat" chapter.
6. A contact sheet and a beat strip (eight frames around the flash) that you checked it with.

### How to run it

**Budget: about $1.07 per reveal on fal: the after still $0.15, the 5 s 480p draft $0.36 and the 5 s silent Kling 3 pro final $0.56. Up to about $1.78 with one still re-roll and one final re-roll. The people version has no draft: about $0.71.**

Tell the person the model and the cost before every paid step, and wait for a yes. Every request goes to fal with the user's own FAL_KEY: `scripts/fal-run.mjs --request <file> --out <dir> --approved` POSTs the printed `input` to `https://queue.fal.run/<model id>`, waits and downloads, and embeds local photos as data URIs so nothing is put on a public URL. Or paste the prompts and images into any Nano-Banana-class image editor and Kling-class image-to-video tool. The Skill never stores keys.

1. **Intake (free).** Look at the photo with your image tool. Decide the kind (`pet`, `room`, `product`, `sketch` or `person`) and the reveal, and write the lock in one line (`references/prompt-patterns.md`). Refuse what `references/safety-and-likeness.md` rules out, and say why. For a person, ask for the consent first.
2. **Prepare the before photo (free).** `node scripts/prepare-before.mjs --input photo.jpg --output before-9x16.jpg [--focus 0.5]` crops it to 9:16 and strips its metadata. Open it: the subject is whole and sits in the middle, clear of the top 12% and the bottom 20%. Move `--focus` for a wide photo. The after still is made from this crop, so the two frames share one framing.
3. **Plan (free).** `node scripts/build-plan.mjs --out plan.json --kind pet --before before-9x16.jpg --reveal costume --change "a small red hero cape and round goggles" --lock "a small scruffy terrier mix, ..."` (add `--refs product.jpg`, `--hold 2.5`, `--label "loading"`, or `--consent yes` for a person). It copies the reveal's after text and motion from `assets/reveals.json` into plan.json. Fill any `FILL:` field, then describe the reveal to the user in plain words: what the before shows, what changes, how the after moves.
4. **Spend gate (free).** `node scripts/build-requests.mjs --plan plan.json --stage estimate`. Quote the printed total and wait for one explicit yes. Ask again if the spend will pass it.
5. **After still (about $0.15).** `--stage keyframe --out requests --inline`, then send the request (`fal-ai/nano-banana-pro/edit`, `aspect_ratio` "9:16", `resolution` "2K"). Put it beside the before photo: the same place, angle, size and light, the subject unchanged except for the one planned change, no text or logos, no extra people or animals. Re-roll with one change (the lock, the wording of the change, or a reference), at most 3 attempts. Put the approved path in `files.keyframe` and show the user both frames before any video.
6. **Draft (optional, about $0.36).** `--stage draft` (Seedance 2.0 mini, 480p, 5 s, silent). Make a contact sheet (`ffmpeg -i draft.mp4 -vf "fps=2,scale=180:-2,tile=8x2" -frames:v 1 sheet.jpg`) and look: the motion reads, the subject holds its look, nothing spawns. It is a different model from the final, so judge the idea, not the physics. Skip the draft for a person. Seedance can also refuse a harmless start frame: on the sample dog in a cape and goggles it returned a "likenesses of real people" refusal, and the refused call still showed on the lab's ledger. If that happens, do not retry the same image: plan with `--no-draft` and let the Kling final be the test.
7. **Final (about $0.56).** `--stage final` prints the Kling 3 pro request: `fal-ai/kling-video/v3/pro/image-to-video`, `start_image_url`, `duration` "5", `generate_audio` false and a negative prompt. `--model veo` (6 s, about $0.60 silent) and `--model seedance` (about $2.36, no faces) are the alternatives. Contact-sheet the clip at 2 fps and note a clean start and end: skip the first 0.1 s, stop before anything odd. Write them as `files.final`, `files.in` and `files.out`.
8. **Assemble (free).** `node scripts/assemble-reveal.mjs --plan plan.json --out reveal.mp4`. It holds the before photo still for `--hold` seconds with the loading panel, the label and a bar that crawls to 93% and jumps to full, ramps the picture to white over the last two frames, shows a fully white frame on the beat, and clears the white over `--flash` seconds into the after clip. It prints the beat (for example "2.000 s, frame 48"), writes it to `reveal.mp4.edl.json` and sets a "Beat" chapter. `--label`, `--accent`, `--ui-y`, `--dim` and `--max-mb 7.5` for a web preview are in the script's header.
9. **Check (free).** `node scripts/check-reveal.mjs --file reveal.mp4 --edl reveal.mp4.edl.json`, then open the contact sheet and the beat strip and go through every check in `references/output-spec.md`. A light coming on in the after can trip the brightness check: look at the moment, then re-run with `--ignore <seconds>`.
10. **Fail path: at most 3 attempts at each paid step.** Change one thing per attempt and say which: the wording of the change, then the lock, then the model for the final. If the model refuses the still, look for a face, a screen or a hurt-looking pet, remove it and regenerate. After three, deliver the best attempt, trimmed to its clean part, and say exactly what differs. Never deliver a reveal that fails the subject, people, welfare or labelling checks.
11. **Deliver** the reveal, the after still, the before crop and plan.json. Say the beat time, and say how to add the sound: in the app's edit screen, move the chosen sound so its drop (or first loud beat) lands on the first white frame at the printed time. Remind the user to pick the sound from the app's own library, to switch on the platform's AI label, and not to caption it as real. Say what was re-rolled.
12. **Clean up.** Delete the user's photo and its crop from the working folders unless they want them kept. fal keeps request inputs (including inline photos) and outputs for a period: tell the user they can delete them from their fal dashboard, and to check fal's current retention policy.

### Files

- `references/reveal-grammar.md`: why the format works, the 7-second timeline, the loading screen, the flash and the beat, safe zones, adding a sound, and the reveal menu.
- `references/prompt-patterns.md`: how to write the lock, the after-still and motion prompts, failure modes and fixes, drafts versus finals, the fail path.
- `references/safety-and-likeness.md`: what may be shown, the people version, pets, labelling and sound.
- `references/output-spec.md`: sizes, costs, the pass/fail checklist and the posting note.
- `assets/reveals.json`: the five reveals: after text, camera, motion and hazards.
- `assets/preview.mp4`, `assets/poster.webp`: the sample reveal (a synthetic terrier in a red hero cape and goggles), 720 × 1280, 6.9 s, silent. `assets/example-input.jpg` is its before photo (an invented dog, no real pet).
- `assets/prompt-template.md`: the after-still and motion wrappers the script fills.
- `assets/generation-input.json`: model ids, fields, default settings, alternatives and prices.
- `scripts/prepare-before.mjs`: crops the photo to 9:16 and removes its metadata (ffmpeg).
- `scripts/build-plan.mjs`: writes plan.json from the photo, the kind and the reveal. Calls no API.
- `scripts/build-requests.mjs`: the cost estimate, and the after-still, draft and final requests. Calls no API.
- `scripts/fal-run.mjs`: sends one request with the user's FAL_KEY after `--approved`, waits, downloads (never overwrites without `--force`).
- `scripts/loader-ui.mjs`: draws the loading panel, label and bar fill as PNGs with a built-in pixel font (no fonts, no npm packages).
- `scripts/assemble-reveal.mjs`: freeze, loading screen, flash, after clip, beat marker, H.264 `+faststart` (ffmpeg).
- `scripts/check-reveal.mjs`: format, silence, frozen start, filling bar, flash on the beat, moving after, black and flash checks, contact sheet and beat strip (ffmpeg).

## Related

- [Cinematic brand film](https://meleon.ai/skills/cinematic-brand-film.md): Turn a one-line brand story into a graded 30 to 45 second film with a title card, the same cast in every shot.
- [Destination teaser](https://meleon.ai/skills/destination-teaser.md): Turn a place, a season and a mood into a graded 15 to 20 second vertical teaser, text-free and ready for captions.
- [ASMR cut reel](https://meleon.ai/skills/asmr-cut-reel.md): Remake an object or your product in glass, jelly, soap or chocolate and slice it in a looping ASMR reel of about 15 seconds with crisp cutting sound.
- [All Meleon Skills](https://meleon.ai/skills/) (index as JSON: https://meleon.ai/skills.json; site guide: https://meleon.ai/llms.txt)
