---
name: wording-verdict
description: Animate two candidate phrases, then reveal a deliberate editorial choice.
version: 1.0.3
---

# Wording Verdict

Use this parameterized renderer for a short, legible comparison of two written instructions or promises. Supply the decision yourself: this tool does not evaluate truth or choose wording.

## Author the brief

Write JSON with `phrases` (exactly two strings, each at most 160 characters), `chosen` (0 or 1), `relation` (up to 12 characters), `conclusion` (up to 70 characters), and `palette` (`estuary`, `mulberry`, `graphite`, or `clay`). Keep each sentence concise enough to read aloud. The concluding line should explain the choice without introducing unsupported claims.

## Generate and render

Use Node.js 22 or newer. Install the pinned renderer with `npm install --ignore-scripts`. Chromium and FFmpeg must be available to HyperFrames; its first render may need network access to acquire Chromium.

Run `node generate.mjs /absolute/brief.json /absolute/project`. Then run `npx hyperframes render /absolute/project --fps 30 --quality high --strict --workers 1 --output /absolute/project/video.mp4`.

The project contains editable HTML, GSAP animation, local fonts, the resolved brief, and timing data. Its 1920×1080 canvas reveals the first phrase, pauses on a relationship marker, introduces the second phrase, and only then changes emphasis. The reading duration grows with token count. Chinese uses grapheme steps; Latin content uses word steps. A single local Noto Sans SC font covers both.

## Inspect before delivery

Read every line at output size. Confirm the selected index, settled phrase bounds, and conclusion. Test nonmonotonic seeks around each text reveal and emphasis transition. Decode the complete encoded film and play it at normal speed. Reject overflow, lost glyphs, or stale selection states. Record dropped and total frames, playback rate, resolution, buffering state, and browser conditions. Assess visible disruption and investigate likely causes before deciding whether playback needs correction. A nonzero drop count alone does not establish a content defect or require rejection; retain every observed result and state any uncertainty. Shorten excessive content rather than shrinking it into an unreadable caption.

## Scope

This is a fixed comparison grammar with data-driven wording, timing, and color choices. It is not arbitrary story generation. Output is silent: no music, narration, or sound editor is included. Dependencies retain their notices in assets. No credentials or paid services are needed. Very long unbroken strings may require editing even within the character limit; inspect actual output.