Từ dẫn phóng to rồi ráp câu

Mọi thứ trên trang này đều có bản máy đọc. Đưa agent link JSON, hoặc thêm MCP server gimgs rồi bảo nó tìm motion template.

1. Tải JSON — trong đó có công thức cảnh quay (timing, easing, bảng tham số, lỗi hay gặp) và toàn bộ source TSX.
2. Clone repo video-shotcraft; component import fixture chung từ demos/_fixtures và helper từ assets/lib.
3. Đăng ký component thành <Composition> trong project Remotion rồi render bằng `npx remotion render`, hoặc dựng lại công thức bằng engine của bạn.

curl -s https://gimgs.net/motion/lead-word-zoom-assemble.json

git clone https://github.com/Vincentwei1021/video-shotcraft
# component: demos/typography/lead-word-zoom-assemble/LeadWordZoomAssemble.tsx
npx remotion render <CompositionId> out.mp4

claude mcp add --transport http gimgs https://gimgs.net/mcp
# then: search_motion_templates / get_motion_template("lead-word-zoom-assemble")

Use the video-shotcraft shot recipe "Từ dẫn phóng to rồi ráp câu" (Remotion composition, Apache-2.0) published on gimgs.net.
Recipe JSON (prompt = timing/easing/parameter table/pitfalls, files = TSX source): https://gimgs.net/motion/lead-word-zoom-assemble.json
Raw TSX: https://gimgs.net/motion/lead-word-zoom-assemble/template.html
Repository: https://github.com/Vincentwei1021/video-shotcraft  (component: demos/typography/lead-word-zoom-assemble/LeadWordZoomAssemble.tsx; imports demos/_fixtures and assets/lib)
Spec: 1920x1080, 30 fps, 2.8s

1. Read the JSON and follow its "prompt" section (when to use, intention, parameter table, known pitfalls).
2. Adapt it for: <describe what you want>.
3. Either clone the repo, register the component as a <Composition> in a Remotion project and run npx remotion render <CompositionId> out.mp4 — or re-implement the recipe in your own engine (HTML/WAAPI works: keep the same timing and easing).

JSON · Markdown · llms.txt

// ===== demos/typography/lead-word-zoom-assemble/LeadWordZoomAssemble.tsx =====
// lead-word-zoom-assemble — 首词推近落位组句
// 首词以 2.3 倍字号占据画面中央，hold 期间继续向观众推近 6%，随后一条曲线同时
// 完成「缩回终字号」与「整行左滑到最终排布」，后续词从右侧各自槽位外 0.5em 推入；
// 整行上移的同一时窗里副行浮出，停一拍后整幕 crash-zoom 推近失焦交棒下一镜。
//
// 手感命门是支点：transform-origin 横向钉在**首词中心**、纵向钉在**基线**——
// 这两点是全程唯一不该移动的锚，挂载时实测一次（基线用零尺寸 inline-block 的
// offsetTop 读出），否则缩放过程中字会逐帧抖基线。
// 参数以 1920×1080 标定。
import React, { Fragment, useEffect, useRef, useState } from 'react';
import {
  AbsoluteFill,
  continueRender,
  delayRender,
  Easing,
  getRemotionEnvironment,
  interpolate,
  useCurrentFrame,
  useVideoConfig,
} from 'remotion';

export const LEAD_WORD_ZOOM_ASSEMBLE_DURATION = 84; // 2.8s @30fps

// ---- 编舞常量 ----
const TEXT = 'Introducing Lumen Deck';
const HIGHLIGHT_WORD = 'Lumen'; // 精确匹配的词换强调色：品牌词落位那一下自带高亮
const FONT_SIZE = 96;
const INITIAL_SCALE = 2.3; // 首词起手放大倍数
const INTRO_DURATION = 6; // f：首词淡入
const HOLD_DURATION = 12; // f：首词在峰值停留（推近发生在这一段）
const PUSH_SCALE = 1.06; // hold 期间继续推近的倍数
const RECEDE_DURATION = 12; // f：缩回终字号
const ASSEMBLE_DURATION = 24; // f：整行左滑到位
const WORD_DELAY = 6; // f：缩回起点后第二个词开始推入
const WORD_STAGGER = 4; // f：后续词之间的错峰
const WORD_DURATION = 12; // f：单词推入行程
const WORD_PUSH = 0.5; // em：后续词起始位置在自己槽位右侧多远
const WORD_FADE = 2; // f：后续词淡入（刻意极短——是被推进来的，不是淡进来的）
const LETTER_SPACING = '-0.03em';

// ---- 场景层 ----
const LIFT: [number, number] = [34, 50]; // f：整行上移 + 副行同帧出现
const LIFT_DISTANCE = -56; // px
const SUBLINE = 'One shot card, one motion recipe — copy, paste, render.';
const CRASH_FRAMES = 12; // f：段尾 crash-zoom 占用的收尾帧数
const CRASH_SCALE = 0.2;
const CRASH_BLUR = 9;

const INK = '#1d1d1f';
const INK_DIM = '#7a7a7a';
const ACCENT = '#7A5AF8';
const SANS = '-apple-system, "PingFang SC", BlinkMacSystemFont, sans-serif';
const MESH_BG =
  'radial-gradient(52% 44% at 18% 22%, rgba(122,90,248,0.20) 0%, rgba(122,90,248,0) 70%),' +
  'radial-gradient(46% 42% at 84% 18%, rgba(255,138,178,0.20) 0%, rgba(255,138,178,0) 70%),' +
  'radial-gradient(58% 50% at 78% 84%, rgba(96,190,255,0.20) 0%, rgba(96,190,255,0) 70%),' +
  'radial-gradient(50% 46% at 24% 88%, rgba(255,196,112,0.20) 0%, rgba(255,196,112,0) 70%),' +
  'linear-gradient(180deg, #f7f6f9 0%, #f2f1f5 100%)';

/** 首词向观众漂移：先快后停，峰值处挂住 */
const PUSH_EASE = Easing.bezier(0.25, 1, 0.5, 1);
/** 缩回与左滑共用同一条曲线——两个动作必须读作一次运动，不能各走各的 */
const ZOOM_EASE = Easing.bezier(0.5, 0, 0.05, 1);
/** 后续词推入槽位：起步快，落地长而软 */
const WORD_EASE = Easing.bezier(0.22, 0.8, 0.36, 1);

const TextReveal: React.FC = () => {
  const frame = useCurrentFrame();
  const { width } = useVideoConfig();

  // 整行宽度、首词中心占比、基线位置：只在挂载时量一次，量到之前挂起渲染
  const lineRef = useRef<HTMLSpanElement>(null);
  const leadRef = useRef<HTMLSpanElement>(null);
  const baselineRef = useRef<HTMLSpanElement>(null);
  const [handle] = useState(() => delayRender('lead-word-zoom-assemble: measure line'));
  const [metrics, setMetrics] = useState<{
    lineWidth: number;
    leadRatio: number;
    baseline: number;
  } | null>(null);

  useEffect(() => {
    const line = lineRef.current;
    const lead = leadRef.current;
    if (!line || !lead) {
      continueRender(handle);
      return;
    }
    setMetrics({
      lineWidth: line.offsetWidth,
      leadRatio: (lead.offsetLeft + lead.offsetWidth / 2) / line.offsetWidth,
      baseline: baselineRef.current?.offsetTop ?? line.offsetHeight * 0.8,
    });
  }, [handle]);

  useEffect(() => {
    if (metrics) continueRender(handle);
  }, [metrics, handle]);

  const words = TEXT.split(' ').filter(Boolean);
  const ready = metrics !== null;
  const leadRatio = metrics?.leadRatio ?? 0.14;
  const lineWidth = metrics?.lineWidth ?? width * 0.5;
  const baseline = metrics?.baseline ?? FONT_SIZE * 0.88;

  const zoomStart = HOLD_DURATION;
  // 首词从"画面正中"走到"行内自己的位置"，位移就是这段偏心距
  const slideDistance = lineWidth * (0.5 - leadRatio);

  return (
    <AbsoluteFill style={{ alignItems: 'center', justifyContent: 'center' }}>
      <span
        ref={lineRef}
        style={{
          position: 'relative',
          display: 'inline-block',
          fontSize: FONT_SIZE,
          fontWeight: 600,
          color: INK,
          letterSpacing: LETTER_SPACING,
          lineHeight: 1.1,
          whiteSpace: 'nowrap',
          fontFamily: SANS,
          transformOrigin: `${leadRatio * 100}% ${baseline}px`,
          scale:
            interpolate(frame, [0, HOLD_DURATION], [INITIAL_SCALE, INITIAL_SCALE * PUSH_SCALE], {
              extrapolateLeft: 'clamp',
              extrapolateRight: 'clamp',
              easing: PUSH_EASE,
            }) +
            interpolate(
              frame,
              [zoomStart, zoomStart + RECEDE_DURATION],
              [0, 1 - INITIAL_SCALE * PUSH_SCALE],
              { extrapolateLeft: 'clamp', extrapolateRight: 'clamp', easing: ZOOM_EASE },
            ),
          translate: `${interpolate(
            frame,
            [zoomStart, zoomStart + ASSEMBLE_DURATION],
            [slideDistance, 0],
            { extrapolateLeft: 'clamp', extrapolateRight: 'clamp', easing: ZOOM_EASE },
          )}px`,
          opacity: ready ? 1 : 0,
          textRendering: 'geometricPrecision',
          ...(getRemotionEnvironment().isRendering ? null : { willChange: 'transform' as const }),
        }}
      >
        {words.map((word, i) => {
          const isLead = i === 0;
          const pushStart = zoomStart + WORD_DELAY + (i - 1) * WORD_STAGGER;
          const opacity = isLead
            ? interpolate(frame, [0, INTRO_DURATION], [0, 1], {
                extrapolateLeft: 'clamp',
                extrapolateRight: 'clamp',
              })
            : interpolate(frame, [pushStart, pushStart + WORD_FADE], [0, 1], {
                extrapolateLeft: 'clamp',
                extrapolateRight: 'clamp',
              });
          return (
            <Fragment key={i}>
              <span
                ref={isLead ? leadRef : undefined}
                style={{
                  display: 'inline-block',
                  opacity,
                  color: word === HIGHLIGHT_WORD ? ACCENT : undefined,
                  translate: isLead
                    ? undefined
                    : `${interpolate(
                        frame,
                        [pushStart, pushStart + WORD_DURATION],
                        [WORD_PUSH * FONT_SIZE, 0],
                        { extrapolateLeft: 'clamp', extrapolateRight: 'clamp', easing: WORD_EASE },
                      )}px`,
                }}
              >
                {word}
              </span>
              {/* 词间空格必须放在 inline-block 之外——跟在词里会被行盒裁掉，词会黏在一起 */}
              {i < words.length - 1 ? ' ' : null}
            </Fragment>
          );
        })}
        {/* 基线尺：零尺寸 inline-block，它的 offsetTop 就是基线 */}
        <span ref={baselineRef} style={{ display: 'inline-block', width: 0, height: 0 }} />
      </span>
    </AbsoluteFill>
  );
};

export const LeadWordZoomAssemble: React.FC = () => {
  const frame = useCurrentFrame();
  const { durationInFrames } = useVideoConfig();

  const lift = interpolate(frame, LIFT, [0, 1], {
    extrapolateLeft: 'clamp',
    extrapolateRight: 'clamp',
    easing: Easing.out(Easing.cubic),
  });
  const crash = interpolate(
    frame,
    [durationInFrames - CRASH_FRAMES, durationInFrames - 1],
    [0, 1],
    { extrapolateLeft: 'clamp', extrapolateRight: 'clamp', easing: Easing.in(Easing.quad) },
  );

  return (
    <AbsoluteFill style={{ background: MESH_BG, fontFamily: SANS }}>
      <AbsoluteFill
        style={{
          transform: `scale(${1 + crash * CRASH_SCALE})`,
          filter: crash > 0.01 ? `blur(${crash * CRASH_BLUR}px)` : undefined,
          opacity: 1 - crash * 0.55,
        }}
      >
        <AbsoluteFill style={{ transform: `translateY(${lift * LIFT_DISTANCE}px)` }}>
          <TextReveal />
        </AbsoluteFill>
        <div
          style={{
            position: 'absolute',
            left: 0,
            right: 0,
            top: '50%',
            marginTop: 62,
            textAlign: 'center',
            fontSize: 32,
            color: INK_DIM,
            opacity: lift,
            transform: `translateY(${(1 - lift) * 16}px)`,
          }}
        >
          {SUBLINE}
        </div>
      </AbsoluteFill>
    </AbsoluteFill>
  );
};

1. Read the recipe
   Prompt = when to use, intention, parameter table with typical values and how each one feels, known pitfalls. Treat values as calibrated starting points.
2. Get the code
   tool: git
   git clone https://github.com/Vincentwei1021/video-shotcraft && cd video-shotcraft && npm install
3. Register + render
   tool: remotion
   npx remotion render <CompositionId> out.mp4 # component: demos/typography/lead-word-zoom-assemble/LeadWordZoomAssemble.tsx
4. Or re-implement
   Port the timing/easing from the recipe to HTML + WAAPI (gimgs motion contract) or any engine; keep the same segment windows and easing.