anbeime/skill / skills/anything2explainer/anything2explainer
anything2explainer
Unverified给一个主题,产出一条黑底 MG 风格(幕底可选星点或点阵波)、有配音字幕章节进度条的科普讲解视频(中文或英文;Remotion 代码动画;时长由用户定,常用 3–5 分钟)。内含可编译模板、图元库、配音/分镜/渲染工具、风格与动效规范、多 agent 分工协议与 QC 判据,以及一条完整样片(《RAG 与知识库》)作为质量标尺。Turn any topic into a narrated motion-graphics explainer video in Chinese or English, on a black canvas with a star-field or dot-field backdrop, TTS voiceover, subtitles and a chapter progress bar, every frame drawn in code with Remotion. Use when the user asks for an explainer, educational or science-communication video about a topic, or wants an article or document turned into a video.
examples/rag/shots_src/G2/SC09.tsx
View on GitHubimport React from 'react';
import {useCurrentFrame} from 'remotion';
import {stepKf, clamp01, emphasisPulse, FONT_ORB, FONT_SERIF} from '../../common';
import {CText, Box, Pill, TechText, WHITE, GREY, GREY_LIGHT, GREY_MID, PURPLE, PURPLE_LIGHT, TEXT_GLOW, BLOOM_SOFT, slideUp, fadeIn, SoftIn} from '../../ui';
import {mixHex} from './parts';
/**
* SC09 2020 论文 · N 1130–1328(S09 1138–1326:1138「2020年」/ 1152「Facebook AI 的研究者在论文里」/ 1211「正式提出了这个名字」/ 1238「Retrieval-Augmented Generation」)
* 事实依据(research/RAG调研.md §1):Lewis et al.,《Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks》, Facebook AI Research (Meta), NeurIPS 2020。
* 版式:左 论文卡 (140,140,380,460):三行标题(Times 22/700)+ 作者行「Lewis et al. · NeurIPS 2020」灰 + 摘要灰线两栏 + 右下小柱状图;
* 右 「2020」Orbitron 96 (860,300) → 胶囊「Facebook AI Research」(690,376,340,50) → 大号 Exo 2 两行 Retrieval-Augmented / Generation (860,495/545)。
* 节拍:1138 年份 SoftIn + 2016→2020 计数(每 3 帧 +1,14 帧);1146 论文卡 slideUp(Δ90/22 + 8 帧淡入;QC v1:原 Δ300 在 1151–1158 穿过字幕带);1152 胶囊 SoftIn(本镜头无 glitch,协议 §8);
* 1211 标题下紫色下划线 draw-on 14 帧 + 标题块 emphasisPulse 1.06 + 三词 1211/1215/1219 依次 12 帧白闪(QC v1:原仅 3px 下划线,1167–1240 静止 73 帧);
* 1238/1248/1258 标题中 Retrieval / Augmented / Generation 依次 11 帧灰→紫;1238 大号英文 slideUp(Δ160/22);1323–1328 整组 6 帧淡出。
*/
const F0 = 1130;
const T_YEAR = 1138, T_CARD = 1146, T_PILL = 1152, T_UNDER = 1211, T_WORDS = 1238, T_BIG = 1238, T_EXIT = 1323;
const CARD = {x: 140, y: 140, w: 380, h: 460};
const YEAR_KF: Array<[number, number]> = [[T_YEAR, 2016], [T_YEAR + 3, 2017], [T_YEAR + 6, 2018], [T_YEAR + 9, 2019], [T_YEAR + 13, 2020]];
/** 摘要灰线:[col(0/1), row, 宽度比例] */
const ABS_ROWS = 8;
const ABS_W = [0.98, 0.9, 1, 0.84, 0.95, 1, 0.88, 0.6, 0.97, 0.93, 1, 0.8, 0.9, 0.55];
const CHART_BARS: Array<[number, string]> = [[30, GREY_MID], [44, GREY_MID], [58, PURPLE], [74, PURPLE]];
/** 标题词:k = 灰→紫进度(1238 起);f = 白闪强度 0→1→0(1211 起,与 k 不重叠) */
const Word: React.FC<{k: number; f?: number; children: React.ReactNode}> = ({k, f = 0, children}) => (
<span style={{color: f > 0 ? mixHex(GREY_LIGHT, WHITE, f) : mixHex(GREY_LIGHT, PURPLE_LIGHT, k), textShadow: f > 0 ? `0 0 ${Math.round(6 + 10 * f)}px rgba(255,255,255,${(0.85 * f).toFixed(2)})` : k > 0.5 ? '0 0 8px rgba(161,117,241,.6)' : undefined, fontWeight: 700}}>{children}</span>
);
/** 12 帧三角脉冲 0→1→0 */
const tri = (n: number, len = 12) => (n <= 0 || n >= len ? 0 : n < len / 2 ? n / (len / 2) : 1 - (n - len / 2) / (len / 2));
export const SC09: React.FC = () => {
const N = useCurrentFrame() + F0;
const ex = N - T_EXIT;
const exOp = ex <= 0 ? 1 : Math.max(0, 1 - ex / 6);
const year = stepKf(N, YEAR_KF);
const cn = N - T_CARD;
const cardY = CARD.y + slideUp(cn, 90, 22);
const cardOp = cn < 0 ? 0 : fadeIn(cn, 8);
const under = clamp01((N - T_UNDER) / 14);
const fl = [0, 1, 2].map((i) => tri(N - (T_UNDER + 4 * i), 12));
const titleS = emphasisPulse(N - T_UNDER, {peak: 1.06, up: 8, hold: 4, down: 10});
const kw = [0, 1, 2].map((i) => clamp01((N - (T_WORDS + 10 * i)) / 11));
const gn = N - T_BIG;
const bigDy = slideUp(gn, 160, 22);
const bigOp = gn < 0 ? 0 : fadeIn(gn, 10);
return (
<div style={{position: 'absolute', inset: 0, opacity: exOp}}>
{/* 论文卡 */}
{cardOp > 0 ? (
<Box x={CARD.x} y={cardY} w={CARD.w} h={CARD.h} r={8} sw={3} opacity={cardOp} style={{filter: BLOOM_SOFT}}>
{/* 标题三行 */}
<div style={{position: 'absolute', left: 20, right: 20, top: 30, textAlign: 'center', fontFamily: FONT_SERIF, fontSize: 22, lineHeight: '28px', color: GREY_LIGHT, fontWeight: 700, whiteSpace: 'nowrap', transformOrigin: '50% 50%', transform: titleS === 1 ? undefined : `scale(${titleS.toFixed(4)})`}}>
<div><Word k={kw[0]} f={fl[0]}>Retrieval</Word>-<Word k={kw[1]} f={fl[1]}>Augmented</Word> <Word k={kw[2]} f={fl[2]}>Generation</Word></div>
<div>for Knowledge-Intensive</div>
<div>NLP Tasks</div>
</div>
{/* 下划线(1211 draw-on,左锚) */}
{under > 0 ? <div style={{position: 'absolute', left: 40, top: 122, width: (CARD.w - 80) * under, height: 2.5, background: PURPLE_LIGHT, boxShadow: '0 0 8px rgba(102,45,248,.7)'}} /> : null}
{/* 作者行 */}
<div style={{position: 'absolute', left: 0, right: 0, top: 138, textAlign: 'center', fontFamily: FONT_SERIF, fontSize: 22, lineHeight: '26px', color: GREY, whiteSpace: 'nowrap'}}>Lewis et al. · NeurIPS 2020</div>
<div style={{position: 'absolute', left: 24, top: 176, width: CARD.w - 48, height: 1.5, background: GREY, opacity: 0.7}} />
{/* 摘要两栏灰线 */}
{Array.from({length: ABS_ROWS}, (_, i) => (
<React.Fragment key={i}>
<div style={{position: 'absolute', left: 24, top: 194 + i * 17, width: 156 * ABS_W[i], height: 3, background: GREY_MID}} />
<div style={{position: 'absolute', left: 200, top: 194 + i * 17, width: 156 * ABS_W[(i + 5) % ABS_W.length], height: 3, background: GREY_MID}} />
</React.Fragment>
))}
{/* 左下继续文本;右下小图表 */}
{Array.from({length: 6}, (_, i) => <div key={`b${i}`} style={{position: 'absolute', left: 24, top: 344 + i * 17, width: 156 * ABS_W[(i + 2) % ABS_W.length], height: 3, background: GREY_MID}} />)}
<div style={{position: 'absolute', left: 200, top: 340, width: 156, height: 96, boxSizing: 'border-box', border: `1.5px solid ${GREY}`}}>
{CHART_BARS.map(([h, c], i) => <div key={i} style={{position: 'absolute', left: 16 + i * 32, bottom: 12, width: 20, height: h, background: c, opacity: c === PURPLE ? 0.9 : 0.8}} />)}
<div style={{position: 'absolute', left: 8, right: 8, bottom: 11, height: 1.5, background: GREY}} />
</div>
</Box>
) : null}
{/* 右:年份计数 */}
<SoftIn N={N} f0={T_YEAR}>
<CText cx={860} cy={300} size={96} weight={700} family={FONT_ORB} letterSpacing={4} color={WHITE} shadow={TEXT_GLOW} style={{fontVariantNumeric: 'tabular-nums'}} dy={-2}>
{year}
</CText>
</SoftIn>
{/* 胶囊 Facebook AI Research */}
<SoftIn N={N} f0={T_PILL}>
<Pill x={690} y={376} w={340} h={50} sw={2.5} text="Facebook AI Research" fontSize={26} weight={700} letterSpacing={0.5} style={{filter: BLOOM_SOFT}} />
</SoftIn>
{/* 大号英文两行 */}
{bigOp > 0 ? (
<div style={{position: 'absolute', inset: 0, opacity: bigOp}}>
<TechText cx={860} cy={495 + bigDy} text="Retrieval-Augmented" fontSize={42} scaleX={0.82} weight={700} />
<TechText cx={860} cy={545 + bigDy} text="Generation" fontSize={42} scaleX={0.82} weight={700} />
</div>
) : null}
</div>
);
};
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 469k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |