복고풍 픽셀 스타일의 세 장면 영상을 Hyperframes용 HTML로 만듭니다. 제목과 색상을 바꾸고 장면 전환과 움직임을 갖춘 구성으로 완성합니다.
- 픽셀 스타일의 소개 영상을 만들 때
- 세 장면 영상의 내용을 교체할 때
- HTML에서 장면 전환을 확인할 때
버튼과 메뉴 등 화면의 애니메이션을 설계·구현·검토해 반응과 움직임을 다듬습니다. 녹화에서 속도 변화와 스프링 움직임을 분석하고 재현 코드와 구현 명세를 정리합니다.
Implement, review, and measure UI motion. Reconstruct easing, springs and gestures from a recording using bundled frame extraction, tracking and curve-fitting tools when available.
버튼, 메뉴, 대화 상자, 드래그 조작의 움직임을 설계·구현·검토합니다. 움직임을 넣는 목적을 확인한 뒤 지속 시간, 속도 변화를 나타내는 이징, 스프링 움직임의 설정을 선택합니다. 참고 녹화에서 프레임을 추출하고 요소의 이동을 추적해 곡선을 맞춘 다음, 재현 코드와 구현 명세로 정리하는 절차도 제공합니다.
조작 반응이 느리거나 움직임이 어색할 때, 동작 도중 다시 조작하면 부자연스러워지는 문제를 조사할 때 유용합니다. 기존 프로젝트의 기술 구성을 사용하며, 동작 줄이기 설정과 터치 조작, 렌더링 부하도 확인합니다.
녹화를 정밀하게 측정하려면 ffmpeg, ffprobe와 opencv-python, numpy, scipy를 사용할 수 있는 Python 환경이 필요합니다. 일반적인 움직임 구현에는 필수가 아닙니다. 보고할 때는 눈으로 추정한 값과 실제 측정값을 구분합니다.
이 소개문은 공개된 SKILL.md를 바탕으로 AI(OpenAI Codex)가 작성했습니다. 정확한 사양은 아래 원문을 확인하세요.
AI가 SKILL.md의 내용만을 근거로 4가지 관점에서 판정했습니다. 스타 수나 인기도는 판정에 넣지 않았습니다.
총점은 각 항목을 A=25점, B=15점, C=5점으로 더한 값입니다. 2026년 10월 20일 기준으로 AI(OpenAI Codex)가 판정했습니다. 참고용이므로 설치 전에 원문도 확인해 주세요.
설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.
Use this as a functional helper on the user's current interface or accepted brief. Keep the host's task route, current stage, tool availability, and output contract. Planning describes the motion; implementation happens only in the host's authoring stage. Do not add an extra review, recording, rendering, or approval phase when the host disallows it. Reuse the project's existing stack and motion tokens. Companion skills are optional: use them only if installed; otherwise apply the relevant guidance here without inventing a tool invocation. Report measurements only when actually observed; mark visual estimates and checks that could not be performed explicitly.
ui-design Direction mode), auditing a whole page's UI quality (use ui-design Audit mode), or named text-effect specs (use the external animate-text skill where installed).| File | Read when |
|---|---|
| references/decision-framework.md | Default: deciding whether/why to animate, picking easing character; also the Discovery sweep ("where should this animate") |
| references/spring-animations.md | Spring physics, Motion useSpring, configuring spring params, Apple damping/response values, asymmetric open/close character, interruption mechanics |
| references/component-patterns.md | Buttons, toasts, crossfades, list stagger and removal, hover, step forms, Motion layout morphs and auto height, 3D |
| references/clip-path-techniques.md | clip-path for reveals, tabs, hold-to-delete, comparison sliders |
| references/gesture-drag.md | Drag, swipe-to-dismiss, momentum, pointer capture, velocity handoff, momentum projection, rotary/knob drag, detents, carousel touch-action |
| references/scroll-animations.md | Scroll-triggered reveals, scrubbed/scroll-driven animation (animation-timeline, useScroll), parallax, sticky scrollytelling, and when a scroll animation shouldn't exist |
| references/performance-deep-dive.md | Jank, CSS vs JS, WAAPI, CSS variables trap, Framer Motion caveats |
| references/debugging-symptoms.md | An animation feels off and the cause isn't named: symptom-indexed tables for sluggish, robotic, cheap, jumpy, and misfiring motion |
| references/svg-animation.md | Animating vector art: line drawing (stroke-dashoffset), SVG transform-origin traps, path morphing, shakes, ambient life |
| references/review-format.md | Reviewing animation code: ten standards (each with flag-on-sight triggers), Before/After/Why table, Block/Approve verdict |
| references/contextual-animations.md | Contextual icon swaps, word-level stagger entrances, peripheral de-emphasis, fixed-offset exits |
| references/transition-recipes.md | Installing a CSS transition: container morph, card resize, badge, dropdown and tooltip (with library origin variables), modal (with @starting-style entry), panel and drawer, page slide, icon swap, number pop-in, odometer roll, text swap, success, avatar hover, error shake |
| references/measurement-guide.md | Reverse-engineer: what to measure, eye vs script, reading metrics.json, choosing an ROI |
| references/curve-fitting.md | Reverse-engineer: reading fit_curves.py output, spring vs bezier, judging fit error, asymmetric open/close |
| references/code-output.md | Reverse-engineer: emitting code for CSS, Motion/Framer Motion, SwiftUI, React Native, UIKit |
| references/choreography.md | Reverse-engineer: multi-element/multi-phase motion: staggers, blur-before-move, per-edge settling |
| references/vocabulary.md | Naming a motion effect the user describes vaguely ("what's it called when...") |
| references/interface-sfx.md | Click sounds, interface audio, UI SFX, haptic-plus-sound, or "why is the web afraid of sound" |
requestAnimationFrame); under load CSS stays smooth while JS drops frames.:active at 0ms and set touch-action: manipulation.@starting-style for DOM entry; fall back to a data-mounted attribute where unsupported.filter: blur(2px) hides rough crossfades between swapped content.transform and opacity only; they skip layout and paint.color, background-color, and opacity are acceptable.width, height, top, left); they trigger layout recalc every frame. (Exception: a deliberate container tween, see the card-resize and container-morph recipes.)transition: all; it animates unintended properties and silently adopts future ones. List them explicitly.filter animation for core interactions; if unavoidable keep blur ≤ 20px (heavy blur is expensive, especially in Safari).<g> wrapper with transform-box: fill-box; transform-origin: center; without it they rotate/scale around the canvas origin. Line drawing, path morphing, and the Motion SVG origin override live in references/svg-animation.md.transform: scale() also scales children (icons, text, borders scale proportionally), unlike width/height: a feature for press feedback, but account for it when an inner element must stay fixed-size.[data-theme-switching] * { transition: none !important }), or every themed property animates at once. Force a reflow (void document.body.offsetHeight) after the flip and remove the override on the next frame, or use next-themes disableTransitionOnChange.| Element | Duration | Easing |
|---|---|---|
| Button press feedback | 100-160ms | cubic-bezier(0.22, 1, 0.36, 1) |
| Tooltips, small popovers | 125-200ms | ease-out or enter curve |
| Dropdowns, selects | 150-250ms | cubic-bezier(0.22, 1, 0.36, 1) |
| Modals, drawers | 200-350ms | cubic-bezier(0.22, 1, 0.36, 1) |
| Move/slide on screen | 200-300ms | cubic-bezier(0.25, 1, 0.5, 1) |
| Page transitions | 250-400ms | enter or move curve |
| Hover (colour/opacity), one control | 200ms | ease |
| Hover (transform/scale) | 100-150ms | enter curve |
| Illustrative/marketing | Up to 1000ms | Spring or custom |
Keep routine UI under 300ms; scale duration with distance (a full-screen slide can exceed 300ms, a 6px tooltip shift stays under 150ms).
Named curves
cubic-bezier(0.22, 1, 0.36, 1) for entrances and transform-based hovercubic-bezier(0.25, 1, 0.5, 1) for slides, drawers, panelscubic-bezier(0.32, 0.72, 0, 1) (extremely steep start; the reason its 500ms doesn't read as slow)cubic-bezier(0.19, 1, 0.22, 1) for dramatic reveals, card hovers, text revealscubic-bezier(0.25, 0.46, 0.45, 0.94) for button press feedbackcubic-bezier(0.645, 0.045, 0.355, 1) for back-and-forth movement that stays on screenAvoid ease-in for UI: it starts slow, so the element lags the user's action and feels sluggish. Prefer custom curves from easing.dev over built-in ease/ease-out, whose gentle acceleration reads soft, not decisive.
Match the UI element first, then pick the recipe from references/transition-recipes.md:
| UI pattern | Recipe |
|---|---|
| Trigger + floating dot/count | Notification badge |
| Trigger grows into the surface it opens | Container morph |
| Trigger + anchored surface | Menu dropdown |
| Centred surface on top of page | Modal dialog |
| Panel sliding into existing container | Panel reveal |
| List ↔ detail or wizard steps | Page side-by-side slides |
| Element dimension changes | Card resize |
| Text updating in place | Text state swap |
| Two icons in same slot | Icon swap |
| Number arriving on its own | Number pop-in |
| Number the user is driving | Odometer digit roll |
| Confirmation / success moment | Success celebration |
| Hovering item in horizontal stack | Avatar group hover |
| Form validation error | Error state shake |
Prefer lower-overhead transitions (CSS-only) unless the design requires JS orchestration.
transform-origin at the trigger (modals stay center), dialog/menu entrances from scale(0.9-0.96) not scale(0) (small popovers at the low end, full dialogs at the high end: a large surface already travels far in absolute pixels), and 30-50ms staggers (total under 300ms, most important element leading). Full rules and code in references/transition-recipes.md and references/contextual-animations.md.@media (hover: hover) and (pointer: fine), or touch devices replay hover on tap. Inspect the generated CSS before adding a gate; Tailwind v4 already wraps hover: in @media (hover: hover).IntersectionObserver; they burn GPU even when invisible.will-change only during heavy motion and only for transform/opacity; remove it after. Each promotion costs compositor memory; permanent promotion across many elements is worse than none.transform directly on the moving element.x/y values are the default for axis movement and drag (they bypass React re-renders). Use a full transform string when one owner must combine multiple transform functions, interop with non-Motion code, or survive a busy main thread: the shorthands run on requestAnimationFrame and drop frames when motion coincides with navigation, data loading, or hydration; CSS/WAAPI stay smooth there.transitionend.High-signal failures not covered above:
linear and no duration is correct there, and only there.@media (prefers-reduced-motion) block, so the CSS gate passes review while spatial travel still plays. Wrap the app in <MotionConfig reducedMotion="user"> or branch on useReducedMotion(). The opposite overcorrection, a global * { transition: none !important } under reduce, also strips the opacity and colour feedback that reduced motion keeps.framer-motion for new work: the package is now motion and React imports come from motion/react. The old package still resolves, so a mixed codebase compiles while shipping two copies of the library.Copy and track:
Animation progress:
- [ ] Step 1: Decide whether the interaction should animate
- [ ] Step 2: Choose purpose, easing, and duration
- [ ] Step 3: Pick the implementation style
- [ ] Step 4: Load the relevant component or technique reference
- [ ] Step 5: Validate timing, interruption, and device behavior
Produce evidence for each check (DevTools observations, not "looks fine"):
width, height, top, left) and transition: all.transform-origin issues invisible at full speed.will-change is toggled around animations, not permanently set, and looping animations pause off-screen.prefers-reduced-motion: replace spatial travel with immediate state changes or restrained fades. Pause looping decorations with animation-play-state: paused (do not yank them with display: none). Keep explicit user-triggered feedback. Exercise the same task in that mode.For "where should this animate", run the Discovery sweep in references/decision-framework.md. Report opportunities supported by purpose and usage frequency, plus the rejected ones. Implement a suggestion only when implementation is in scope.
Use this branch to measure an existing animation from a screen recording, then emit code and a handoff spec that reproduce it. The scripts under scripts/ are the canonical, deterministic path; run them rather than reconstructing their logic.
Resolve every scripts/ command below relative to the installed skill directory, not the application working directory.
Dependencies: Check for ffmpeg / ffprobe and a Python environment with
opencv-python, numpy, and scipy before running the measurement scripts.
Do not assume these are installed by Open Design. Reuse an available isolated
Python environment; any dependency setup must respect the host's authorization.
If only ffmpeg is available, extract frames and label the analysis as a visual
estimate. If media inspection is unavailable, state the missing capability;
never claim measured timing, fitted curves, or a recording comparison.
Use a fresh task-local output directory for each extraction so old frames cannot
contaminate the contact sheet. The scripts below are optional measurement tools,
not a requirement for ordinary interface-animation implementation.
Reverse-engineer progress:
- [ ] Step 1: Extract frames + contact sheet (per direction if open differs from close)
- [ ] Step 2: Vision pass: identify element, effects, phases
- [ ] Step 3: Decide precision (eye-only vs scripted)
- [ ] Step 4: Track motion and fit curves (if escalating)
- [ ] Step 5: Annotate choreography (delays, asymmetry)
- [ ] Step 6: Emit code for the target(s)
- [ ] Step 7: Validate against the recording
python3 scripts/extract_frames.py <video> <outdir>. Trim to just the transition with --start/--duration; if the interaction has both an open and a close, trim two windows and run the pipeline once per direction (they are almost never mirror images). Match --fps to the source (probe with ffprobe), never sampling above the source rate. Open contact_sheet.png first.references/measurement-guide.md.python3 scripts/track_motion.py <outdir> for metrics.json (pass --bbox X,Y,W,H to isolate one element), then python3 scripts/fit_curves.py <outdir>/metrics.json for spring params, cubic-bezier, and per-property fit error. Pass the same --fps you extracted with. Read references/curve-fitting.md to pick the model; high error on both means multi-phase motion (split and fit each segment).references/choreography.md. Build the timing-offset table (when each property starts and settles); lead/lag gaps and over-stretch carry more feel than any single curve.references/code-output.md for the target. Keep movement on transform/opacity. Emit two transitions when open and close differ, plus the consolidated handoff spec so it can be implemented without the video.extract_frames.py, and compare contact sheets side by side. Slow to 0.1x to confirm phase order and over-stretch survive. Confirm the code only animates transform, opacity, and filter.Reverse-engineer gotchas:
fit_curves.py defaults to --fps 30: extract at 60 but fit at the default and every duration_ms doubles while fitted stiffness drops to a quarter. Always pass the extraction fps to the fit.metrics.json. Probe and match the source rate.references/choreography.md). Treat a fit error above 0.08 as suspect.product-design: which states exist, what an action affects, and whether it is reversible. Route here first when a gesture replaces a control, since swipe-to-delete and hold-to-confirm change what the user can do before they change how it moves.ui-design Direction mode: visual direction, palettes, typography; settle the visual system before tuning motion.ui-design Audit mode: page/feature-level UI quality audit. Motion craft and fixes belong here.animate-text skill where installed: curated named text effects (typewriter, line reveal, stagger builds) with exact JSON specs.Upstream evaluation fixtures are not part of this runtime bundle. Source and adaptation details are recorded in SOURCES.md.
아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.
로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.
아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.
복고풍 픽셀 스타일의 세 장면 영상을 Hyperframes용 HTML로 만듭니다. 제목과 색상을 바꾸고 장면 전환과 움직임을 갖춘 구성으로 완성합니다.
소셜 미디어 광고와 검색 광고의 제목, 설명, 본문을 작성하고 표현을 반복해서 수정합니다. 광고 문구 후보를 준비하고 개선안을 검토할 때 활용합니다.
어두운 배경과 세리프 글꼴로 패션 잡지 같은 세 페이지 동적 연출을 만듭니다. 글자 단계 표시와 장면 전환을 포함한 단일 HTML 파일로 출력합니다.
브라우저에서 페이지를 살펴보고 클릭, 입력, 화면 캡처, 정보 추출을 수행합니다. OpenDesign의 로컬 미리보기를 중심으로 웹 앱의 동작과 품질을 확인합니다.
AI 음악 앨범의 콘셉트 구상부터 가사 초안, 곡 순서 검토, 내보내기까지 다룹니다. 독립 제작 실험과 브랜드 음악 기획에 활용합니다.
p5.js에서 계산 규칙과 난수로 생성 예술을 만듭니다. 난수의 초깃값을 고정해 같은 그림을 재현하며, 포스터와 동적인 느낌의 정지 이미지, 구도 시안 제작에 활용합니다.