@autoaiai/aicut 0.1.11 → 0.1.13

package/skills/cinematic-product-promo-skill/references/shots/camera/depth-layer-moves.md

@@ -1,40 +0,0 @@
----
-name: depth-layer-moves
-一句话: 分层深度两款运镜——多层视差滑轨（3 层速度梯度横移出纵深）与伪 dolly-zoom（主体钉死、背景膨胀压来）
-适用: 平面截图要"有厚度"的段落；戏剧性蓄力时刻用 dolly-zoom（一支片 ≤1 次）
-时长: 视差滑轨 4–5s 持续；dolly-zoom 3–4s 单向行程
-能量: 视差=中（质感型）；dolly-zoom=中高（压迫感渐强）
----
-
-## 意图
-整页平移是"看页面"，分层深度是"在页面的空间里"。视差滑轨给横移
-加纵深（迪士尼多平面摄影机原理，Linear 片同款质感）；伪 dolly-zoom
-反过来——主角纹丝不动、全世界压过来，给戏剧性时刻蓄力。
-
-## 动效核心
-- **视差滑轨（3 层）**：背景整页（系数 0.35 + blur 2px + 降饱和
-  0.92 + opacity 0.85，退成环境）、中景真实卡切片（系数 0.7，
-  主阅读层，无 blur）、前景浮块（系数 1.4 + blur 3px，掠过镜头）。
-  同一条 drive 位移乘各层系数
-- **伪 dolly-zoom**：主体高清卡固定 px 钉屏中（不参与任何变换）；
-  背景层（整页 + 卡群一体）scale 1→2.25 从画面中心膨胀 + blur 0→3.5px
-  渐深；主体落影随进程加深（12→28px offset）强化"钉住感"
-- 两款都不拆真实页面排版：整页作为一体成层，独立元素用切片另起层
-  ——层间关系是"景深"，不是"页面被撕开"
-
-## 参数表
-| 参数 | 典型值 | 调节手感 |
-|------|--------|----------|
-| 视差层系数 | 0.35 / 0.7 / 1.4 | 梯度 ≥2 倍层间才可辨；层数 ≤4 防过载 |
-| 层次锚 | 前后景 blur + 背景降饱和 | 没有 blur/饱和锚，读作"贴片乱飞" |
-| dolly 膨胀行程 | scale 1→2.0–2.5 | 行程小了压迫感不足；背景 opacity ≤0.6 防读作故障 |
-
-## 已知坑
-- 主阅读层/主体必须高清纹理且无 blur——观众在读它
-- 视差与 PageCam rot 3D 的组合（斜拍视差）未验证，实战先渲 still 确认
-- dolly-zoom 是"戏剧调料"：一支片 ≤1 次，日常段落用视差滑轨
-- 参数经占位素材调校转正，非实战定稿，首次实战后回验
-
-## 参考实现
-demos/camera/depth-layer-moves/
-（DollyZoomReal.tsx / MultiplaneReal.tsx）

package/skills/claude-gif/SKILL.md

@@ -1,211 +0,0 @@
----
-name: claude-gif
-description: "claude-gif -- The Ultimate GIF Creator: the method the original skill taught, run as an AICUT film — pages that draw any moment t, cut on tracks. Runs in AICUT: the film is film.html, a page per shot, drawn by `aicut look` and `aicut render`."
-license: MIT
-source: "AgriciDaniel/claude-gif"
----
-
-# claude-gif -- The Ultimate GIF Creator
-
-## Running it in AICUT
-
-AICUT's film is `film.html`: it cuts pages, footage, pictures and sound together on tracks. A page is a web page that
-sets `window.film = { frame(t) }` and draws that moment — no framework, no build step, no composition object, no
-`registerRoot`. Five shots are five pages cut together, not one page that changes scene, so each shot stays trimmable
-and replaceable in Studio.
-
-```sh
-npx -y @autoaiai/aicut@latest          # the manual: the format and the commands, always current
-npx -y @autoaiai/aicut@latest open .   # make this folder a film and print Studio's address
-npx -y @autoaiai/aicut@latest look     # a contact sheet of the whole film, its sound, and a determinism check
-npx -y @autoaiai/aicut@latest look 4.5 # one frame, full size — look at the PNG, not at the code
-npx -y @autoaiai/aicut@latest render   # the MP4, beside film.html
-```
-
-Run the manual before the first film: it is authoritative for the format, and this skill is about the method above it.
-
-Where the words below name another program's part, it becomes this:
-
-| elsewhere | in AICUT |
-|---|---|
-| composition, timeline, project file | `film.html`; tracks and clips inside it |
-| a scene, a sequence, a slide | a page (`pages/<name>.html`) cut onto a track |
-| `useCurrentFrame()`, `interpolate()` | `frame(t)` — the picture depends only on `t`, never on state or `Date.now()` |
-| a still, a screengrab | `aicut look <t>` |
-| `render` / `build` / `export` | `aicut render` (the person can also export from Studio) |
-| a dev server, a preview window | `aicut open` and the Studio address it prints |
-
-Hard rules that hold whatever the method says: the picture depends only on `t`; every value in the person's prompt is
-binding; look at the frames `look` writes before saying a film is done.
-
-
-Orchestrates five sub-skills to create, generate, convert, optimize, and edit GIFs.
-Every path from concept to final optimized GIF runs through this skill.
-
-## Quick Reference
-
-| Command | Description | Sub-Skill |
-|---------|-------------|-----------|
-| `/gif create <idea>` | Programmatic animation via Remotion (text, spinners, memes, motion graphics) | claude-gif-create |
-| `/gif generate <prompt>` | AI video generation via Veo 3.1, then convert to GIF | claude-gif-generate |
-| `/gif convert <path>` | Convert video, image sequence, or animated SVG to GIF | claude-gif-convert |
-| `/gif optimize <path>` | Reduce GIF file size for a target platform | claude-gif-optimize |
-| `/gif edit <path>` | Modify an existing GIF (speed, reverse, crop, text, trim) | claude-gif-edit |
-| `/gif setup` | Install/verify all dependencies | (setup.sh) |
-| `/gif <idea>` | Auto-detect best approach from description | (orchestrator routes) |
-
-## Orchestration Logic
-
-When the user invokes `/gif`, determine intent and route to the correct sub-skill.
-
-### Intent Detection
-
-1. **Has a file path to a video/image/SVG?** --> `/gif convert`
-2. **Has a file path to an existing GIF?**
-   - Mentions size, compress, platform fit --> `/gif optimize`
-   - Mentions speed, reverse, crop, text, trim --> `/gif edit`
-3. **Describes a text animation, spinner, counter, meme, logo animation, progress bar?** --> `/gif create`
-4. **Describes realistic/cinematic motion, product demo, abstract art, nature scene?** --> `/gif generate`
-5. **Ambiguous description of something to animate?**
-   - If it involves text, geometric shapes, UI elements --> `/gif create` (Remotion)
-   - If it involves photorealistic subjects, complex motion --> `/gif generate` (Veo)
-6. **Mentions "optimize", "compress", "reduce", "fit", platform name?** --> `/gif optimize`
-7. **Mentions "edit", "speed", "reverse", "crop", "text overlay"?** --> `/gif edit`
-8. **Says "setup" or "install"?** --> Run `bash ~/.claude/skills/claude-gif/scripts/setup.sh`
-
-### Routing Commands
-
-```
-/gif create "bouncing hello world text"      --> claude-gif-create
-/gif generate "campfire flickering at night"  --> claude-gif-generate
-/gif convert ~/Videos/clip.mp4               --> claude-gif-convert
-/gif convert ~/Design/logo.svg               --> claude-gif-convert (Mode D: SVG)
-/gif optimize ~/Documents/gif_output/big.gif --> claude-gif-optimize
-/gif edit ~/Documents/gif_output/clip.gif    --> claude-gif-edit
-/gif setup                                   --> scripts/setup.sh
-```
-
-### Bare `/gif <description>` Routing
-
-When no explicit sub-command is given, analyze the description:
-- Keywords "text", "spinner", "loading", "counter", "meme", "typewriter", "logo reveal", "progress" --> create
-- Keywords "realistic", "cinematic", "product", "nature", "abstract", "person", "animal" --> generate
-- A file path to .mp4/.webm/.mov/.avi/.mkv/.svg --> convert
-- A file path to .gif + size/platform keywords --> optimize
-- A file path to .gif + edit keywords --> edit
-
-## Quality Presets
-
-| Preset | Width | FPS | Colors | Dither | Max Size | Best For |
-|--------|-------|-----|--------|--------|----------|----------|
-| `discord` | 320px | 10 | 128 | bayer:3 | 256 KB | Discord inline embeds |
-| `slack` | 400px | 12 | 192 | floyd_steinberg | 500 KB | Slack messages |
-| `twitter` | 480px | 15 | 256 | floyd_steinberg | 15 MB | Twitter/X posts |
-| `web` | 480px | 15 | 256 | floyd_steinberg | 2 MB | General web usage |
-| `hq` | 640px | 20 | 256 | sierra2 | 10 MB | High-quality display |
-
-## Safety Rules
-
-1. **No-overwrite**: Always pass `-n` to FFmpeg (skip if output exists). Only pass `-y` when user explicitly says overwrite.
-2. **Source protection**: NEVER write output to the same path as input. Preflight catches this.
-3. **Cost confirmation**: Veo generation costs money. Always confirm with user before calling generate.py.
-4. **Temp isolation**: All intermediate files go in `/tmp/claude-gif/`. Clean up on completion.
-5. **Output directory**: Final GIFs go to `~/Documents/gif_output/` unless user specifies a different path.
-6. **Preflight check**: Run `bash ~/.claude/skills/claude-gif/scripts/preflight.sh <input> <output>` before any file write.
-
-## Multi-Step Pipelines
-
-Complex requests may chain sub-skills:
-
-```
-generate --> convert --> optimize
-  (Veo)    (to GIF)   (for Discord)
-
-create --> optimize
-(Remotion)  (shrink)
-
-convert --> edit --> optimize
-(video)   (trim)   (platform)
-
-convert (SVG) --> optimize
-  (Playwright)    (for web)
-```
-```
-Always plan the full pipeline before starting. Tell the user the steps.
-```
-
-## Pre-Flight
-
-Before ANY write operation:
-```bash
-bash ~/.claude/skills/claude-gif/scripts/preflight.sh <input_file> <output_file>
-```
-```
-On `/gif setup` or first use:
-```
-```bash
-bash ~/.claude/skills/claude-gif/scripts/check_deps.sh
-```
-```
-If dependencies are missing:
-```
-```bash
-bash ~/.claude/skills/claude-gif/scripts/setup.sh
-```
-
-## Reference Files
-
-Load on demand -- only read the reference relevant to the current task:
-
-| Reference | Path | When to Load |
-|-----------|------|-------------|
-| GIF Optimization | `~/.claude/skills/claude-gif/references/gif-optimization.md` | Before optimize, or choosing palette/dither settings |
-| Platform Specs | `~/.claude/skills/claude-gif/references/platform-specs.md` | When targeting a specific platform |
-| Remotion Patterns | `~/.claude/skills/claude-gif/references/remotion-gif.md` | Before writing Remotion components |
-| Prompt Engineering | `~/.claude/skills/claude-gif/references/prompt-engineering.md` | Before constructing Veo prompts |
-| Perfect Loops | `~/.claude/skills/claude-gif/references/perfect-loops.md` | When making seamless loops |
-
-## Scripts
-
-| Script | Path | Purpose |
-|--------|------|---------|
-| `setup.sh` | `scripts/setup.sh` | Install and verify all dependencies |
-| `check_deps.sh` | `scripts/check_deps.sh` | Check dependency status (JSON, no installs) |
-| `preflight.sh` | `scripts/preflight.sh` | Safety check before writes (input/output validation) |
-| `gif_convert.sh` | `scripts/gif_convert.sh` | FFmpeg two-pass palette video-to-GIF conversion |
-| `gif_optimize.py` | `scripts/gif_optimize.py` | Multi-strategy GIF size optimizer (Python) |
-| `gif_loop.py` | `scripts/gif_loop.py` | Perfect loop creator: crossfade, ping-pong, freeze blend (Python) |
-| `gif_frames.py` | `scripts/gif_frames.py` | Assemble PNG/JPEG frames into optimized GIF (Python) |
-
-All scripts are in `~/.claude/skills/claude-gif/scripts/`.
-Python scripts use `~/.video-skill/bin/python3` (venv with Pillow, numpy).
-
-## Sub-Skills
-
-| Sub-Skill | When to Use |
-|-----------|-------------|
-| `claude-gif-create` | Programmatic animations: text, spinners, counters, memes, logos, progress bars, motion graphics. Uses Remotion (React + Node.js). |
-| `claude-gif-generate` | AI-generated motion: cinematic loops, product demos, abstract art, realistic subjects. Uses Veo 3.1 (costs money). |
-| `claude-gif-convert` | File conversion: video to GIF, image sequence to GIF, AI image sequence to GIF, animated SVG to transparent GIF. |
-| `claude-gif-optimize` | Size reduction: platform auto-fit, lossy compression, color/dimension/frame reduction, dither tuning. |
-| `claude-gif-edit` | Modify existing GIFs: speed change, reverse, ping-pong, crop, resize, text overlay, frame extraction, loop control, trim. |
-
-## Workflow Example
-
-User: "Make me a Discord-sized GIF of a cozy campfire loop"
-
-1. Route to `claude-gif-generate` (realistic motion = Veo)
-2. Read `references/prompt-engineering.md` for loop-friendly prompt patterns
-3. Confirm cost with user (~$0.60 for 4s Veo Fast)
-4. Generate 4s campfire video via Veo
-5. Convert to GIF: `gif_convert.sh --input campfire.mp4 --preset discord`
-6. Perfect loop: `gif_loop.py --input campfire.gif --method crossfade --frames 5`
-7. Verify size < 256KB. If over, route through `claude-gif-optimize`
-8. Deliver to `~/Documents/gif_output/campfire_discord.gif`
-
-## Source
-
-Rewritten for AICUT from [`AgriciDaniel/claude-gif`](https://github.com/AgriciDaniel/claude-gif)
-(MIT), skills/claude-gif/SKILL.md. The method is that skill's own; the way it runs is AICUT's.
-
-The repository holds 6 skills; this is the one its entry stands for.

package/skills/claude-gif/references/gif-optimization.md

@@ -1,209 +0,0 @@
-# GIF Optimization Reference
-
-Comprehensive technical reference for GIF optimization. Load this before any
-optimization task or when choosing palette/dither settings during conversion.
-
-## Palette Generation Strategies (FFmpeg palettegen)
-
-FFmpeg's `palettegen` filter creates a 256-color palette from the source frames.
-The `stats_mode` parameter controls HOW colors are sampled:
-
-| stats_mode | Method | Best For | File Size Impact |
-|------------|--------|----------|-----------------|
-| `full` | Sample all pixels from all frames equally | General purpose, evenly animated content | Baseline |
-| `diff` | Weight pixels that change between frames | Animations with static backgrounds | **Smaller** (static areas get fewer palette slots) |
-| `single` | Generate a new palette per frame | Content with dramatic color changes | **Larger** (more complex encoding) |
-
-### When to Use Each
-
-- **`full`** (default): Safe choice for most content. Every frame contributes equally to the palette.
-- **`diff`**: Best for GIFs where a subject moves against a static background. The background gets fewer palette slots, so the moving subject gets higher color fidelity. Also produces smaller files because unchanged pixels compress better.
-- **`single`**: Only use when colors vary drastically between frames (e.g., rainbow transitions, scene cuts). Produces larger files but prevents color banding in individual frames.
-
-**Recommendation**: Start with `diff` for most GIFs. Fall back to `full` if you see color banding in static areas.
-
-## Dithering Algorithms (FFmpeg paletteuse)
-
-Dithering distributes quantization error to simulate more colors than the palette contains.
-
-| Algorithm | Quality | File Size | Speed | Visual Style |
-|-----------|---------|-----------|-------|-------------|
-| `floyd_steinberg` | Highest | Largest | Fast | Natural, film-like grain |
-| `sierra2` | High | Large | Fast | Slightly less grain than Floyd-Steinberg |
-| `sierra2_4a` | Medium-High | Medium | Fast | Simplified Sierra, good balance |
-| `bayer:bayer_scale=N` | Medium | **Smallest** | Fastest | Ordered/crosshatch pattern, retro look |
-| `none` | Lowest | Smallest | Fastest | Hard color boundaries, posterized |
-
-### bayer_scale Parameter
-
-The `bayer_scale` parameter (0-5) controls the dithering matrix size for bayer dithering:
-- `bayer_scale=0`: 2x2 matrix (very visible pattern, smallest files)
-- `bayer_scale=1`: 4x4 matrix
-- `bayer_scale=2`: 8x8 matrix
-- `bayer_scale=3`: 16x16 matrix (recommended balance)
-- `bayer_scale=4`: 32x32 matrix
-- `bayer_scale=5`: 64x64 matrix (subtlest pattern, largest files)
-
-### Choosing Dithering
-
-| Content Type | Recommended Dither | Reason |
-|-------------|-------------------|--------|
-| Photography/video | `floyd_steinberg` | Natural error diffusion hides banding |
-| Flat illustrations | `bayer:3` or `none` | Fewer colors = less error to diffuse |
-| Text animations | `none` or `bayer:3` | Crisp edges, no grain on text |
-| Pixel art / retro | `bayer:2` | Matches aesthetic, intentional pattern |
-| Size-critical (Discord) | `bayer:3` | Smallest files with acceptable quality |
-| Quality-critical | `floyd_steinberg` | Best perceived quality |
-| Transparent GIFs | `bayer:3` | Handles transparency edges cleanly |
-
-## Color Quantization
-
-### Global Palette vs Per-Frame Palette
-
-- **Global palette** (default): One 256-color palette shared across all frames. Smaller files, but colors are a compromise across all frames.
-- **Per-frame palette** (`stats_mode=single`): Each frame gets its own palette. Better color accuracy per frame, but larger files and potential flickering at frame boundaries.
-
-### Color Count Impact
-
-| Colors | Quality | Size Reduction vs 256 |
-|--------|---------|----------------------|
-| 256 | Best | Baseline |
-| 192 | Very Good | ~10-15% smaller |
-| 128 | Good | ~20-30% smaller |
-| 64 | Acceptable | ~40-50% smaller |
-| 32 | Noticeable loss | ~55-65% smaller |
-| 16 | Significant loss | ~65-75% smaller |
-
-### Color Reduction Strategy
-
-1. Start with 256 colors
-2. If file is too large, try 128 first (barely noticeable for most content)
-3. Only go below 64 for extreme size constraints (Discord)
-4. Pair fewer colors with `bayer` dithering (makes reduction less visible)
-
-## diff_mode=rectangle
-
-The `diff_mode` parameter in `paletteuse` controls how frames are encoded:
-
-- **`diff_mode=none`** (default): Each frame is encoded in full.
-- **`diff_mode=rectangle`**: Only the rectangular region that changed between frames is encoded.
-
-**Always use `diff_mode=rectangle`** for animations. It dramatically reduces file size for GIFs where only part of the frame changes (most GIFs). The savings compound:
-- Subject moving on static background: 40-70% reduction
-- Full-frame motion: 5-10% reduction (still worth enabling)
-- Zero cost to quality
-
-## Gifsicle Optimization
-
-### Optimization Levels
-
-| Level | Description | Speed | Savings |
-|-------|-------------|-------|---------|
-| `-O1` | Basic optimization | Fast | 5-10% |
-| `-O2` | Moderate optimization | Medium | 10-15% |
-| `-O3` | Maximum optimization | Slow | 15-20% |
-
-Always use `-O3` -- the speed difference is negligible for typical GIFs.
-
-### Lossy Compression (--lossy=N)
-
-Gifsicle's `--lossy` parameter introduces controlled quality loss for size reduction.
-The value is NOT a percentage -- it's an artifact tolerance scale.
-
-| Value | Quality Impact | Typical Savings |
-|-------|---------------|----------------|
-| 30 | Imperceptible | 10-20% |
-| 60 | Barely visible | 20-30% |
-| 80 | Minor artifacts | 25-40% |
-| 120 | Noticeable | 35-50% |
-| 150 | Visible artifacts | 45-60% |
-| 200 | Significant artifacts | 55-70% |
-
-**Strategy**: Start at 30, increase by 30 until target size is reached. Preview after each step.
-
-### Additional Gifsicle Flags
-
-```bash
---no-comments      # Strip comment metadata
---no-names         # Strip frame name metadata
---no-extensions    # Strip extension blocks
---color-method=diversity  # Better color selection (default)
---color-method=blend-diversity  # Blend similar colors (slightly better quality)
-```
-
-## Transparency Handling
-
-GIF supports binary transparency (each pixel is either fully opaque or fully transparent).
-No semi-transparency / alpha blending.
-
-### FFmpeg Transparent GIF Settings
-
-```
-palettegen:
-  reserve_transparent=1    # Reserve one palette slot for transparency
-  max_colors=255           # 255 colors + 1 transparent = 256
-
-paletteuse:
-  alpha_threshold=128      # Pixels with source alpha < 128 become transparent
-```
-
-### Alpha Threshold Tuning
-
-- `alpha_threshold=0`: Only fully transparent pixels (alpha=0) become transparent
-- `alpha_threshold=128`: Default; pixels with <50% opacity become transparent
-- `alpha_threshold=255`: Only fully opaque pixels remain; everything else transparent
-
-For clean edges on transparent GIFs, use `alpha_threshold=128` with `dither=bayer:bayer_scale=3`.
-
-## File Size Estimation
-
-Rough formula for estimating GIF file size before creation:
-
-```
-estimated_bytes = width * height * frame_count * color_factor * dither_factor * motion_factor
-
-color_factor:
-  256 colors = 0.5
-  128 colors = 0.4
-  64 colors  = 0.3
-
-dither_factor:
-  floyd_steinberg = 1.0
-  sierra2         = 0.95
-  bayer:3         = 0.7
-  none            = 0.5
-
-motion_factor (with diff_mode=rectangle):
-  Full-frame motion  = 0.8
-  Partial motion     = 0.3
-  Mostly static      = 0.15
-```
-
-This is very approximate. Actual compression depends heavily on content complexity.
-
-## Optimization Decision Flowchart
-
-```
-Is GIF too large?
-  |
-  +--> Apply gifsicle -O3 --lossy=30 (always first)
-  |     |
-  |     +--> Still too large?
-  |           |
-  |           +--> Is dither floyd_steinberg?
-  |           |     +--> Yes: Switch to bayer:3 + stats_mode=diff
-  |           |     +--> No: Reduce colors (256->128->64)
-  |           |
-  |           +--> Still too large?
-  |           |     +--> Reduce dimensions (640->480->320)
-  |           |
-  |           +--> Still too large?
-  |           |     +--> Increase lossy (60->80->120->150)
-  |           |
-  |           +--> Still too large?
-  |           |     +--> Reduce FPS (20->15->10->8)
-  |           |
-  |           +--> Still too large?
-  |                 +--> Content is too complex for target size
-  |                 +--> Suggest: trim duration, split into parts, or use video format
-```