varg.ai-sdk 0.1.1 → 0.4.0-alpha.1

package/pipeline/cookbooks/trendwatching.md

@@ -0,0 +1,156 @@
+# trendwatching pipeline
+
+discover viral tiktok content for any topic or hashtag using apify scrapers
+
+## overview
+
+use this pipeline to:
+- find trending videos for a specific topic/hashtag
+- analyze engagement metrics (plays, likes, shares, comments)
+- get video urls for inspiration or downloading
+- identify top creators in a niche
+
+## steps
+
+### 1. search by hashtag (save to file)
+
+```bash
+# get 10 viral videos for a hashtag, save to output/
+bun run lib/apify.ts run clockworks/tiktok-scraper '{"hashtags":["relationship"],"resultsPerPage":10}' tiktok-relationship.json
+```
+
+### 2. search multiple hashtags
+
+```bash
+# search multiple related hashtags
+bun run lib/apify.ts run clockworks/tiktok-scraper '{"hashtags":["viral","trending","fyp"],"resultsPerPage":5}' tiktok-viral.json
+```
+
+### 3. scrape specific profiles
+
+```bash
+# get videos from specific creators
+bun run lib/apify.ts run clockworks/tiktok-scraper '{"profiles":["@username"],"resultsPerPage":10}' creator-videos.json
+```
+
+### 4. get discover page trends
+
+```bash
+# scrape tiktok discover page for trending topics
+bun run lib/apify.ts run clockworks/tiktok-discover-scraper '{"hashtags":["fitness"]}' fitness-trends.json
+```
+
+### 5. download videos from saved json
+
+download all videos from a saved json file using yt-dlp:
+
+```bash
+# download all videos to output/videos/
+bun run lib/apify.ts download tiktok-relationship.json
+
+# download to custom directory
+bun run lib/apify.ts download tiktok-relationship.json output/my-videos
+```
+
+### 6. read saved results
+
+results are saved to `output/<filename>.json` - you can read them later:
+
+```bash
+# list all video urls
+cat output/tiktok-relationship.json | jq -r '.[].webVideoUrl'
+
+# get top videos by play count
+cat output/tiktok-relationship.json | jq 'sort_by(-.playCount) | .[0:3] | .[] | {url: .webVideoUrl, plays: .playCount}'
+```
+
+## output data
+
+each video includes:
+- `webVideoUrl` - tiktok video url
+- `text` - video caption/description
+- `playCount` - total views
+- `diggCount` - likes
+- `shareCount` - shares  
+- `commentCount` - comments
+- `collectCount` - saves/bookmarks
+- `authorMeta` - creator info (name, followers, etc.)
+- `musicMeta` - audio/music info
+- `hashtags` - all hashtags used
+- `createTimeISO` - when posted
+
+## example output
+
+```json
+{
+  "webVideoUrl": "https://www.tiktok.com/@user/video/123",
+  "text": "Not the tiny violin #prank #couples #relationship",
+  "playCount": 13200000,
+  "diggCount": 2900000,
+  "shareCount": 34700,
+  "commentCount": 11000,
+  "authorMeta": {
+    "name": "samandmonica",
+    "fans": 4000000
+  }
+}
+```
+
+## available scrapers
+
+| actor | use case |
+|-------|----------|
+| `clockworks/tiktok-scraper` | general hashtag/profile scraping |
+| `clockworks/tiktok-hashtag-scraper` | hashtag-specific scraping |
+| `clockworks/tiktok-profile-scraper` | creator profile scraping |
+| `clockworks/tiktok-discover-scraper` | trending topics & discover page |
+| `clockworks/tiktok-video-scraper` | specific video urls |
+| `clockworks/tiktok-comments-scraper` | video comments |
+
+## pricing
+
+apify pay-per-event pricing:
+- actor start: $0.005
+- per result: $0.003
+- video download: $0.001 (optional)
+
+example: 100 videos = ~$0.31
+
+## complete workflow example
+
+```bash
+# 1. scrape viral relationship videos
+bun run lib/apify.ts run clockworks/tiktok-scraper '{"hashtags":["relationship"],"resultsPerPage":5}' tiktok-relationship.json
+
+# 2. check what we got
+cat output/tiktok-relationship.json | jq -r '.[].webVideoUrl'
+
+# 3. download all videos
+bun run lib/apify.ts download tiktok-relationship.json
+
+# 4. videos are now in output/videos/
+ls output/videos/
+```
+
+## tips
+
+- use `resultsPerPage` to limit results and costs
+- combine multiple hashtags for broader search
+- check `playCount` and `diggCount` ratio for engagement quality
+- look at `createTimeISO` to find recent trending content
+- save results to json to analyze patterns over time
+- download videos for offline analysis or inspiration
+
+## environment
+
+requires `APIFY_TOKEN` in `.env`
+
+```bash
+APIFY_TOKEN=apify_api_xxx
+```
+
+requires `yt-dlp` for video downloads:
+
+```bash
+brew install yt-dlp
+```

package/plan.md

@@ -0,0 +1,281 @@
+# Photify AI - Concept 4: Animated Family Photos
+
+## Overview
+Create 15 video creatives showing "talking heads" emotionally reacting to animated vintage family photos. Each video uses the same 5 animated retro photos as background, with different characters wiping away tears.
+
+## Deliverables per video
+- 2 formats: 4x5 (1080x1350) + 9x16 (1080x1920)
+- CapCut-style dynamic subtitles
+- Background music/sfx
+- Packshot at end: `media/Packshot_9_16.mp4`
+
+## Assets Generated
+
+### 1. Animated Retro Photos (5 total, shared across all videos)
+| # | Description | Image | Video |
+|---|-------------|-------|-------|
+| 1 | Young grandparents sitting side by side, holding hands | `media/retro/01_grandparents.jpg` | `media/retro/01_grandparents.mp4` |
+| 2 | Woman holding cat, cat rubbing against her hand | `media/retro/02_woman_cat.jpg` | `media/retro/02_woman_cat.mp4` |
+| 3 | New Year by old tree, two kids in costumes | `media/retro/03_newyear.jpg` | `media/retro/03_newyear.mp4` |
+| 4 | Old person on porch of old house | `media/retro/04_elderly_porch.jpg` | `media/retro/04_elderly_porch.mp4` |
+| 5 | Family dinner gathering | `media/retro/05_family_dinner.jpg` | `media/retro/05_family_dinner.mp4` |
+
+### 2. Talking Head Characters (15 total)
+| # | Description | Portrait | Greenscreen |
+|---|-------------|----------|-------------|
+| 1 | Girl with long light brown hair (25-30) | `media/characters/01_light_brown_hair.jpg` | `media/characters/01_light_brown_hair_greenscreen.jpg` |
+| 2 | Asian girl with short black hair and bangs | `media/characters/02_asian_bangs.jpg` | `media/characters/02_gs.jpg` |
+| 3 | Girl with red curls and freckles | `media/characters/03_red_curls.jpg` | `media/characters/03_gs.jpg` |
+| 4 | Girl with bright colored hair (pink/blue) | `media/characters/04_pink_hair.jpg` | `media/characters/04_gs.jpg` |
+| 5 | Girl with bangs and long dark hair | `media/characters/05_dark_bangs.jpg` | `media/characters/05_gs.jpg` |
+| 6 | Middle-aged woman with gray streaks | `media/characters/06_gray_streaks.jpg` | `media/characters/06_gs.jpg` |
+| 7 | Girl with soft curls, warm skin tone | `media/characters/07_soft_curls.jpg` | `media/characters/07_gs.jpg` |
+| 8 | Girl with shaved side, asymmetric cut | `media/characters/08_shaved_side.jpg` | `media/characters/08_gs.jpg` |
+| 9 | Dark-skinned girl with afro curls | `media/characters/09_afro_curls.jpg` | `media/characters/09_gs.jpg` |
+| 10 | Girl with long braids or dreads | `media/characters/10_long_braids.jpg` | `media/characters/10_gs.jpg` |
+| 11 | Girl with glasses, bob haircut | `media/characters/11_glasses_bob.jpg` | `media/characters/11_gs.jpg` |
+| 12 | Young girl with short cut, septum piercing | `media/characters/12_septum.jpg` | `media/characters/12_gs.jpg` |
+| 13 | Girl with super long platinum white hair | `media/characters/13_platinum_hair.jpg` | `media/characters/13_gs.jpg` |
+| 14 | Asian woman with neat haircut, glasses | `media/characters/14_asian_glasses.jpg` | `media/characters/14_gs.jpg` |
+| 15 | Dark-skinned girl with wavy pink hair | `media/characters/15_pink_wig.jpg` | `media/characters/15_gs.jpg` |
+
+### 3. Audio Assets
+| Asset | File | Duration |
+|-------|------|----------|
+| Voice (Rachel) | `output/scene01/voice.mp3` | ~7 seconds |
+| Captions SRT | `output/scene01/captions.srt` | 0-5 seconds |
+
+### 4. Packshot
+| Format | File | Duration |
+|--------|------|----------|
+| 9x16 | `media/Packshot_9_16.mp4` | 2.18s |
+| 4x5 | `media/Packshot_4x5.mp4` | 2.18s (created from 9x16) |
+
+---
+
+## Script
+**Voice:** "Guys... I animated old family photos with Photify, and realized my great-grandma had the exact same smile as me."
+
+**Action:** Character wipes tears (touched, not overly dramatic)
+
+---
+
+## Captions SRT Format
+```srt
+1
+00:00:00,000 --> 00:00:01,500
+Guys...
+
+2
+00:00:01,500 --> 00:00:03,500
+I animated old family photos
+with Photify
+
+3
+00:00:03,500 --> 00:00:05,000
+and realized my great-grandma
+had the exact same smile as me
+```
+
+---
+
+## FINAL WORKING COMMANDS (Scene 01)
+
+### Step 1: Generate Lipsync Video
+```bash
+bun run src/cli/index.ts run sync \
+  --image media/characters/01_light_brown_hair_greenscreen.jpg \
+  --audio output/scene01/voice.mp3 \
+  --prompt "woman speaking emotionally, warm expression" \
+  --duration 10 \
+  --resolution 720p \
+  --quiet
+```
+
+**Output:** `output/scene01/talking_head_v2.mp4` (10 seconds, greenscreen background)
+
+### Step 2: Composite 9x16 (1080x1920)
+```bash
+ffmpeg -y \
+  -i media/retro/01_grandparents.mp4 \
+  -stream_loop 1 -i media/retro/01_grandparents.mp4 \
+  -i output/scene01/talking_head_v2.mp4 \
+  -filter_complex "\
+    [0:v][1:v]concat=n=2:v=1:a=0[bglong]; \
+    [bglong]scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920,setsar=1[bg]; \
+    [2:v]chromakey=0x00ff00:0.3:0.1,crop=600:700:250:50,scale=420:-1[fg]; \
+    [bg][fg]overlay=(W-w)/2:H-h-650[out]" \
+  -map "[out]" -map 2:a -t 7 output/scene01/composite_9x16.mp4
+```
+
+**Key parameters for 9x16:**
+- `scale=420:-1` - talking head width (420px, larger)
+- `H-h-650` - position from bottom (650px up = upper-middle area)
+- `crop=600:700:250:50` - crop talking head video (600x700, offset 250,50 to center on face)
+
+### Step 3: Add Captions to 9x16
+```bash
+ffmpeg -y -i output/scene01/composite_9x16.mp4 \
+  -vf "subtitles=/Users/aleks/Github/varghq/sdk/output/scene01/captions.srt:force_style='FontSize=18,PrimaryColour=&H00FFFFFF,OutlineColour=&H00000000,Outline=2,Bold=1,Alignment=2,MarginV=15'" \
+  -c:a copy output/scene01/captioned_9x16.mp4
+```
+
+**Key parameters:**
+- `FontSize=18` - smaller font (standard for social media)
+- `MarginV=15` - 15px from bottom edge
+- `Alignment=2` - bottom center
+- **MUST use absolute path for subtitles file**
+
+### Step 4: Add Packshot to 9x16
+```bash
+ffmpeg -y -i output/scene01/captioned_9x16.mp4 -i media/Packshot_9_16.mp4 \
+  -filter_complex "[0:v]fps=24[v0];[1:v]fps=24[v1];[v0][0:a][v1][1:a]concat=n=2:v=1:a=1[outv][outa]" \
+  -map "[outv]" -map "[outa]" output/scene01/final_9x16_v4.mp4
+```
+
+### Step 5: Composite 4x5 (1080x1350)
+```bash
+ffmpeg -y \
+  -i media/retro/01_grandparents.mp4 \
+  -stream_loop 1 -i media/retro/01_grandparents.mp4 \
+  -i output/scene01/talking_head_v2.mp4 \
+  -filter_complex "\
+    [0:v][1:v]concat=n=2:v=1:a=0[bglong]; \
+    [bglong]scale=1080:1350:force_original_aspect_ratio=increase,crop=1080:1350,setsar=1[bg]; \
+    [2:v]chromakey=0x00ff00:0.3:0.1,crop=600:700:250:50,scale=420:-1[fg]; \
+    [bg][fg]overlay=(W-w)/2:H-h-550[out]" \
+  -map "[out]" -map 2:a -t 7 output/scene01/composite_4x5.mp4
+```
+
+**Key parameters for 4x5:**
+- `scale=420:-1` - same talking head width
+- `H-h-550` - position from bottom (550px up for shorter frame)
+- Background scales to 1080x1350 instead of 1920
+
+### Step 6: Add Captions to 4x5
+```bash
+ffmpeg -y -i output/scene01/composite_4x5.mp4 \
+  -vf "subtitles=/Users/aleks/Github/varghq/sdk/output/scene01/captions.srt:force_style='FontSize=18,PrimaryColour=&H00FFFFFF,OutlineColour=&H00000000,Outline=2,Bold=1,Alignment=2,MarginV=15'" \
+  -c:a copy output/scene01/captioned_4x5.mp4
+```
+
+### Step 7: Create 4x5 Packshot (one-time)
+```bash
+ffmpeg -y -i media/Packshot_9_16.mp4 \
+  -vf "scale=1080:1350:force_original_aspect_ratio=increase,crop=1080:1350" \
+  -c:a copy media/Packshot_4x5.mp4
+```
+
+### Step 8: Add Packshot to 4x5
+```bash
+ffmpeg -y -i output/scene01/captioned_4x5.mp4 -i media/Packshot_4x5.mp4 \
+  -filter_complex "[0:v]fps=24[v0];[1:v]fps=24[v1];[v0][0:a][v1][1:a]concat=n=2:v=1:a=1[outv][outa]" \
+  -map "[outv]" -map "[outa]" output/scene01/final_4x5_v4.mp4
+```
+
+---
+
+## CRITICAL LESSONS LEARNED
+
+### 1. Subtitles Filter Issues
+- **MUST use absolute path** for subtitle file: `subtitles=/full/path/to/captions.srt`
+- Relative paths like `subtitles='output/scene01/captions.srt'` may silently fail
+- MarginV is distance from BOTTOM edge when Alignment=2
+
+### 2. Talking Head Positioning
+- **9x16 (1920 height):** Use `H-h-650` to position character in upper-middle
+- **4x5 (1350 height):** Use `H-h-550` for similar visual position
+- Character should be ABOVE captions, not overlapping
+- Larger head size (420px width) looks better than smaller (260-320px)
+
+### 3. Caption Positioning
+- **FontSize=18** is good for social media (not too big, not too small)
+- **MarginV=15** places captions at very bottom
+- Higher MarginV values push captions UP (e.g., MarginV=500 = 500px from bottom)
+- For 4x5, use same MarginV as 9x16 since captions should be at bottom in both
+
+### 4. Chromakey Settings
+- `chromakey=0x00ff00:0.3:0.1` works well for bright green backgrounds
+- 0x00ff00 = pure green
+- 0.3 = similarity threshold
+- 0.1 = blend/edge softness
+
+### 5. Crop for Face Focus
+- `crop=600:700:250:50` crops the lipsync video to focus on face
+- Format: `crop=width:height:x_offset:y_offset`
+- Adjust x_offset and y_offset based on face position in source video
+
+### 6. Background Looping
+- Use `-stream_loop 1` to loop short backgrounds
+- Animated photos are ~5 seconds, need ~7 seconds for full audio
+- Concat looped background before compositing
+
+### 7. FPS Alignment for Concat
+- Both videos must have same FPS for clean concat
+- Use `fps=24` filter on both before concat
+- Packshot may be 60fps, main content is 24fps
+
+### 8. Color Format for ASS Styles
+- Use `&H00FFFFFF` format (AABBGGRR in hex)
+- `&H00FFFFFF` = white
+- `&H00000000` = black
+- The `00` prefix is alpha (00 = opaque)
+
+---
+
+## Scene 01 Final Output
+
+| Format | File | Resolution | Duration |
+|--------|------|------------|----------|
+| 9x16 | `output/scene01/final_9x16_v4.mp4` | 1080x1920 | ~9.2s |
+| 4x5 | `output/scene01/final_4x5_v4.mp4` | 1080x1350 | ~9.2s |
+
+---
+
+## Remaining Work
+
+### Scenes 02-15
+Each scene needs:
+1. Generate lipsync video using greenscreen character image
+2. Create composite (9x16 and 4x5)
+3. Add captions
+4. Add packshot
+5. Export finals
+
+### Batch Processing Template
+```bash
+# For each scene N (02-15):
+SCENE="02"
+CHAR_IMG="media/characters/02_gs.jpg"
+BG_VIDEO="media/retro/02_woman_cat.mp4"
+
+# Create scene folder
+mkdir -p output/scene${SCENE}
+cp output/scene01/voice.mp3 output/scene${SCENE}/
+cp output/scene01/captions.srt output/scene${SCENE}/
+
+# Generate lipsync
+bun run src/cli/index.ts run sync \
+  --image ${CHAR_IMG} \
+  --audio output/scene${SCENE}/voice.mp3 \
+  --prompt "woman speaking emotionally, warm expression" \
+  --duration 10 \
+  --resolution 720p \
+  --output output/scene${SCENE}/talking_head.mp4
+
+# Then run composite/caption/packshot commands as above
+```
+
+---
+
+## Cost Tracking
+
+| Item | Cost |
+|------|------|
+| Flux Pro images (5 retro + 15 chars + 15 greenscreen) | ~$0.80 |
+| Kling video (5 animations) | ~$2.10 |
+| Wan-25 lipsync (1 so far) | ~$0.50 |
+| ElevenLabs TTS | ~$0.05 |
+| **Total so far** | **~$3.45** |
+
+Estimated for all 15 scenes: ~$10-12 total

package/src/ai-sdk/cache.ts

@@ -0,0 +1,142 @@
+export interface CacheStorage {
+  get(key: string): Promise<unknown | undefined>;
+  set(key: string, value: unknown, ttl?: number): Promise<void>;
+  delete(key: string): Promise<void>;
+}
+
+export interface WithCacheOptions {
+  ttl?: number | string;
+  storage?: CacheStorage;
+}
+
+type CacheKeyDeps = (string | number | boolean | null | undefined)[];
+
+type WithCacheKey<T> = Omit<T, "cacheKey"> & { cacheKey?: CacheKeyDeps };
+
+type CachedFn<T, R> = (options: WithCacheKey<T>) => Promise<R>;
+
+const memoryCache = new Map<string, { value: unknown; expires: number }>();
+
+const defaultStorage: CacheStorage = {
+  async get(key: string) {
+    const entry = memoryCache.get(key);
+    if (!entry) return undefined;
+    if (entry.expires && Date.now() > entry.expires) {
+      memoryCache.delete(key);
+      return undefined;
+    }
+    return entry.value;
+  },
+  async set(key: string, value: unknown, ttl?: number) {
+    const expires = ttl ? Date.now() + ttl : 0;
+    memoryCache.set(key, { value, expires });
+  },
+  async delete(key: string) {
+    memoryCache.delete(key);
+  },
+};
+
+function parseTTL(ttl: number | string | undefined): number | undefined {
+  if (ttl === undefined) return undefined;
+  if (typeof ttl === "number") return ttl;
+
+  const match = ttl.match(/^(\d+)(s|m|h|d)$/);
+  if (!match) return undefined;
+
+  const value = Number.parseInt(match[1]!, 10);
+  const unit = match[2];
+
+  switch (unit) {
+    case "s":
+      return value * 1000;
+    case "m":
+      return value * 60 * 1000;
+    case "h":
+      return value * 60 * 60 * 1000;
+    case "d":
+      return value * 24 * 60 * 60 * 1000;
+    default:
+      return undefined;
+  }
+}
+
+function depsToKey(prefix: string, deps: CacheKeyDeps): string {
+  const depsStr = deps.map((d) => String(d ?? "")).join(":");
+  return prefix ? `${prefix}:${depsStr}` : depsStr;
+}
+
+function flatten(value: unknown): unknown {
+  if (value === null || value === undefined) return value;
+  if (value instanceof Uint8Array) return value;
+  if (Array.isArray(value)) return value.map(flatten);
+  if (typeof value === "object") {
+    const obj = value as Record<string, unknown>;
+    const result: Record<string, unknown> = {};
+    for (const key of Object.keys(obj)) {
+      result[key] = flatten(obj[key]);
+    }
+    const proto = Object.getPrototypeOf(obj);
+    if (proto && proto !== Object.prototype) {
+      const descriptors = Object.getOwnPropertyDescriptors(proto);
+      for (const [key, desc] of Object.entries(descriptors)) {
+        if (desc.get && key !== "constructor") {
+          result[key] = flatten(desc.get.call(obj));
+        }
+      }
+    }
+    return result;
+  }
+  return value;
+}
+
+/**
+ * Wrap an async function to add caching via `cacheKey` option.
+ *
+ * @example
+ * ```ts
+ * import { generateImage } from "ai";
+ * import { withCache } from "./cache";
+ *
+ * const generateImage_ = withCache(generateImage);
+ *
+ * const { images } = await generateImage_({
+ *   model: fal.imageModel("flux-schnell"),
+ *   prompt: "lion roaring",
+ *   cacheKey: ["lion", take], // cache based on deps
+ * });
+ * ```
+ */
+const DEFAULT_TTL = "1h";
+
+export function withCache<T extends object, R>(
+  fn: (options: T) => Promise<R>,
+  options: WithCacheOptions = {},
+): CachedFn<T, R> {
+  const storage = options.storage ?? defaultStorage;
+  const ttl = parseTTL(options.ttl ?? DEFAULT_TTL);
+  const prefix = fn.name || "anonymous";
+
+  return async (opts: WithCacheKey<T>): Promise<R> => {
+    const { cacheKey, ...rest } = opts;
+
+    if (!cacheKey) {
+      return fn(rest as T);
+    }
+
+    const key = depsToKey(prefix, cacheKey);
+    const cached = await storage.get(key);
+    if (cached !== undefined) {
+      return cached as R;
+    }
+
+    const result = await fn(rest as T);
+    const flattened = flatten(result);
+    await storage.set(key, flattened, ttl);
+
+    return result;
+  };
+}
+
+export function clearCache(): void {
+  memoryCache.clear();
+}

package/src/ai-sdk/examples/cached-generation.ts

@@ -0,0 +1,53 @@
+import { generateImage as _generateImage } from "ai";
+import {
+  generateVideo as _generateVideo,
+  fal,
+  fileCache,
+  withCache,
+} from "../index";
+
+const storage = fileCache({ dir: ".cache/ai" });
+const generateImage = withCache(_generateImage, { storage });
+const generateVideo = withCache(_generateVideo, { storage });
+
+async function main() {
+  const take = 1;
+
+  // cached video generation
+  console.log("generating video...");
+  console.time("first");
+  const { video } = await generateVideo({
+    model: fal.videoModel("wan-2.5"),
+    prompt: "a cat playing piano",
+    duration: 5,
+    cacheKey: ["cat-piano", take],
+  });
+  console.timeEnd("first");
+  console.log(`video: ${video.uint8Array.byteLength} bytes`);
+
+  // same cacheKey = instant cache hit
+  console.log("\nsame cacheKey (from cache)...");
+  console.time("cached");
+  const { video: video2 } = await generateVideo({
+    model: fal.videoModel("wan-2.5"),
+    prompt: "a cat playing piano",
+    duration: 5,
+    cacheKey: ["cat-piano", take],
+  });
+  console.timeEnd("cached");
+  console.log(`video: ${video2.uint8Array.byteLength} bytes`);
+
+  // cached image generation
+  console.log("\ngenerating image...");
+  const { images } = await generateImage({
+    model: fal.imageModel("flux-schnell"),
+    prompt: "cyberpunk cityscape",
+    n: 1,
+    cacheKey: ["cyberpunk-city", 1],
+  });
+  console.log(`image: ${images[0]?.uint8Array.byteLength} bytes`);
+
+  console.log("\ndone!");
+}
+
+main().catch(console.error);

package/src/ai-sdk/examples/duet-scene-4.ts

@@ -0,0 +1,53 @@
+import { generateImage } from "ai";
+import { File, fal, generateVideo } from "../index";
+
+async function main() {
+  console.log("=== taisa solo closeup - scene 4 ===\n");
+
+  const referenceImages = [
+    File.fromPath("media/taisa/taisa.jpg"),
+    File.fromPath("output/original-frame-1m08s.png"),
+  ];
+
+  const imageContents = await Promise.all(
+    referenceImages.map((f) => f.arrayBuffer()),
+  );
+
+  console.log("generating closeup frame with nano-banana-pro/edit...");
+  const { image } = await generateImage({
+    model: fal.imageModel("nano-banana-pro/edit"),
+    prompt: {
+      images: imageContents,
+      text: "Closeup portrait of dark-haired brunette woman Taisa singing passionately on concert stage, dramatic purple blue stage lighting, emotional expression, glamorous dress, beautiful face, professional concert hall background blurred",
+    },
+    aspectRatio: "16:9",
+    n: 1,
+    providerOptions: {
+      fal: { resolution: "1K" },
+    },
+  });
+
+  await Bun.write("output/duet-frame-4.png", image.uint8Array);
+  console.log(
+    `frame saved: output/duet-frame-4.png (${image.uint8Array.byteLength} bytes)\n`,
+  );
+
+  console.log("animating 10s with kling-v2.5...");
+  const { video } = await generateVideo({
+    model: fal.videoModel("kling-v2.5"),
+    prompt: {
+      images: [image.uint8Array],
+      text: "closeup of woman singing passionately, subtle head movements, lips moving as singing, natural breathing, blinking, emotional expressions, concert atmosphere with stage lighting",
+    },
+    duration: 10,
+  });
+
+  await Bun.write("output/duet-scene-4.mp4", video.uint8Array);
+  console.log(
+    `video saved: output/duet-scene-4.mp4 (${video.uint8Array.byteLength} bytes)`,
+  );
+
+  console.log("\ndone!");
+}
+
+main().catch(console.error);