tibetan-cinematic-video

Tibetan Cinematic Video (google/veo)

Purpose

Image → google/veo video prompt → ~/.openclaw/workspace/tibetanProc/yymmddHHMM_{title}.mp4

VALID INPUTS ONLY

1. IMAGE (MANDATORY)
2. EXACTLY: "word1,word2,word3" (Chinese characters)

Invalid → reject gracefully

EXECUTION FLOW

1. VERIFY: Ensure dir ~/.openclaw/workspace/tibetanProc/ exists; CREATE ONLY IF ABSENT. REJECT/ERROR if any file/path touches ~/.openclaw/workspace/ parent. GENERATE → yymmddHHMM_{title}.mp4
2. ANALYZE image → extract TITLE (thangka|monastery|mandala|stupa)
3. MAP 3 words → Clip1(historical), Clip2(human), Clip3(modern)
4. BUILD prompt for Veo (3 clips + required motions, camera, lighting, mood, 9:16)
5. CALL Veo:
   • Use Veo‑3.1‑generate‑preview (or your internal Veo endpoint).
   • Pass:
     • prompt (plain text, UTF‑8, Chinese + English allowed).
     • image (input image bytes/Base64 as first frame).
     • aspect_ratio: "9:16" (portrait) if configurable.
   • Poll the operation until done, with a reasonable timeout (e.g., 120 seconds).
6. DOWNLOAD the generated video from uri → save as ~/.openclaw/workspace/tibetanProc/yymmddHHMM_{title}.mp4

MANDATORY PROMPT ELEMENTS

• MOTIONS: slow zoom‑in, camera pan left/right, dolly zoom, clouds moving.
• CAMERA: Drone shot, panning shot, dolly zoom; emphasize smooth transitions between scenes.
• LIGHT: Soft cinematic lighting, natural Himalayan ambient light.
• MOOD: Ethereal, contemplative, reverent.
• RATIO: 9:16 vertical (portrait).
• MODEL: Google Veo (Veo‑3.1‑generate‑preview) ← for telemetry only, not as style.

NOTE: Veo behavior

• Veo‑3.1‑generate‑preview produces ~8‑second clips by default, up to 1080p or 4K depending on configuration.
• Aspect ratio can be forced to 9:16 in the API call if supported.
• Longer videos are possible by chaining multiple 8‑second segments (e.g., via Veo‑extension or Flow‑style tooling), but this skill currently outputs a single 8‑second clip per call.

Google Veo Prompt Format

Generate a continuous 9:16 vertical cinematic video set in Tibet:

1. Historical scene (from WORD1): Drone shot slowly zooming in on an ancient Tibetan scene: thangka, carved mani script, prayer wheels, weathered monastery walls.
2. Cultural scene (from WORD2): Camera pans left across a cultural Tibetan scene: maroon‑robed monks, butter lamps, temple murals, chanting in the background.
3. Modern scene (from WORD3): Panning right with a dolly zoom effect on a modern Tibetan scene: pilgrims spinning prayer wheels, electric lights, clouds moving over Himalayan peaks.

Style: Soft cinematic lighting, ethereal and reverent mood, high‑speed slow motion illusion, smooth transitions between scenes. Maintain Tibetan cultural authenticity; avoid Western fantasy and anime stylization.

FILE SPEC (NEVER DEVIATE)

GLOBAL RULE: Do not create ANYTHING under ~/.openclaw/workspace/ parent dir. ALL artifacts (videos, temps) MUST be in ~/.openclaw/workspace/tibetanProc/. Deviate → abort. DIRECTORY: ~/.openclaw/workspace/tibetanProc/ BASENAME: yymmddHHMM_{title}.mp4 EXAMPLES:

• 2604081747_thangka.mp4
• 2604081752_monastery.mp4
• 2604081801_mandala.mp4

TIMESTAMP: UTC, 10-digit (yymmddHHMM) TITLE: Single word from image dominant subject

CULTURAL GUARDRAILS

✅ Monasteries, prayer wheels, thangka, mani stones, butter lamps, maroon robes, Himalayan peaks. ❌ Western fantasy, anime stylization, overly stylized “golden hour” clichés.

EXECUTION EXAMPLE

INPUT IMAGE: Thangka scroll
INPUT WORDS: "歷史,人文,當今"
OUTPUT FILE: ~/.openclaw/workspace/tibetanProc/2604081747_thangka.mp4

• Historical scene (from "歷史"): Drone shot slowly zooming in on an ancient thangka scroll, carved mani script, prayer wheels, and weathered monastery walls.
• Cultural scene (from "人文"): Camera pans left across maroon‑robed monks, flickering butter lamps, and temple mural backgrounds.
• Modern scene (from "當今"): Panning right with a dolly zoom on modern pilgrims spinning prayer wheels; clouds move over Himalayan peaks.

Style: 9:16 vertical, soft cinematic lighting, ethereal mood, high‑speed slow motion illusion, smooth transitions. Maintain Tibetan cultural authenticity.

ERROR HANDLING

NO IMAGE → "Please upload Tibetan image first" <3 WORDS → "Need exactly 3 Chinese theme words (e.g. 歷史,人文,當今)" WRONG FORMAT → "Use: word1,word2,word3 format" WRONG PATH → "ERROR: No creation of ALL artifacts allowed under ~/.openclaw/workspace/. Use ONLY ~/.openclaw/workspace/tibetanProc/yymmddHHMM_{title}.*" for all artifacts, not just video