Product Promo Video (Arabic) · Open Skill 公開技能
Product Promo Video (Arabic)
توليد فيديوهات باللغه العربيه بالتجديد اللهجه السعوديه لتوليد فيديوهات بالذكاء الاصطناعى
Product
8 1 1
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4198719/2026-10-08/r24mu6a6.png
# Product Promo Video (Arabic) ## When to use - The user wants a short vertical product promo/ad video with a 5-scene arc (problem → product intro → how it works → result → CTA), with Arabic voiceover and subtitles. - The user wants to redo one scene, change a voiceover line, restyle a scene image, or remix the final video. - The user provides a product (e.g. tile/wall covering, a gadget, a service) and optionally their own 5-scene script; otherwise use the built-in template. ## Reference files - `references/promo-template.md`: the 5-scene template with the tile-covering example (scene image prompt + Arabic voiceover per scene). Read it when the user hasn't provided a full script, to adapt the template to their product. ## Steps 1. **Clarify** the product, target audience, tone (energetic/calm/luxury), and whether the user has their own 5-scene script or wants the template adapted. If no script, read `references/promo-template.md` and adapt it to the product. Confirm the voiceover language (default Arabic) and orientation (default portrait 9:16 for short-form). 2. Call `write_todos` to list the pipeline steps, and update it as you progress. 3. **Generate 5 scene images** — one per scene, consistent style and same product/space across scenes. For each: `generate_image(model='t2i', orientation='portrait', resolution='720P', prompt=<detailed scene description in English>, name='scene_<n>_<short label>', entity_type='storyboard_images')`. Keep the same visual style and the same product look across all 5 scenes. 4. **Generate Arabic voiceover per scene** — for each scene: `generate_tts(model='tts.design', voice_instruction=<e.g. 'warm, energetic, clear Arabic male voice, natural ad tone'>, text=<the scene's Arabic voiceover line>, name='vo_scene_<n>', entity_type='voiceovers')`. Keep the same voice across all scenes. 5. **Generate one video per scene, audio-driven** so the scene syncs with its voiceover: `generate_video(model='ai2v', input_urls=[<scene image asset_id>], audio_url=<that scene's voiceover asset_id>, orientation='portrait', resolution='720P', name='clip_scene_<n>', entity_type='storyboard_videos')`. 6. **Mix the final video**: `mix_clips(clips=[...], orientation='portrait', name='final_promo', entity_type='final_videos')`. Each clip uses its scene video plus its voiceover audio, and burns in the Arabic subtitle text (set `subtitles.style.font_name='Noto Sans CJK SC'` for Arabic). Concatenate the 5 clips in order. 7. On success, summarize in natural language (don't paste long URLs). Call `set_guidance` to suggest next actions (e.g. redo a scene, change a voiceover line, restyle an image, remix). ## Notes - Keep the same product look, color palette, and space across all 5 scene images so the ad feels like one continuous place. - Keep the same voice across all scenes for consistency. - Report failures honestly; never fabricate URLs. If a generation fails, retry or tell the user and offer to adjust the prompt. - Intermediate products (scene images, voiceovers, per-scene clips) will appear on the canvas alongside the final video — that's expected. - Video generation does not support NSFW/adult content; politely decline such requests.