Anime Scene Video · Open Skill 公開スキル
Anime Scene Video
Anime Scene Video は Digen 上の公開 Open Skill です。開くと、このワークフローが何を作るか分かります。
Anime
60 5 10
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4080505/2026-10-05/j7vvpc47.jpg
# Anime Scene Video
Turn a visual script (an image + text) into an anime-style 2D/3D animation scene video. Defaults: English prompts, 9:16 portrait, ~10-15s output, anime 2D/3D style.
## When to use
- The user wants to create an anime animation scene from an image plus text (scene description, narration, or dialogue).
- The user wants to restyle an existing image into anime and animate it into a short clip.
- The user wants a short animated scene with voiceover narration and/or background music.
## Steps
1. **Gather inputs.** The visual script = an image (optional) + text (scene description / narration / dialogue). If the user uploaded an image, call `recognize_image(image_url=<asset URL>, prompt="Describe the scene, characters, composition, and mood in detail")` to understand it. If the user uploaded an audio narration instead of text, call `transcribe_audio(audio_url=<asset URL>)` to get the script text. Use `write_todos` to track the multi-step pipeline.
2. **Generate the anime scene image (9:16 portrait).**
- No reference image: `generate_image(model='t2i', prompt=<scene description in English, anime 2D/3D style, detailed>, orientation='portrait', resolution='720P', name='Scene Image', entity_type='storyboard_images')`.
- Reference image provided: `generate_image(model='i2i', input_urls=[<image asset_id>], prompt=<restyle into anime 2D/3D, keep composition and characters>, orientation='portrait', resolution='720P', name='Scene Image', entity_type='storyboard_images')`.
3. **Generate voiceover narration** from the script text (English): `generate_tts(model='tts.design', text=<narration>, voice_instruction=<calm, expressive anime narrator voice>, name='Voiceover', entity_type='voiceovers')`. If the user provided a reference voice, use `model='tts.clone'` with `reference_audio_url=<voice asset>` instead.
4. **Generate background music**: `generate_music(prompt=<anime background music matching the scene mood, e.g. emotional/upbeat/mysterious>, duration=<10-15>, name='Music', entity_type='music')`.
5. **Generate the animation video** from the scene image: `generate_video(model='i2v', input_urls=[<scene image asset_id>], prompt=<describe the animation motion in English, e.g. hair and clothes swaying, camera slowly pushing in>, orientation='portrait', duration=<10-15>, name='Animation Scene', entity_type='storyboard_videos')`. If the user wants the motion driven by the voiceover, use `model='ai2v'` with `input_urls=[<scene image>]` and `audio_url=<voiceover asset_id>` instead.
6. **Mix the final clip**: `mix_clips(clips=[{video:{url:<video asset_id>}, audios:[{url:<voiceover asset_id>, delay:0}, {url:<music asset_id>, delay:0, volume:0.5}]}], orientation='portrait', name='Final Animation Scene', entity_type='final_videos')`.
7. **Reply** with a natural-language summary of the finished animation scene (image, voiceover, music, final clip). Call `set_guidance` to suggest next actions (e.g. restyle the character, generate the next scene, adjust the music).
## Notes
- All prompts and narration are in English.
- Default output is 9:16 portrait, ~10-15 seconds, anime 2D/3D style.
- Video generation does not support NSFW/adult content on any channel — politely decline such requests.
- Report failures honestly; never fabricate URLs or asset IDs.
- Intermediate products (scene image, voiceover, music, animation clip) will be visible on the canvas as separate cards before the final mix.