途中まで良い感じで生成できたんですが、踊らせるところがうまく行かず、少しイメージしていたのと違う感じになりました。ダンスまたは踊りをさせるならもっと具体的に指定しないとダメなようです。動きのあるものはAI任せではうまく行かず、プロンプトで具体的に指示する方が良いみたいです。
以下、基にした画像です。

この画像をもとに以下のプロンプトでGrok Imagine動画生成をしました。
Two 19-year-old Japanese girls standing side by side in a luxurious dark-wood maid cafe, exactly matching the reference image.
Left: exact likeness of genius actress Hirose Suzumi (広瀬すず美), 19yo, delicate beautiful face identical to Suzu Hirose, shoulder-length wavy dark brown hair with bangs, wearing the exact black maid dress with white lace frills and large black bow. She gently bows while speaking elegantly: 「いらっしゃいませご主人様、天才女優の広瀬すず美でございます」
Right: exact likeness of world top idol Hashimoto Kana (橋本かな), 19yo, bright cheerful beautiful face identical to Kanna Hashimoto, long straight black hair with bangs, wearing the exact navy blue maid dress with white frilly apron and large navy bow. She bounces cutely, tilts her head, winks and makes a small heart gesture while saying playfully: 「お帰りなさい!ご主人様。世界のトップアイドル橋本かなだよん。よろしくニャン♪」
Natural subtle movements only, gentle bowing, cute bouncing, head tilt, wink, small heart gesture, soft breathing, slight hair sway, very smooth and coherent motion, exactly 8 to 10 seconds duration, seamless loop capable, photorealistic, ultra-detailed 8K, perfect face and clothing consistency with reference image, cinematic lighting, 24fps.
Negative prompt: low quality, blurry, deformed, bad anatomy, extra limbs, fused fingers, text, watermark, cartoon, anime, plastic skin, overexposed, different outfit, age change, body distortion, jittery motion, fast motion, too long duration, short clip
2本目のプロンプト
Two 19-year-old Japanese girls standing side by side in a luxurious dark-wood maid cafe, exactly matching the reference image composition, lighting, and camera angle.
Left: exact likeness of genius actress Hirose Suzumi (広瀬すず美), 19yo, delicate beautiful face identical to Suzu Hirose, shoulder-length wavy dark brown hair with bangs, wearing the exact black maid dress with white lace frills and large black bow from reference. She first says with a gentle smile: 「練習は、これでバッチリだねかなちゃん」 then after hearing Kana’s reply, she makes a classic “yare yare” exasperated and slightly dumbfounded expression, sighing softly while looking at Kana.
Right: exact likeness of world top idol Hashimoto Kana (橋本かな), 19yo, bright cheerful beautiful face identical to Kanna Hashimoto, long straight black hair with bangs, wearing the exact navy blue maid dress with white frilly apron and large navy bow from reference. She responds brightly, energetically and playfully: 「バツイチ、だよすず美ちゃん。まだ結婚もしていないのに~」 while tilting her head cutely, smiling widely and giggling.
Natural conversation flow: Suzumi speaks first, Kana replies energetically, then Suzumi reacts with yare-yare exasperated face, subtle head movements, soft breathing, gentle hair sway, very smooth and coherent motion, exactly 8 to 10 seconds duration, seamless loop capable, photorealistic, ultra-detailed 8K, perfect face and clothing consistency with reference image, cinematic lighting, 24fps.
Negative prompt: low quality, blurry, deformed, bad anatomy, extra limbs, fused fingers, text, watermark, cartoon, anime, plastic skin, overexposed, different outfit, age change, body distortion, jittery motion, fast motion, too long duration, short clip
3本目用プロンプト
Two 19-year-old Japanese girls standing side by side in a luxurious dark-wood maid cafe, exactly matching the reference image composition, lighting, and camera angle.
Left: exact likeness of genius actress Hirose Suzumi (広瀬すず美), 19yo, delicate beautiful face identical to Suzu Hirose, shoulder-length wavy dark brown hair with bangs, wearing the exact black maid dress with white lace frills and large black bow from reference.
Right: exact likeness of world top idol Hashimoto Kana (橋本かな), 19yo, bright cheerful beautiful face identical to Kanna Hashimoto, long straight black hair with bangs, wearing the exact navy blue maid dress with white frilly apron and large navy bow from reference.
Natural conversation flow with smooth movements:
- Suzumi speaks cheerfully: 「かなちゃん、暇だからダンスの練習しようよ」
- Kana replies energetically and cutely: 「そうだねすず美ちゃん、練習たいせつだね。」
- Suzumi asks curiously: 「どんなダンスが良い?」
- Kana responds brightly while thinking and raising her finger: 「ちょっと待ってAIに聞いてみる」
- Then Kana continues playfully: 「AIさん、二人でできるかわいいダンス教えて~!」 while smiling at the viewer.
Gentle natural movements: head turns, cute gestures, soft breathing, slight hair sway, natural conversation rhythm, very smooth and coherent motion, exactly 8 to 10 seconds duration, seamless loop capable if possible, photorealistic, ultra-detailed 8K, perfect face and clothing consistency with reference image, cinematic lighting, 24fps.
Negative prompt: low quality, blurry, deformed, bad anatomy, extra limbs, fused fingers, text, watermark, cartoon, anime, plastic skin, overexposed, different outfit, age change, body distortion, jittery motion, fast motion, too long duration, short clip
3本目10秒延長用プロンプト
Two 19-year-old Japanese girls standing side by side in a luxurious dark-wood maid cafe, exactly matching the reference image composition, lighting, and camera angle.
Left: exact likeness of genius actress Hirose Suzumi (広瀬すず美), 19yo, delicate beautiful face identical to Suzu Hirose, shoulder-length wavy dark brown hair with bangs, wearing the exact black maid dress with white lace frills and large black bow.
Right: exact likeness of world top idol Hashimoto Kana (橋本かな), 19yo, bright cheerful beautiful face identical to Kanna Hashimoto, long straight black hair with bangs, wearing the exact navy blue maid dress with white frilly apron and large navy bow.
The two girls are energetically dancing a big, lively Japanese Bon Odori from the start with large exaggerated arm circles, wide stepping movements, dynamic body sways, and festival energy while wearing maid outfits. Traditional Japanese Bon Odori festival music is playing loudly and clearly in the background.
Conversation while dancing:
- Suzumi dances with a slightly annoyed and exasperated expression: 「これって、盆踊りじゃねえか!」
- Kana continues dancing playfully, does a cute tehepero pose with tongue out and finger on cheek: 「いいじゃん、盆踊りでも、てへぺろ~♪」
Comedic and joyful atmosphere, big dynamic Bon Odori movements, strong energetic dancing, clear facial expressions, soft hair sway, very smooth and coherent motion, exactly 8 to 10 seconds duration, photorealistic, ultra-detailed 8K, perfect face and clothing consistency with reference image, cinematic lighting, 24fps.
Negative prompt: low quality, blurry, deformed, bad anatomy, extra limbs, fused fingers, text, watermark, cartoon, anime, plastic skin, overexposed, different outfit, age change, body distortion, jittery motion, small movements, weak dancing, no music

