Soraのプロンプトに対する仮説

目次

1. 動画　Cherry Blossom Serenity
2. プロンプト詳細
3. 他の動画
4. まとめ

前の記事で書いたように、Soraにアクセスできるようになってすぐ試してみたのですが、そのあと、今更ながらStable Diffusionに触り始めて思ったことがあるのでちょっと書いてみます。

動画　Cherry Blossom Serenity

一番良くできたなと思っている動画です。

https://sora.com/g/gen_01jf1r9g2gfndtr5ftn5xevr9n

https://www.youtube.com/watch?v=tEgh0GYhZ6c

Soraにアカウントがある方ならそちらで見られるのかな？自分ではよくわからないのでお好きな方で見てやってください。

プロンプト詳細

1. Character Features
Basic Information
Subject: Japanese woman in her 20s.
Body Type: Slim and natural, maintaining a calm and composed posture.
Clothing:
A white off-shoulder shirt paired with long denim jeans for a casual yet clean and elegant spring style.
Hairstyle
Hair:
A ponytail that sways gently in the spring breeze.
Expression
Smile:
Subtle, soft, and slightly shy. Warm and modest.
Face:
Natural beauty with distinctly Japanese features.
2. Actions and Timing
(All movements and gaze should include very small, subtle variations.)

0.0–2.0 Seconds: Calm Beginning
Pose:

The woman stands still, facing the camera directly.
Her head is slightly lowered, and her gaze is directed downward.
Subtle Variations:

Even while standing still, slight natural movements in her shoulders or neck should be present for realism.
2.0–5.0 Seconds: Lifting the Gaze
Action:

She slowly raises her head and lifts her gaze toward the camera.
Add a brief glance to the side or downward before fully meeting the camera’s gaze.
Expression:

A small, shy smile appears as her gaze meets the camera.
Subtle Variations:

The movement of her gaze and head should include minor, smooth shifts to avoid a rigid appearance.
5.0–7.0 Seconds: Subtle Shyness
Action:

While maintaining the smile, her gaze briefly shifts away from the camera, looking downward or slightly to the side.
A slight tilt of her head adds a natural touch.
Subtle Variations:

Her gaze and head movements should include tiny, natural sways for a lifelike effect.
7.0–10.0 Seconds: Closing Gesture
Action:

She brings her gaze back to the camera, continuing to smile gently.
To finish, she slightly lowers her gaze or tilts her head modestly.
Expression:

The closing smile is gentle, warm, and leaves a lasting impression of kindness.
Subtle Variations:

Even in her final gaze and movements, minor variations should be present to enhance naturalness.
3. Background and Environment
Setting
A cherry blossom-lined path or park, evoking a serene and spring-like Japanese ambiance.
Time of Day
Bright daytime with soft, natural sunlight filtering through the trees.
Atmosphere
Calm and slightly fantastical:
Incorporate floating light particles and gently falling cherry blossom petals to enhance the tranquil setting.
4. Effects and Enhancements
Cherry Blossom Petals
Softly drifting petals surround the woman, creating a dynamic yet gentle visual effect.
Light Particles
Subtle, glowing light particles float in the air, adding to the serene and dreamlike atmosphere.
Wind Effects
Her ponytail and clothing move slightly in the breeze to enhance realism.
5. Essential Guidelines
Subject Must Be a Japanese Woman in Her 20s:

The appearance and demeanor should authentically reflect a Japanese woman.
Minimal Movements:

All actions must be small, controlled, and elegant, staying within a natural range.
Subtle Variations in All Movements and Gaze:

Include slight, smooth shifts in gaze, head, or shoulders to prevent rigidity and create natural realism.
Focus on a Gentle Smile:

The smile should remain soft, modest, and inviting, emphasizing warmth and sincerity.

Stable DiffusionをSoraの後に触り始めて思ったのですが、SoraはChatGPT並みに理解してくれるということです。指示が出しやすい。Stable Diffusionのように単語の組み合わせでギリギリの作品を作る感覚ではなく、ChatGPTに相談して作った長文を貼り付ける形でいけるのではないでしょうか？より人間の指示の出し方を理解してくれてますね。

そうは言っても動画の動きと言う複雑なものに対する指示なので、人間の意志とAIの意志が乖離しやすい指示は実行しにくいというのが現状かなと思います。例えば、バイバイの手の振り方は具体的にどうするのと言うと具体的な指示にするには複雑だし、大まかな指示では伝わらないかと思います。

ただ、入力欄がこのような長文を入力するようにはできていなさそうなので、想定されていない指示の出し方な気もしますがどうでしょうか？Soraで動画を開いてプロンプトを見れますが、長文は文字がスクロールできないのでCTRL+Aでコピーするしか見れない気がするのはうちの環境だけですか？ただ、あまりに長文ははじかれますので、ご注意ください。ん、他のサービスの話とごっちゃになってるか？長文いけるか気になる方は試して下さい。

他の動画

Cherry Blossom Serenity

https://sora.com/g/gen_01jf1ma2bjfrday6c1748tmy2j

https://www.youtube.com/watch?v=VbOrg2KoqFU

1. Female Character Features
Age: Early 20s, youthful and cheerful.
Clothing: Casual, spring-like outfit (e.g., pastel dress or white shirt with a denim skirt).
Hairstyle: Ponytail slightly swaying in the breeze.
Expression: Shy yet smiling, conveying a gentle and warm demeanor.
2. Actions and Timing
0.0–2.0 seconds:

The woman stands in the center, gazing downward shyly with a gentle smile.
Background: Cherry blossoms swaying slightly with petals falling.
2.0–4.0 seconds:

She begins waving her right hand delicately, paired with a small, shy smile.
Light particles softly drift in the air, emphasizing a fantasy-like atmosphere.
4.0–6.0 seconds:

She lowers her hand, tilting her head slightly while glancing away.
Her ponytail and outfit gently move with the breeze, adding a natural touch.
6.0–8.0 seconds:

She takes a small step back, lightly bows, or clasps her hands in front of her.
Background: Cherry blossom petals swirl around as light particles reflect in the sunlight.
8.0–10.0 seconds:

She looks back at the camera for a final smile before slowly averting her gaze.
The scene fades out with a wide shot of the cherry blossom trees.
3. Environment and Effects
Setting: A cherry blossom-lined path on a sunny spring day.
Ambiance: A soft, glowing fantasy-like vibe with light particles in the air.
Effects: Subtle wind moves her hair and outfit, with petals gracefully falling around her.

Cherry Blossom Elegance

https://sora.com/g/gen_01jf1mwpzbfecrfxp8gnjexxkb

https://www.youtube.com/watch?v=CbrGQprOv78

1. Female Character Features
Age: Early 20s, youthful and healthy.
Clothing:
A white off-shoulder shirt paired with long denim jeans.
The outfit should convey a casual, spring-like style.
Hairstyle: A ponytail that sways gently in the breeze.
Facial Features:
Clearly Japanese facial structure with a soft, natural appearance.
Her expression is shy but warm, with stable and steady movements.
Subtle gestures, like small head tilts or natural eye movements, can be included to match her emotions.
2. Actions and Timing
The woman remains standing in the same position, facing forward, without moving from her spot.
The camera is fixed, focusing on her from the front.
0.0–2.0 seconds

The woman stands facing the camera with her gaze slightly lowered.
She slowly lifts her head and makes eye contact with the camera, revealing a shy and gentle smile.
2.0–4.0 seconds

She begins to wave her right hand delicately.
The wave is small and slow, reflecting a sense of shyness and grace, while her smile remains natural.
4.0–6.0 seconds

After finishing the wave, she lowers her hand gracefully.
She tilts her head slightly and momentarily glances away from the camera, then looks back with a soft expression.
6.0–8.0 seconds

She adjusts her posture slightly, keeping both hands either relaxed at her sides or lightly clasped in front of her.
Her expression remains calm and approachable, with her smile lingering.
8.0–10.0 seconds

For the final moment, she looks directly at the camera with a warm, shy smile before slowly averting her gaze.
The background begins to fade out as the scene transitions to a soft ending.
3. Background and Effects
Setting: A cherry blossom-lined path or a park with full-bloom sakura trees.
Time: Bright spring daytime, with warm sunlight creating a cheerful atmosphere.
Atmosphere: A light, fantasy-like touch with glowing particles softly drifting in the air.
Effects:
Cherry blossom petals float gently in the breeze, occasionally brushing past her hand or shoulder.
Her hair and outfit sway slightly, emphasizing the presence of a soft spring wind.

Cherry Blossom Serenity

https://sora.com/g/gen_01jf1n4fb3frz90ewqec2n1a47

https://www.youtube.com/watch?v=AAsGObt6CwU

1. Female Character Features
Age: Early 20s, youthful and cheerful.
Clothing:
A white off-shoulder shirt paired with long denim jeans, reflecting a casual yet spring-appropriate look.
Hairstyle: A ponytail that moves slightly with the breeze.
Facial Features and Eye Movements:
Natural Japanese facial structure with a soft, gentle expression.
Eye Movements: Instead of keeping her gaze entirely fixed, the eyes can subtly shift to reflect her emotions, such as glancing momentarily to the side or downward before returning to the camera.
Head Movement: Gentle, slight tilts or nods to avoid rigidity while maintaining composure. Movements should be smooth and consistent to prevent unnatural breaks.
2. Actions and Timing
The woman remains in the same position from start to finish, but her gaze and expressions show natural, subtle variability.
0.0–2.0 seconds

She starts by looking downward with a shy expression.
Gradually, her eyes lift toward the camera while her face shows a soft, hesitant smile. During this, her gaze may shift slightly to the side before settling on the camera.
2.0–4.0 seconds

She begins waving her right hand in a small, slow, and delicate motion.
Her eyes lightly follow the motion of her hand, shifting back to the camera intermittently. Her smile grows slightly warmer as she waves.
4.0–6.0 seconds

After finishing the wave, she lowers her hand and tilts her head slightly to one side.
Her eyes momentarily glance downward or to the side before returning to the camera with a soft, shy expression.
6.0–8.0 seconds

She gently adjusts her posture, perhaps clasping her hands lightly in front of her or letting them rest naturally by her sides.
Her eyes show slight movement, as if she’s thoughtfully observing something nearby, then briefly returning to the camera.
8.0–10.0 seconds

For the final moment, she gazes warmly at the camera with a small, shy smile.
Her eyes softly drift to the side before the scene fades out, giving a natural and relaxed conclusion.
3. Background and Effects
Setting: A cherry blossom-lined path or a tranquil park with full-bloom sakura trees.
Time: Bright spring daytime, with warm sunlight and a cheerful, serene atmosphere.
Atmosphere: A subtle fantasy-like tone with glowing light particles drifting gently in the air.
Effects:
Cherry blossom petals float softly in the breeze, occasionally brushing past her hand or shoulder.
Her hair and outfit sway gently with the wind, enhancing realism and movement.

Cherry Blossom Serenity

https://sora.com/g/gen_01jf1ntxs5fz7vmf00tte73j2k

https://youtu.be/5gdCrGlwapA

1. Key Features of the Woman
Facial Features and Appearance
Ethnicity: The woman must have distinctly Japanese facial features, including soft, natural contours, almond-shaped eyes, and a fair complexion.
Expression: Her expression is shy and reserved, conveying warmth and modesty through a gentle, genuine smile.
Body Type
Build: A petite and slender build typical of Japanese women, with a height and proportions that reflect a natural and realistic appearance.
Posture: Upright but relaxed, maintaining a calm and composed demeanor throughout.
Clothing
A white off-shoulder shirt paired with long denim jeans, creating a casual yet elegant spring look that feels contemporary and approachable.
Hair
A neatly tied ponytail, gently swaying in the breeze to enhance a sense of natural movement and softness.
2. Actions and Timing
0.0–2.0 seconds: Introductory Shyness
Pose:

The woman stands in place, facing the camera directly.
Her head is slightly lowered, and her gaze is directed downward, conveying modesty and introversion.
Action:

Gradually, she raises her head and makes eye contact with the camera.
Her lips curve into a soft, shy smile, showing warmth without exaggeration.
Details:

Her gaze may momentarily shift to the side before returning to the camera, adding a natural sense of uncertainty.
2.0–4.0 seconds: Beginning the Bai-Bai Gesture
Pose:

She raises her right hand slowly, positioning it near shoulder height.
The fingers are relaxed, with the palm facing outward.
Action:

She performs a small, gentle waving motion, using only her wrist to move her hand side-to-side.
The gesture is controlled and understated, reflecting a polite and modest tone.
Expression:

Her smile brightens slightly but remains reserved, maintaining the culturally appropriate sense of humility.
Eye Movement:

Her eyes briefly follow the motion of her hand before returning to meet the camera’s gaze, adding a subtle dynamic to her expression.
4.0–6.0 seconds: Ending the Gesture
Action:

After completing 2–3 small waves, she lowers her hand slowly and gracefully back to her side.
Pose:

She tilts her head slightly to one side, reflecting a gentle, shy demeanor.
Expression:

Her smile softens, and her eyes glance momentarily downward or to the side before returning to the camera.
6.0–8.0 seconds: Resting and Repose
Pose:

Standing still, she lightly adjusts her posture by clasping her hands in front of her or allowing them to rest naturally at her sides.
Expression:

Her gaze shows slight movement, such as a brief glance downward or away, before returning to the camera with a thoughtful and kind expression.
8.0–10.0 seconds: Final Smile and Farewell
Pose and Action:

She concludes by offering a final, soft smile directly at the camera.
Her head dips slightly in a modest bow, conveying humility and politeness.
Her gaze drifts gently away from the camera just before the scene fades out.
Expression:

Her closing smile is genuine and shy, leaving an impression of warmth and friendliness.
Transition:

The background begins to fade out with softly falling cherry blossom petals as the focus on her smile gently dissolves.
3. Background and Effects
Setting:

A cherry blossom-lined path or park with full-bloom sakura trees, representing a serene and quintessentially Japanese spring setting.
Time:

Bright daytime with warm, natural sunlight filtering through the trees.
Effects:

Cherry Blossom Petals: Gentle, randomized drifting of sakura petals enhances the visual softness.
Light Particles: Subtle glowing particles drift lightly in the air, adding a whimsical and dreamlike quality.
Wind Effects: Her hair and outfit sway slightly in the breeze, enhancing the realism and natural ambiance.
4. Cultural Emphasis for Non-Japanese Creators
Gestural Scale:

The wave should remain small and delicate, with movement restricted to the wrist. Large or exaggerated arm motions are not culturally appropriate in this context.
Expression and Tone:

Japanese expressions are often reserved and understated. The smile should be gentle and modest, avoiding overly broad or dramatic expressions.
Natural Movements:

The woman’s movements should be smooth and controlled, avoiding abrupt or exaggerated gestures. The overall tone must convey elegance and composure.
Eye and Head Movement:

Slight variations in gaze (e.g., momentary glances downward or sideways) create a sense of realism and emotional depth without breaking the serene flow of the video.

Cherry Blossom Grace

https://sora.com/g/gen_01jf1qpp23enc9rkp4n8ytnkd6

https://www.youtube.com/watch?v=dU5kPbwAk_w

1. Character Features
Basic Information
Subject: A Japanese woman in her 20s.
Body Type: Slim and natural, with a calm and composed posture.
Clothing:
A white off-shoulder shirt paired with long denim jeans for a casual yet clean and elegant spring look.
Hairstyle
Hair:
A ponytail, gently swaying in the spring breeze for a natural effect.
Expression
Smile:
A subtle and slightly shy smile. The smile should be small and soft, conveying warmth and a gentle demeanor.
2. Actions and Timing
0.0–2.0 Seconds: Initial Calmness
Pose:

The woman stands still, facing the camera directly, maintaining a stable and composed posture.
Her head is slightly lowered, and her gaze is directed downward.
Action:

She remains motionless at the beginning, creating a serene and tranquil atmosphere.
2.0–5.0 Seconds: Lifting the Gaze
Action:

She slowly raises her head, gradually lifting her downward gaze to meet the camera.
This movement should be smooth and deliberate, with a natural and unhurried rhythm.
Eye Movement:

As her gaze approaches the camera, she briefly glances to the side or slightly away before making full eye contact, adding a sense of hesitation or shyness.
Expression:

As her eyes meet the camera, a small, soft smile begins to appear, conveying warmth and modesty.
5.0–7.0 Seconds: Subtle Shyness
Action:

While maintaining the smile, she gently shifts her gaze away from the camera, looking downward or slightly to the side.
A slight tilt of her head or a small nod adds a touch of natural rhythm to her movements.
Expression:

The shy smile remains, reflecting a gentle and approachable demeanor.
7.0–10.0 Seconds: Closing Gesture
Action:

She gradually returns her gaze to the camera, maintaining her soft smile.
To conclude, she slightly lowers her gaze or tilts her head modestly, emphasizing a calm and reserved impression.
Expression:

The final smile is gentle and warm, leaving a lasting impression of kindness and softness.
3. Background and Environment
Setting
Cherry Blossom Scenery: A natural background featuring a cherry blossom-lined path or park, evoking the essence of spring in Japan.
Time of Day
A bright spring day with soft sunlight filtering through the trees, creating warm and gentle shadows.
Atmosphere
Serene and Fantastical:
Floating light particles and gently falling cherry blossom petals create a quiet, dreamlike ambiance.
4. Effects and Enhancements
Cherry Blossom Petals
Softly drifting cherry blossom petals naturally float around the woman, adding dynamic movement to the scene.
Light Particles
Subtle glowing particles enhance the scene’s gentle and fantastical atmosphere.
Wind Effects
The breeze lightly moves her ponytail and clothing, emphasizing a natural spring setting.
5. Essential Guidelines
Subject Must Be a Japanese Woman in Her 20s:

The appearance and demeanor should authentically reflect a Japanese woman.
Subtle and Small Movements:

The overall motion should be minimal and elegant, avoiding any large or exaggerated gestures.
Incorporate Eye Movement Variability:

The gaze should not remain fixed; include small, natural shifts such as brief glances to the side or downward.
Focus on a Gentle Smile:

The smile should remain soft and modest, with a warm and inviting impression.

Cherry Blossom Serenity

https://sora.com/g/gen_01jf1r2e33fbj9kphxmmd2ttah

https://www.youtube.com/watch?v=8MYNFbbdAa0

1. Character Features
Basic Information
Subject: Japanese woman in her 20s.
Body Type: Slim and natural with a stable, composed posture.
Clothing:
White off-shoulder shirt and long denim jeans for a casual spring look.
Hairstyle
Hair:
Ponytail, lightly swaying in the breeze.
Expression
Smile:
Subtle, soft, and slightly shy. Warm and modest.
Face:
Distinctly Japanese features with natural, understated beauty.
2. Actions and Timing
0.0–2.0 Seconds: Initial Stillness
Pose:
Standing still, facing the camera.
Head slightly lowered, gaze directed downward.
2.0–5.0 Seconds: Lifting Gaze
Action:

Slowly raise the head and lift the gaze toward the camera.
Add a small pause or glance sideways before making eye contact.
Expression:

A gentle, shy smile appears as the gaze meets the camera.
5.0–7.0 Seconds: Subtle Shyness
Action:
Maintain the smile while slightly averting the gaze downward or to the side.
Add a small head tilt for a natural touch.
7.0–10.0 Seconds: Closing Gesture
Action:
Gaze returns to the camera, accompanied by a soft smile.
Conclude with a slight tilt or lowering of the head.
3. Background and Environment
Setting
Cherry blossom-lined path or park for a spring ambiance.
Time of Day
Bright daytime with soft sunlight filtering through.
Atmosphere
Serene and slightly fantastical with floating light particles.
4. Effects and Enhancements
Cherry Blossom Petals
Gentle petals drifting around the woman.
Light Particles
Subtle glowing particles to enhance the serene atmosphere.
Wind Effects
Hair and clothing sway lightly in the breeze.
5. Essential Guidelines
Subject Must Be a Japanese Woman in Her 20s:

Features should authentically reflect Japanese beauty.
Minimal Movements:

Movements must be small, subtle, and elegant.
Natural Gaze and Expression:

Include slight variability in the gaze for natural realism.
Focus on a Gentle Smile:

The smile should remain modest, warm, and inviting.

Cherry Blossom Serenity

https://sora.com/g/gen_01jf1rq83efkhrbxefvc2vjv2c

https://www.youtube.com/watch?v=yVPEvN3ctnQ

1. Character Features
Basic Information
Subject: Japanese woman in her 20s.
Body Type: Slim and natural, with a composed and stable posture.
Clothing:
A white off-shoulder shirt paired with long denim jeans for a casual, clean spring look.
Facial Features
Authentic Japanese Appearance:
Gentle contours with a natural oval or slightly rounded face shape.
Almond-shaped eyes that reflect warmth and modesty.
Clear and natural skin tone, emphasizing simplicity and elegance.
Hairstyle
Hair:
A neatly tied ponytail that sways lightly in the breeze.
Expression
Smile:
A small, soft, and slightly shy smile that is warm and inviting.
2. Actions and Timing
(All movements and gaze should include very small, subtle variations. Camera remains fixed for a stable, close-up focus on the subject.)

0.0–2.0 Seconds: Calm Beginning
Pose:

The woman stands still, facing the camera directly.
Her head is slightly lowered, and her gaze is directed downward.
Subtle Variations:

Even in stillness, minor natural movements in her shoulders or neck ensure a lifelike appearance.
2.0–5.0 Seconds: Lifting the Gaze
Action:

She slowly raises her head and lifts her gaze toward the camera.
Before fully meeting the camera’s gaze, she briefly glances sideways or downward to add natural variability.
Expression:

A small, shy smile appears as her gaze meets the camera, exuding warmth and modesty.
Subtle Variations:

Smooth, minor shifts in her head and gaze prevent rigidity, adding natural charm.
5.0–7.0 Seconds: Subtle Shyness
Action:

While maintaining her smile, she shifts her gaze briefly away from the camera, looking downward or slightly to the side.
A slight tilt of her head complements the gaze shift, emphasizing a soft, natural demeanor.
Subtle Variations:

Tiny, smooth adjustments in her head and gaze make the motion feel lifelike and fluid.
7.0–10.0 Seconds: Closing Gesture
Action:

She returns her gaze to the camera with a warm smile, maintaining her shy yet gentle expression.
To conclude, she slightly lowers her gaze or tilts her head in a modest gesture.
Expression:

The final smile is soft and genuine, leaving a warm and lasting impression.
Subtle Variations:

Small, natural shifts in her gaze and posture add a touch of realism to her closing gesture.
3. Background and Environment
Setting
A cherry blossom-lined path or park, evoking a tranquil and spring-like Japanese ambiance.
Time of Day
Bright daytime with soft, natural sunlight filtering through the trees.
Atmosphere
Serene and slightly fantastical:
Subtle light particles and softly drifting cherry blossom petals enhance the dreamlike setting.
4. Effects and Enhancements
Cherry Blossom Petals
Softly drifting petals move naturally around the subject, complementing the scene’s gentleness.
Light Particles
Subtle, glowing light particles float in the air, adding to the serene and tranquil atmosphere.
Wind Effects
The ponytail and clothing sway gently in the breeze for a natural spring effect.
5. Essential Guidelines
Subject Must Be a Japanese Woman in Her 20s:

The appearance and demeanor should authentically reflect Japanese elegance.
Minimal Movements:

All actions should be small, controlled, and elegant, emphasizing natural and understated charm.
Subtle Variations in Movements and Gaze:

Slight, smooth shifts in gaze, head, or posture enhance realism without being distracting.
Camera Work:

The camera remains fixed and stable, focusing closely on the subject to highlight her subtle expressions and movements.
Focus on a Gentle Smile:

The smile should be soft, warm, and modest, emphasizing authenticity and sincerity.

まとめ

こんな風に書いてますが、私はプロンプトを読んでません。ChatGPTに丸投げしています。日本語状態で確認してはいますが、修正をChatGPTに投げているだけなので修正の加減はよく分かりません。逆に、その微調整をChatGPTが担ってしまうのもすごいと思います。

ちなみに他の動画サービスも少し触ってみたのですが、それらはStable Diffusion的なプロンプトのアプローチが必要かもしれません。Soraのすごさって動画作成能力じゃなくて、プロンプトの理解力なんじゃないかなと思ったりします。