映画 · 新着

And I Need You

追加するもの：すべての被写体が所定の位置に写った一枚の開始画像。スタジオでアップロードします。

開始画像を 16:9 にトリミングしてください。このモデルはアップロード画像の構図を維持します。

話題の作品からの着想

高品質な五秒の参考プレビュー。5sまたは10sのショットを生成するか、30–120sの複数ショット映画を計画できます。

コミュニティの原文プロンプト

UNIVERSAL VIDEO REMAKE — COMPLETE CHARACTER REPLACEMENT

1. CHARACTER REFERENCE ASSIGNMENT — ABSOLUTE LOCK

Use @video1 as the source for motion, performance, timing, choreography, and overall cinematic energy.

Completely replace EVERY original human character visible in @video1 with the corresponding character reference listed below.

CHARACTER MAPPING — EDIT THIS SECTION ONLY:

The character reference images are provided in the following order:

@image1 - driver songer. @image2 - passenger a passenger with a small dog in their arms. 

Add or remove character assignments according to the number of replacement characters required. Each reference corresponds to exactly one original character identity.

The order of the assignments is absolute. Match each reference to its intended original character and maintain that mapping consistently throughout the entire video, including cuts, camera movements, temporary occlusions, close-ups, wide shots, reflections, and reappearances.

Never swap identities between characters. Never merge two identities. Never create hybrid faces or transfer distinctive features between different references.

Each reference image is the exclusive source of visual identity for its assigned replacement character.

Transfer and preserve the reference character's recognizable identity, including:

- Face and facial structure
- Eyes, eyebrows, nose, lips, and ears
- Skin tone and natural skin characteristics
- Hair, hairstyle, and facial hair
- Body type and proportions
- Clothing, footwear, accessories, and distinctive styling
- Age appearance and other defining visual characteristics

Preserve the reference identity even when the character turns away, moves rapidly, changes pose, enters unusual lighting, or appears in a different camera angle.

Do not inherit the original character's face, hairstyle, facial hair, clothing, body identity, or distinctive appearance unless a specific feature independently exists in the assigned reference.

Facial hair must come exclusively from the assigned reference. Never transfer a mustache, beard, or stubble from an original character to a clean-shaven reference character.

2. COMPLETE VISUAL RECONSTRUCTION — NO ORIGINAL CHARACTER TRACES

Recreate the video as a new, fully realized cinematic production rather than a superficial face swap or character overlay.

The final video must contain no recognizable visual remnants of the original characters.

Completely reconstruct each character's visible appearance, including the face, head, hair, neck, body, hands, clothing, shoes, and accessories whenever visible.

Remove the original characters' distinctive visual identity from every frame, including peripheral appearances, background figures, reflections, shadows that reveal their original silhouettes, and partially occluded body parts.

All visible instances of a replaced character must use the same assigned reference identity.

Do not preserve original character textures beneath the replacement. Do not leave original facial fragments, hairlines, skin patches, clothing patterns, silhouettes, or blended features.

The result must look like a genuine live-action performance by the replacement characters, not an AI face pasted onto the original footage.

3. NEUTRAL LOCATION RECONSTRUCTION — SUBTLE REDESIGN ONLY

Create ONE coherent, neutral, understated location for the entire video.

The original @video1 is the absolute authority for every shot, camera angle, camera position, camera movement, framing, composition, edit, and moment in time. Do not invent any new visual coverage.

Reconstruct the environment with a very subtle redesign: preserve the general spatial logic required by the original performance, but alter environmental details just enough to make the location feel independently created rather than an exact reproduction of the original setting.

The location must be simple, believable, visually restrained, and secondary to the characters.

Use a neutral, minimally designed environment with coherent architecture, understated surfaces, natural spatial depth, and only the environmental details necessary to support the action. Avoid elaborate set design, recognizable landmarks, distinctive architecture, decorative clutter, dramatic environmental features, or visually dominant backgrounds.

STRICT LOCATION RULES

- Create one consistent location that remains spatially coherent across every original shot and cut.
- Do not create a different location for different shots.
- Do not introduce new rooms, new areas, new environments, or new establishing shots that are absent from the source video.
- Do not radically redesign the original spatial layout.
- Do not reproduce the original location's most distinctive identifying details literally.
- Reinterpret the original environment only slightly, using restrained changes to surface finishes, minor architectural details, textures, and other nonessential elements.
- Preserve the spatial relationships required for the original performance: walls, floors, furniture, vehicles, props, and other objects must remain positioned and proportioned appropriately for the actions performed.
- Retain any environmental object essential to the original choreography or physical interaction, adapting its appearance only as much as necessary.
- Do not add unnecessary furniture, props, signage, decorations, background characters, visual effects, or other invented elements.
- Do not make the environment more elaborate, luxurious, futuristic, stylized, cinematic, or visually complex than the source footage warrants.

The redesign must be deliberately restrained. The result should feel like the same type of physical space, subtly reimagined, not a completely different production design.

The new location must not become a visual concept of its own. It exists to support the performance, provide believable spatial continuity, and integrate the replacement characters naturally.

4. REFERENCE IMAGE ISOLATION — IDENTITY ONLY

The images @image1, @image2, and all additional character reference images are used EXCLUSIVELY to determine the appearance and identity of their assigned characters.

Extract only the character-specific visual information required for the replacement, such as facial identity, hair, facial hair, skin tone, body appearance, clothing, footwear, and accessories.

NEVER use character reference images as environmental, location, or production-design references.

Do not copy, reproduce, borrow, or derive any of the following from the reference images:

- Backgrounds, rooms, studios, streets, buildings, landscapes, or interiors
- Walls, floors, ceilings, architecture, furniture, or set dressing
- Vehicles, environmental objects, props, or surrounding objects that are not part of the character's actual appearance
- Background lighting setups, environmental color schemes, or distinctive location-specific visual treatments
- Camera angles, framing, compositions, or photographic setups from the reference images
- Any recognizable environment or visual setting visible behind or around the reference character

Do not reconstruct the reference photograph. Do not place the replacement character into the environment shown in their reference image. Do not blend the reference background with the source video's environment.

The replacement character must be placed exclusively within the newly reconstructed neutral location, following the original video's camera perspective, framing, lighting logic, and spatial layout.

When clothing or accessories are visually difficult to separate from the reference background, preserve only the elements that clearly belong to the character. Never interpret background elements as part of the new scene.

Identity comes from the reference images. Every shot, camera angle, composition, performance, and environmental requirement comes from @video1 or the neutral-location rules in this prompt.

5. ORIGINAL CAMERA AND SHOT LOCK — NO INVENTED ANGLES

@video1 is the sole and absolute source of camera direction and shot selection.

Reproduce ONLY the shots and camera views that actually exist in the original video.

Do not generate additional angles, alternative perspectives, coverage shots, reverse angles, inserts, close-ups, wide shots, reaction shots, establishing shots, or cutaways that are not present in the source.

Do not imagine how the scene might look from another position. Do not show the action from a new viewpoint simply because it would improve visual variety or make the reconstructed location easier to understand.

For every original shot, preserve:

- The exact original camera angle and viewing direction
- The original framing, composition, and shot scale
- The original camera-to-subject distance and perspective
- The original camera height, tilt, and orientation
- The original camera movement, trajectory, and speed
- The original subject positions within the frame
- The original depth relationships and foreground/background arrangement
- The original shot duration, cut points, and sequence
- The original focus behavior, depth of field, and motion blur, as closely as possible

If the source cuts from one angle to another, reproduce that same cut and use only the corresponding original angle. Do not insert intermediate views or create a more comprehensive coverage of the action.

If the source contains a single continuous shot, keep it as a single continuous shot. Do not add cuts or change the camera position during the shot.

If a character or object is partially outside the frame, preserve that framing. Do not widen the shot, pull the camera back, or reveal additional parts of the location that the original video did not show.

If the original camera movement is limited, keep it limited. Do not add artificial camera movement, zooms, pans, orbiting, tracking shots, handheld shake, or cinematic reframing.

The reconstructed location must adapt to the original camera views. The camera must NEVER be moved or redesigned to accommodate the new location.

Do not invent new shots to show off the replacement characters or the reconstructed environment. Reproduce the original visual coverage, not an imagined alternative edit.

6. LOCATION CONTINUITY ACROSS ORIGINAL SHOTS

Although the location is subtly reimagined, it must represent ONE consistent physical place.

Every original camera angle must correspond to the same coherent environment. Walls, floors, doors, furniture, vehicles, props, and architectural features must maintain consistent placement, appearance, scale, and orientation whenever visible from different angles.

Do not independently invent a background for each shot. Do not create unrelated environments that happen to share a similar color palette.

Use the original video's spatial information to maintain continuity. Where the source does not reveal a particular part of the location, keep the unseen area simple and neutral rather than inventing elaborate architecture or distinctive set design.

Only reconstruct environmental information that is necessary to make the existing shots physically coherent. Do not use unseen areas as an opportunity to add new visual concepts.

The result must feel as though all the original performances were filmed in one understated location, using exactly the same camera coverage and editing structure as the source.

7. CAMERA AND EDITING CONTINUITY

Preserve the original video's temporal structure and cinematic camera behavior.

Match the original:

- Shot order and edit timing
- Shot duration and transition timing
- Camera movement and movement speed
- Framing, shot scale, and composition
- Camera-to-subject distance
- Camera angle, perspective, and lens character
- Focus behavior, depth of field, and motion blur
- Character screen position and spatial relationships

Retain the original camera choreography even when rebuilding the environment.

Do not add new cuts, slow motion, speed ramps, artificial zooms, camera shakes, or transitions that are absent from the source.

Do not freeze, skip, repeat, or arbitrarily extend moments.

Where the new environment requires reconstructed visual elements, adapt the background and set geometry around the preserved camera path.

Maintain consistent spatial continuity across all shots.

8. FULL CINEMATIC LIVE-ACTION QUALITY

The finished result must have the visual quality and emotional credibility of a professionally directed, fully photographed music video.

AESTHETIC DIRECTION:

Early-2000s music-video atmosphere, authentic hip-hop visual culture, confident and distinctive art direction, tactile photographic textures, cinematic lighting, rich but controlled color, natural imperfections, and a strong sense of physical presence.

Use a coherent combination of:

- Filmic highlight roll-off and natural shadow detail
- Organic 35mm-film-inspired grain
- Realistic skin texture, pores, and fine facial detail
- Natural skin tones without artificial smoothing
- Physically plausible lighting, reflections, and shadows
- Realistic lens rendering, depth of field, and motion blur
- Subtle optical imperfections where appropriate
- Authentic material textures in clothing, hair, skin, vehicles, and surroundings
- Carefully controlled contrast and cinematic color grading
- A tactile, slightly imperfect, early-2000s music-video finish

The exact color palette should remain compatible with the source video's mood while being reinterpreted for the new environment.

Avoid excessive digital sharpness, sterile perfection, artificial HDR, exaggerated clarity, and overly processed colors.

 9. ANTI-PLASTIC / ANTI-AI QUALITY LOCK

The result must look photographed, not synthetically rendered.

Absolutely avoid:

- Plastic, waxy, rubber-like, or porcelain skin
- Excessive skin smoothing or beauty-filter effects
- Artificial facial symmetry
- Unnaturally glossy faces or lips
- CGI-like hair, skin, clothing, or environments
- Uncanny eyes, inconsistent pupils, or artificial eye reflections
- Warped anatomy, malformed hands, extra fingers, or disappearing limbs
- Facial identity drift or unstable facial proportions
- Flickering textures, crawling details, or inconsistent clothing
- Melting objects, background deformation, or unstable geometry
- Floating feet, sliding contact points, or incorrect body weight
- Fake reflections, incorrect shadows, or inconsistent lighting
- Visible face-swap boundaries, masks, seams, halos, or compositing artifacts
- Unmotivated lens flares, glow, particles, glitch effects, or excessive bloom

Preserve natural microtexture and physically plausible imperfections.

Skin must retain realistic pores, subtle tonal variation, and natural specular response. Hair must move and respond to light naturally. Clothing must show believable folds, fabric weight, and motion.

Do not beautify, idealize, stylize, or redesign the reference characters beyond what is necessary to integrate them convincingly into the scene.

10. PHYSICAL INTEGRATION AND FRAME-TO-FRAME STABILITY

Every replacement must be integrated into the reconstructed footage with consistent physical and optical behavior.

Maintain:

- Correct anatomy and body proportions
- Correct subject-to-environment scale
- Consistent perspective and depth
- Accurate occlusion between characters and objects
- Correct contact points and weight distribution
- Natural foot placement and ground contact
- Physically plausible cloth and hair movement
- Consistent lighting direction, intensity, and color
- Realistic cast shadows and contact shadows
- Coherent reflections on glass, mirrors, vehicles, and other reflective surfaces
- Natural motion blur consistent with the original movement and camera
- Stable identity, wardrobe, accessories, and body proportions across frames

When characters overlap, cross in front of one another, or temporarily disappear behind objects, preserve the correct depth order and identity.

Reconstruct reflections and other visible representations of replaced characters using the assigned reference identity. Never allow the original character to remain visible in a mirror, window, vehicle surface, or background reflection.

No flicker, morphing between identities, temporal warping, inconsistent anatomy, or sudden appearance changes.

11. STRICT PRIORITY ORDER

When requirements conflict, follow this priority order:

1. Preserve original performance timing, body motion, facial expression dynamics, and lip-sync.
2. Preserve the assigned reference character's identity without blending in the original character.
3. Preserve the original shot timing, camera movement, framing, and spatial choreography.
4. Reconstruct the environment and integrate all subjects with convincing physical continuity.
5. Achieve the cinematic early-2000s aesthetic and natural photographic texture.

Resolve conflicts by reconstructing the new character and environment around the original performance, not by altering the performance to accommodate the reconstruction.

12. FINAL OUTPUT REQUIREMENTS

Deliver one coherent, fully reconstructed cinematic video.

Every original human character must be replaced by its assigned reference character.

Every replaced character must remain visually consistent throughout the entire video.

The original characters' recognizable identities must be completely absent.

The environment must be newly reconstructed and visually distinct while retaining the original video's broad cinematic mood and energy.

The original performance, gestures, facial expressions, choreography, shot timing, and lip-sync must be preserved as faithfully as the source footage allows.

The finished video must feel like a genuine early-2000s-inspired live-action music video starring the assigned reference characters, filmed in a new location, with natural skin, realistic materials, seamless compositing, and professional cinematic image quality.

No superficial face swap. No identity blending. No generic motion. No plastic skin. No recognizable original character appearance.

Rebuild the visual world. Replace every character. Preserve the performance.

この原文プロンプトには、このモデルでは受け取れない必須の参照メディアまたは保持音声が必要です。コピーして別の場所で使用してください。

生成プロンプト

以下のモデルであなたの画像から類似コンテンツを作成できるよう調整済みです。結果は画像と生成ごとに異なります。

被写体の案内に合う鮮明な開始画像をアップロードしてください。
このテンプレートで編集可能な十秒のプロンプトと設定を読み込んでください。
5sまたは10sのショットを生成するか、Film Studioでより長い複数ショットの映画を制作してください。

費用: 64 cr
直接クリップ

もっと長い動画が必要ですか？

編集したプロンプトを連続する10秒ショットとして計画し、30–120秒の映画に組み立てます。

このテンプレートの調整を手伝ってほしいですか？

編集したプロンプトをLLM Chatに引き継ぎ、Prompt writer、Storyboard、Model pickerなど、すべてのStudioツールを利用できます。

LLM Chatで開く →

生成を押すまで課金されません。生成に失敗した場合は全額返金されます。