Cinematic scene with Topview
Nano Banana 2 for the characters, Seedance 2.5 for the video.
Tools in this guide: Topview
A simple step-by-step guide + the exact Seedance 2.5 prompt I used.
The main tool: Topview AI
I used Topview AI as the workspace for the whole creation process: generating the character references with Nano Banana 2, then turning those references into the final continuous cinematic scene with Seedance 2.5.
What you need
Tool · What it does
Topview AI · The main workspace for building the scene.
Nano Banana 2 · Create the character/reference images you want to keep consistent.
Seedance 2.5 · Generate the final 30-second cinematic video from your references + prompt.
Reference images · Character, environment, bodyguards, waiter/bouquet, and other visual references used by the prompt.
Step-by-step
1. Open Topview AI. Start your project inside Topview AI.
2. Create your characters with Nano Banana 2. Generate the people you want in the scene. Keep the faces, clothing, and overall look consistent because these images become your identity references.
3. Create your main starting frame. Build one strong restaurant frame that establishes the table, camera angle, lighting, composition, clothing, and environment. This becomes the visual anchor for the entire video.
4. Prepare all your references. Upload your character images and any important supporting references. In my scene, the prompt also uses bodyguards, a waiter carrying roses, and an additional character.
5. Choose Seedance 2.5 in Topview AI. Switch to Seedance 2.5 for the video generation stage, then add the reference images in the order that matches your prompt.
6. Set the video to 30 seconds. This scene was designed as one continuous 30-second cinematic take, with no cuts or scene changes.
7. Paste the prompt below. Use the full prompt as your starting point. Replace my image references, characters, environment, actions, and dialogue with your own when making a different scene.
8. Hit Generate. Generate the video, review character consistency and timing, then refine your references or prompt if needed.
Reference images I used (create or upload these inside Topview AI, then attach them as references for Seedance 2.5): Character B, Bodyguards, Waiter + roses, Character C, Starting frame, Character A.
Tip: keep the same people, outfits, lighting, and restaurant style across your reference images.
Before you generate
- Keep your image numbering consistent. The prompt refers to specific image numbers, so those references need to match what you upload.
- A strong starting frame matters. It tells the model what should stay visually stable throughout the shot.
- Be specific about continuity: faces, clothing, table position, lighting, camera angle, props, and who enters when.
- For a continuous scene, clearly say no cuts, no camera-angle changes, no time jumps, and no teleporting characters.
The exact prompt I used
Copy this as a reference. The image numbers below correspond to my own uploaded reference images, so replace them with your own references when recreating the workflow.
Animate the attached restaurant image into a single continuous 30-second photorealistic video while preserving the exact identities of all referenced characters and the exact restaurant environment throughout the entire shot.
Use @[Image 1](image_1) as the exact identity reference for Character B, the beautiful young woman who is Character A’s date.
Use @[Image 2](image_2) as the exact identity reference for Character A, the bald male character wearing the expensive luxury suit.
Use @[Image 3](image_3) as the exact starting frame and primary scene reference. This image defines the restaurant, table, seating positions, camera angle, composition, lighting, lens, perspective, clothing, and overall visual atmosphere.
Use @[Image 4](image_4) as the exact visual reference for the two professional bodyguards who appear later in the scene. Preserve their black suits, sunglasses, serious appearance, body proportions, and security earpieces.
Use @[Image 5](image_5) as the exact visual reference for the waiter carrying the large luxury bouquet of red roses who enters later in the scene.
Use @[Image 4](image_4) as the exact identity and appearance reference for Character C, the attractive young woman in the short red dress who appears only near the end of the scene.
Primary continuity requirements
Preserve the exact:
faces and identities of Character A and Character B
hairstyles
facial hair
clothing
body proportions
skin tones
seating positions
restaurant table
chairs
table decorations
glasses and plates
restaurant architecture
background
lighting
colors
shadows
depth of field
perspective
camera position
framing
lens characteristics
Character A and Character B must remain seated at the same table in the same restaurant for the entire 30 seconds.
The restaurant itself must never transform, change location, morph, or disappear.
The entire video must feel like one real continuous event unfolding naturally at the same restaurant table.
The shot must be one uninterrupted continuous 30-second take.
No cuts.
No scene changes.
No alternate camera angles.
No montage.
No time jumps.
Keep the same moderately wide 4:3 cinematic composition from @[Image 3](image_3) throughout the entire generation.
The framing must continuously keep Character A and Character B clearly visible while also preserving enough surrounding space for the bodyguards, waiter, bouquet, and Character C to naturally enter the scene.
The camera remains approximately eye-level and mostly stationary, with only extremely subtle natural cinematic camera movement.
Do not zoom, reframe, pan dramatically, cut to close-ups, or change the composition.
Overall performance
Character A and Character B are having a luxurious first-date conversation.
Their performances should feel natural, playful, understated, flirtatious, and humorous, not theatrical and not like actors performing a commercial.
Character B calmly explains what she likes in a man.
Each thing she says causes Character A to casually demonstrate that he apparently has exactly what she wants.
The comedy should come from Character A becoming increasingly absurdly prepared for every requirement while behaving as if everything is completely normal.
Character A should remain confident, composed, and slightly amused.
Character B should begin confident and impressed, but gradually become more surprised by each escalation.
Dialogue and actions in exact chronological order
At the beginning of the shot, Character A and Character B are already seated naturally across from each other exactly as shown in @[Image 3](image_3).
They are having a relaxed upscale dinner-date conversation.
Character A looks naturally toward Character B and asks:
A: “So... what do you like?”
Character B looks at Character A and replies casually:
B: “I like expensive taste.”
Character A gives her a subtle confident look.
Without standing up, Character A naturally reaches inside his suit jacket and pulls out a small luxurious black velvet ring box.
He opens or presents the ring box just enough to reveal a large beautiful diamond ring.
The movement must be physically realistic: one hand enters the jacket, retrieves the small ring box, brings it naturally into view, and presents it toward Character B.
The ring must not magically appear.
Character B looks at the diamond ring with a pleasantly surprised reaction.
She then continues:
B: “I like power.”
Character A remains completely calm.
Immediately after she says this, the two bodyguards from @[Image 4](image_4) naturally walk into the restaurant scene from behind or from the sides of Character A.
They move into position and stand seriously behind Character A, one slightly to his left and one slightly to his right.
They must clearly match @[Image 4](image_4).
Both bodyguards wear black suits, dark sunglasses, and professional security earpieces.
They stand still, alert, serious, and intimidating, like elite personal security.
They do not speak.
Character A barely reacts, as though having two bodyguards suddenly stand behind him is completely normal.
Character B looks briefly at the guards, visibly more impressed and slightly confused.
She then continues:
B: “I like attention.”
A moment later, the waiter from @[Image 5](image_5) naturally walks into frame carrying the same large luxurious bouquet of red roses shown in @[Image 5](image_5).
The waiter approaches the table naturally.
He carefully presents or places the bouquet into the intentionally empty section of the restaurant table near Character B.
The bouquet must remain physically realistic and must occupy the available open tabletop space without clipping through glasses, plates, hands, or characters.
Character B looks at the bouquet, then back toward Character A.
Her reaction should suggest that she is now genuinely impressed by how ridiculous this is becoming.
Character A remains composed and confident.
Character B then says:
B: “I like drama.”
After a short natural beat, Character C from @[Image 4](image_4) walks naturally into the restaurant scene from the side.
Character C must preserve the exact young adult identity, hairstyle, facial appearance, body proportions, and short red dress from @[Image 4](image_4).
She confidently approaches Character A’s side of the table.
She looks directly toward Character A with familiarity and casually says:
C: “Hi.”
Character A freezes for a very brief moment.
Character B slowly shifts her eyes from Character C toward Character A.
Her expression changes from impressed to suspicious and mildly annoyed, creating the comedic payoff.
The two bodyguards remain completely serious behind Character A.
The waiter can naturally finish his action and move slightly away without distracting from the main characters.
The bouquet remains on the table.
The diamond ring remains physically consistent wherever Character A placed or held it.
Character C remains beside the table.
Character A realizes the situation has finally become awkward.
He then turns his head and looks directly into the camera for the first time in the entire video, naturally breaking the fourth wall.
Character A says:
A: “This whole scene was generated with Seedance two point five on Topview.”
Then, still looking directly into the camera:
A: “Here’s how I made it. Back to the studio.”
End the shot naturally immediately after this line.
Character behavior requirements
Character A must speak only Character A’s dialogue.
Character B must speak only Character B’s dialogue.
Character C must speak only the single word “Hi.”
The bodyguards must not speak.
The waiter must not speak.
Do not swap voices or dialogue between characters.
Maintain accurate lip synchronization for every spoken line.
Use natural:
eye contact
blinking
breathing
head turns
small hand gestures
posture changes
facial reactions
timing
pauses
interaction between characters
The comedy must remain subtle and believable.
Do not make any character dramatically overact.
Character A continuity
Character A must remain visually identical to @[Image 1](image_1) and his appearance in @[Image 3](image_3) throughout the entire shot.
Preserve his:
bald head
facial features
facial hair
skin tone
expensive suit
shirt
accessories
body proportions
age
Do not change Character A’s face or clothing when new characters enter.
Character B continuity
Character B must remain visually identical to @[Image 1](image_1) and her appearance in @[Image 3](image_3) throughout the entire shot.
Preserve her:
young adult appearance
cute facial features
hairstyle
makeup
skin tone
dress
body proportions
jewelry
age
Her reactions may naturally change, but her identity and physical appearance must remain unchanged.
Bodyguard continuity
The two bodyguards must appear only after Character B says:
“I like power.”
They must visually match @[Image 4](image_4).
Do not duplicate them.
There must be exactly two bodyguards.
They walk into position naturally rather than suddenly teleporting into existence.
Once positioned behind Character A, they remain physically stable and consistent.
They must not block Character A’s face.
They must not block Character B.
Waiter and bouquet continuity
The waiter must appear only after Character B says:
“I like attention.”
The waiter and bouquet should visually match @[Image 5](image_5).
The waiter must walk naturally into the restaurant environment and physically carry the bouquet.
The bouquet must not magically appear.
He places or presents the bouquet naturally.
The bouquet then remains physically present and consistent for the rest of the video.
Do not duplicate the bouquet.
Character C continuity
Character C must appear only after Character B says:
“I like drama.”
She must visually match @[Image 4](image_4) exactly.
She walks naturally toward the table.
She does not teleport or suddenly materialize.
She speaks only:
C: “Hi.”
She should behave confidently and naturally, creating immediate awkward tension between Character A and Character B.
Do not confuse Character C with Character B.
Their faces, hairstyles, dresses, voices, and identities must remain clearly different.
Object continuity
The restaurant table must stay in exactly the same location.
The diamond ring box must remain physically consistent after Character A removes it from his jacket.
Do not generate multiple ring boxes or multiple rings.
The bouquet must remain where the waiter places it.
Glasses, plates, silverware, chairs, and table decorations must not randomly disappear, duplicate, move, or morph.
Characters must interact physically correctly with all props.
Hands must remain anatomically correct.
Objects must not pass through hands, bodies, the table, or each other.
Camera and scene continuity
The entire 30-second video must visually feel like one uninterrupted shot filmed by one real camera.
Maintain the exact restaurant environment from @[Image 3](image_3) throughout.
Do not replace the background when guards enter.
Do not change the restaurant when the waiter arrives.
Do not change the environment when Character C arrives.
Do not alter the lighting between beats.
Do not create sudden exposure changes.
Do not change camera height.
Do not change focal length.
Do not change depth of field.
Do not move Character A or Character B to different seats.
Every new person must enter and interact with the existing physical scene.
Negative prompt
Do not change Character A or Character B’s identities.
Do not change their faces, hairstyles, clothing, age, skin tone, body proportions, or accessories.
Do not confuse Character B with Character C.
Do not swap faces.
Do not swap voices.
Do not create duplicate characters.
Do not create more than two bodyguards.
Do not create additional waiters.
Do not create extra women.
Do not add random restaurant customers near the main table.
Do not add background faces that become prominent.
Do not teleport characters into the scene.
Do not make the bodyguards suddenly materialize.
Do not make the waiter suddenly materialize.
Do not make Character C suddenly materialize.
All entering characters must physically walk into frame naturally.
Do not make the diamond ring magically appear.
Do not duplicate the diamond ring or ring box.
Do not duplicate the bouquet.
Do not make props float.
Do not create extra arms, hands, fingers, legs, heads, or bodies.
Do not merge characters together.
Do not allow bodies to pass through furniture.
Do not allow hands to pass through objects.
Do not change the restaurant.
Do not change the table.
Do not change the lighting.
Do not change the camera position.
Do not change the composition.
Do not introduce cuts, close-ups, B-roll, transitions, montages, zooms, new angles, subtitles, captions, logos, UI elements, or on-screen text.
Do not make the performance exaggerated, slapstick, cartoonish, or unrealistic.
The entire video must remain photorealistic, cinematic, physically believable, temporally consistent, and visually continuous for the full 30 seconds.
Made with Topview AI • Nano Banana 2 • Seedance 2.5
More guides
All guides · All AI tools · Prompts · HandlyTech
