Paste one prompt and two photos into a general AI video tool and you get the elevator scene: one person in headphones, the other saying “I love the Smiths,” then saying it again louder and walking out.
Image 1 is the listener, on the left. Image 2 is the speaker, on the right. That order is the whole prompt. If you swap the photos, you swap the roles. Writing a longer prompt will not fix a swapped upload.
These prompts are for ordinary image-to-video tools: anywhere you can attach reference images and type a prompt. Set 15 seconds, 720p, 9:16, and turn sound on if the tool has that switch. Use 16:9 only for the widescreen version below.
You do not need a prompt at all on I Love the Smiths AI. Upload the two photos there and the template casts the scene for you. Everything under this line is for a general video model.
Before you paste
| Setting | Use this |
|---|---|
| Images | Two. One face each. Image 1 = listener. Image 2 = speaker. |
| Length | 15 seconds. A 10-second cut is further down if the tool cannot do 15. |
| Frame | 9:16 for a phone post. 16:9 if you want the movie frame. |
| Quality | 720p if you can choose it. |
| Sound | On. The prompt writes the lines and the room tone. |
Photos that work: front-facing, even light, chest up, eyes visible, no sunglasses. Crop a couple selfie into two pictures first. The headphones do not need to be in the photo. The prompt puts them on the listener.
If the tool has only one image slot, do it in two passes. First make a still of both people already standing in the elevator, listener left, speaker right. Then animate that still with the short prompt at the bottom. Do not send a photo that still contains both faces as “Image 1.”
Prompt 1 — Two people
Use this for friends, coworkers, or anyone who is not playing it as a flirt. Image 1 stays in the elevator. Image 2 says the line and leaves.
A photoreal 15-second vertical scene, filmed like a real movie, inside a small old office elevator. Warm dark wood panels, brass trim, a brass handrail, brushed-metal doors, soft warm overhead light, a quiet morning. Natural skin, natural color, light film grain, a warm cinematic grade. Not CGI. Not a cartoon. Both people stand side by side facing the doors, about an arm's length apart, framed from the waist up so both faces stay large, sharp, and well lit.
Image 1 is the listener and stays on the LEFT in every frame. Copy that person's face exactly: shape, eyes, eyebrows, nose, mouth, skin tone, age, hair color, hair length, hairstyle, and any beard, glasses, or freckles. Keep the clothes from Image 1. Add large plain over-ear headphones with no logo.
Image 2 is the speaker and stays on the RIGHT in every frame. Copy that person's face and clothes the same way. They hold a thin stack of folders.
If a reference image shows an animal, keep that animal's real face, fur, and markings. Never blend the two faces. Never swap their clothes. Each face stays the same from the first frame to the last.
Shots:
0-3s. Medium two-shot from the front. The elevator hums upward. The listener stares ahead and nods slightly to the music. Faint tinny guitar leaks from the headphones. The speaker glances over, hears it, and a small smile starts.
3-5s. Medium close-up of the speaker, turned three-quarters toward the listener, warm and a little shy: "I love the Smiths."
5-7s. Medium close-up of the listener, who did not hear. They slide one headphone cup off one ear, puzzled, and say "Sorry?"
7-11s. The speaker, louder, with a playful confident smile: "I said I love the Smiths." Then, eyes half closed, they hum a short wordless tune and sway a little.
11-13s. A soft bell. The doors open on a bright office hallway. The speaker gives a small sideways smile and steps out without looking back.
13-15s. The listener alone, from the chest up, frozen, eyes wide, mouth slightly open, as the doors slide shut. They whisper "Wow."
Audio: elevator hum, faint tinny guitar from the headphones as an original melody, two natural English voices with exact lip sync, the soft hum, the bell, and the doors. No other dialogue. No other music. No famous melody. No lyrics.
Only these two people. Nobody else in the elevator or the hallway. No text, no subtitles, no logos, no brand names, no watermark.
Prompt 2 — Couple
Same elevator, same lines, softer acting. Image 1 is you in the headphones. Image 2 is your partner or crush. They look a beat too long, lean in, wave on the way out, and leave you with a slow grin.
A photoreal 15-second vertical scene, filmed like a real movie, inside a small old office elevator. Warm dark wood panels, brass trim, a brass handrail, brushed-metal doors, soft warm overhead light, a quiet morning. Natural skin, natural color, light film grain, a warm cinematic grade. Not CGI. Not a cartoon. Both people stand side by side facing the doors, about an arm's length apart, framed from the waist up so both faces stay large, sharp, and well lit.
Image 1 is the listener and stays on the LEFT in every frame. Copy that person's face exactly: shape, eyes, eyebrows, nose, mouth, skin tone, age, hair color, hair length, hairstyle, and any beard, glasses, or freckles. Keep the clothes from Image 1. Add large plain over-ear headphones with no logo.
Image 2 is the crush and stays on the RIGHT in every frame. Copy that person's face and clothes the same way. They hold a thin stack of folders.
Never blend the two faces. Never swap their clothes. Each face stays the same from the first frame to the last.
Shots:
0-3s. Medium two-shot from the front. The elevator hums upward. The listener stares ahead and nods slightly. Faint tinny guitar leaks from the headphones. The crush glances over, hears it, and looks at the listener a beat too long.
3-5s. Medium close-up of the crush, leaning a little closer, soft flirty smile: "I love the Smiths."
5-7s. Medium close-up of the listener, who did not hear. They slide one headphone cup off one ear, puzzled, and say "Sorry?"
7-11s. The crush, louder, holding the listener's eyes with a teasing smile: "I said I love the Smiths." Then they hum a short wordless tune, eyes still on the listener.
11-13s. A soft bell. The doors open on a bright office hallway. The crush steps out, looks back over one shoulder, and gives a small smile and a tiny wave.
13-15s. The listener alone, from the chest up, frozen, then a slow smitten grin. One hand comes up to the chest as the doors slide shut. They whisper "Wow."
Audio: elevator hum, faint tinny guitar from the headphones as an original melody, two natural English voices with exact lip sync, the soft hum, the bell, and the doors. No other dialogue. No other music. No famous melody. No lyrics.
Only these two people. Nobody else in the elevator or the hallway. No text, no subtitles, no logos, no brand names, no watermark.
Prompt 3 — You and your pet
Image 1 is the pet. Image 2 is you. The pet sits on a low bench so its head reaches your chest, wears tiny headphones, tilts its head, and makes one puzzled sound. It never talks. You say both lines and walk out.
A photoreal 15-second vertical scene, filmed like a real movie, inside a small old office elevator. Warm dark wood panels, brass trim, a brass handrail, brushed-metal doors, soft warm overhead light, a quiet morning. Natural skin, natural fur, natural color, light film grain, a warm cinematic grade. Not CGI. Not a cartoon. The pet and the person are side by side facing the doors, about an arm's length apart, framed from the waist up so both faces stay large, sharp, and well lit.
Image 1 is the listener and stays on the LEFT in every frame. It is the pet from Image 1, sitting upright on a small wooden bench so its head is about at the person's chest. Keep the same species, breed, face, eyes, ears, muzzle, fur color, fur length, and markings. Keep a natural animal body. Do not flatten it, stretch it, or turn it into a person. Add small plain over-ear headphones sized for its head, with no logo.
Image 2 is the speaker and stays on the RIGHT in every frame, standing. Copy that person's face exactly: shape, eyes, eyebrows, nose, mouth, skin tone, age, hair, and any beard, glasses, or freckles. Keep the clothes from Image 2. They hold a thin stack of folders.
Never blend the pet and the person. Never swap their looks. Each stays recognizable from the first frame to the last.
Shots:
0-3s. Medium two-shot from the front. The elevator hums upward. The pet sits still and stares ahead. Faint tinny guitar leaks from its headphones. The person glances over, hears it, and a small smile starts.
3-5s. Medium close-up of the person, turned three-quarters toward the pet, warm and a little shy: "I love the Smiths."
5-7s. Close-up of the pet, who did not hear. It tilts its head, one headphone cup slips off one ear, and it makes one small puzzled sound of its own kind.
7-11s. The person, louder, with a playful confident smile: "I said I love the Smiths." Then, eyes half closed, they hum a short wordless tune and sway a little.
11-13s. A soft bell. The doors open on a bright office hallway. The person gives a small sideways smile and steps out without looking back.
13-15s. The pet alone on the bench, frozen, eyes wide, ears up, staring as the doors slide shut.
Audio: elevator hum, faint tinny guitar from the headphones as an original melody, the person's natural English voice with exact lip sync, the pet's one small sound, the soft hum, the bell, and the doors. The pet never talks. No other dialogue. No other music. No famous melody. No lyrics.
Only this pet and this person. Nobody else in the elevator or the hallway. No text, no subtitles, no logos, no brand names, no watermark.
A sitting, front-facing pet photo works better than a photo from above. The whole head has to be visible.
Prompt 4 — Widescreen, no reference clip
Use this when you want the movie’s 16:9 frame but your tool only takes photos, not a source video. Change the project setting to 16:9.
A photoreal 15-second widescreen 16:9 scene inside a small old office elevator. Warm dark wood panels, brass trim, a brass handrail, brushed-metal doors, soft warm overhead light. Natural skin, natural color, light film grain, a warm cinematic grade. Not CGI. Not a cartoon. One steady medium two-shot from the front, waist up, both faces large and sharp. No zoom, no whip pans, no extra cuts unless noted.
Image 1 is the listener on the LEFT in every frame. Copy the face, hair, and clothes from Image 1. Add plain over-ear headphones with no logo.
Image 2 is the speaker on the RIGHT in every frame. Copy the face, hair, and clothes from Image 2. They hold a thin stack of folders.
Never blend the faces. Never swap clothes. Never add a third person.
0-3s. The elevator hums. The listener nods slightly. Faint tinny guitar leaks from the headphones. The speaker glances over and smiles.
3-5s. The speaker turns three-quarters and says "I love the Smiths."
5-7s. The listener slides one cup off one ear and says "Sorry?"
7-11s. The speaker, a little louder, smiling: "I said I love the Smiths." Then they hum a short wordless tune.
11-13s. A soft bell. The doors open. The speaker steps out into a bright hallway without looking back.
13-15s. The listener stays, frozen, eyes wide, and whispers "Wow." as the doors slide shut.
Audio: elevator hum, an original faint tinny guitar in the headphones, natural English lip sync, the hum, the bell, and the doors. No famous melody. No lyrics. No other dialogue. No text, subtitles, logos, or watermark.
Prompt 5 — Shot-for-shot, only with a reference video
Some tools have a video slot plus two image slots. This prompt keeps the reference clip’s camera, cuts, and timing, and replaces only the two people.
Only upload a reference video you have the right to use. The prompt also refuses a famous melody on purpose. General video models are bad at licensed songs, and the spoken lines are what make the clip readable. If you want the trend audio, add that sound afterward in the app where you post.
Set 16:9, 15 seconds. Image 1 replaces the person on the left. Image 2 replaces the person on the right.
Recreate the reference video shot for shot. Same elevator, same camera positions, framing, cuts, timing, and mouth timing. The only change: the two people are now the people in Image 1 and Image 2.
Lock the source camera and room. Keep the wall color, door color, reflections, exposure, and lighting from the reference in every shot. Keep the same lens, crop, and subject size. Do not zoom. Do not reframe. Do not invent a new camera move. Keep every cut on its original frame.
Image 1 replaces only the person on the LEFT, the one with the headphones. Image 2 replaces only the person on the RIGHT. Image 2's person is the one who walks out, at the same moment as the reference. The last shot shows only Image 1. Do not bring the other person back.
Copy each face exactly: shape, eyes, eyebrows, nose, mouth, skin, age, hair, glasses, facial hair. Rebuild the head and the clothes from that person's own photo. Do not leave the reference actor's hair, facial hair, or outfit unless the uploaded photo shows it. The listener may keep plain over-ear headphones as a prop. Never drift back to the faces in the reference. Never blend the two uploaded faces.
Match the reference poses, hands, gaze, head turns, and walk. Mouths move only on that person's own line:
0-4.7s. Front two-shot. The person on the right, smiling: "I love the Smiths." The person on the left slides the headphones down: "Sorry?" The person on the right: "I said I love the Smiths."
4.7-11s. Hard cut, same two-shot. The person on the right says "Yeah.", faces forward, hums a short wordless phrase with a smile, then says "I love them." The person on the left stares, stunned.
11-15s. Hard cut to a wider view. The doors open, the person on the right walks out, the person on the left stays alone and mutters two quiet words.
Photoreal live action. Stable faces and outfits. No extra people, no swapped faces, no merged bodies, no extra limbs. No other dialogue. No background music. No famous melody. No lyrics. No text, captions, logos, or watermark.
Name people by seat, left and right. If a template still says “man” or “woman,” delete it, or the model will try to reshape a face to match that gender.
Short I Love the Smiths AI prompt
Use this when the box cuts you off around 800–1,000 characters.
15-second vertical photoreal elevator scene, warm wood panels, brass rail, soft overhead light, film grain, not CGI. Waist-up two-shot, faces large.
Image 1 = LEFT listener in every frame. Exact face, hair, and clothes from Image 1, plus plain over-ear headphones, no logo.
Image 2 = RIGHT speaker in every frame. Exact face, hair, and clothes from Image 2, holding folders.
Never blend faces, never swap clothes, no extra people.
0-3s. Listener nods to faint tinny guitar leaking from the headphones. Speaker glances over and smiles.
3-5s. Speaker: "I love the Smiths."
5-7s. Listener slides one cup off one ear: "Sorry?"
7-11s. Speaker, louder, smiling: "I said I love the Smiths." Then a short wordless hum.
11-13s. Bell, doors open, speaker steps out without looking back.
13-15s. Listener alone, eyes wide, whispers "Wow." as the doors shut.
Audio: elevator hum, original tinny guitar, English lip sync, hum, bell, doors. No famous melody, no lyrics, no other dialogue, no text, no watermark.
If the tool only does 10 seconds
Replace the shot list with this and set the duration to 10.
0-2s. Two-shot. Listener nods. Faint guitar leaks from the headphones. Speaker smiles.
2-3.5s. Speaker: "I love the Smiths."
3.5-5s. Listener lifts one cup: "Sorry?"
5-7.5s. Speaker: "I said I love the Smiths." Short wordless hum.
7.5-8.5s. Bell. Doors open. Speaker steps out.
8.5-10s. Listener alone, eyes wide, whispers "Wow."
What to change, and what to leave
Change the frame if you need to. Replace “vertical” with “widescreen 16:9” and set the tool to match. Do not leave the word “vertical” in a 16:9 project.
Change the tone with a few words, not a new plot. “Playful, confident smile” can become “deadpan” or “trying not to laugh.” Do not add a kiss, a second location, or a third friend.
Leave these lines in:
- Image 1 stays on the left. Image 2 stays on the right.
- Never blend the faces. Never swap the clothes.
- The three spoken beats: “I love the Smiths.” / “Sorry?” / “I said I love the Smiths.”
- “No famous melody. No lyrics.”
That last line is deliberate. The prompt asks for an original thin guitar tone inside the headphones, plus the spoken lines. It does not ask the model to sing a real song. Add any trending sound later, in the editor you post from.
Negative prompt
Paste this into a separate negative box if the tool has one.
blurry face, blended faces, swapped clothes, swapped roles, extra people, third person, watermark, subtitles, captions, logos, cartoon, CGI, plastic skin, distorted hands, extra fingers, famous song, lyrics, headphone logo, morphing identity
When the prompt misses
| What you got | What to fix |
|---|---|
| The wrong person is in the headphones | Swap the uploads. Image 1 is always the listener. |
| Faces melt together | One person per image. Crop the group photo. |
| A face changes after the head turns | Use a sharper, front-facing photo with both eyes visible. |
| It looks filtered or waxy | Remove beauty filters from the source photo. |
| Words appear on screen | Put “no text, no subtitles” back. It was cut off. |
| It hums a real song or shows lyrics | Delete the song title. Keep “original melody” and “no lyrics.” |
| The speaker never leaves | The tool was shorter than 15s and dropped the last shots. Use the 10-second list. |
| Hands go wrong on the headphones | Keep the gesture to one cup, one ear. Do not add “takes the headphones apart.” |
Left listens. Right speaks. Each run is a new take. The elevator color can drift. The seats should not.
Questions
What is the I Love the Smiths AI prompt?
A 15-second image-to-video prompt. Two reference photos, listener on the left, speaker on the right, three spoken lines, then the speaker walks out.
Which prompt should I use?
Friends: prompt 1. A crush or partner: prompt 2. A pet in the headphones: prompt 3. A 16:9 frame and no source video: prompt 4. A tool with a legal motion reference: prompt 5.
Do I need headphones in my photo?
No. The prompt adds plain over-ear headphones to Image 1.
Can I name the real song in the prompt?
You can, and the clip usually gets worse. Keep the original-melody line.
What if my tool has no image slots, only text?
The prompt cannot lock a likeness from nothing. Use a tool that accepts reference images.
Create Your I Love the Smiths AI Video
Turn photos of yourself, your partner, or friends into the iconic elevator scene from (500) Days of Summer. Upload your photos and bring the viral trend to life!
Create Your Video