Skip to content

Open source

We publish the prompts and settings behind AI videos made with Sera. Pick a scene you like and make your own with the same settings.

Create a video

0–4sWhite dragon over the city

Medium-wide shot from behind the rider. He stands alone at the edge of a skyscraper rooftop, hands in his pockets, looking out over the city, the large black S logo on the back of his t-shirt clearly visible.

Hard cut. Wide shot from a neighboring tower. The rider is now seated on the base of the dragon's neck as it dives off the rooftop and plunges between the glass skyscrapers…

Close-up on the rider's face, sunset light glinting on his glasses and pearl necklace, curly hair whipping in the wind, shouting with pure exhilaration…

Extreme wide shot. The dragon bursts up through a layer of golden clouds above the skyline and rises into the open sky, swimming through the air with slow powerful undulations, legs tucked against its body…

Prompts

0:00 / 0:20

Trending

Trump and Netanyahu dancing in the street

Model
Genjutsu
Resolution
720p
Length
20s
Aspect ratio
Auto
Sound
Yes

Prompt

Use @Video 1 as the complete source video. Replace ONLY the two main men in @Video 1: CHARACTER 1: Replace the first man speaking in the opening scene with the exact man from @Image 1. @Image 1 is the sole authority for his face, complete head geometry, skin, beard, hairstyle, black fedora, body proportions and wardrobe. Keep his exact black blazer, white dress shirt, blue tie, gray trousers, black shoes and black fedora from @Image 1. CHARACTER 2: Replace the man positioned opposite Character 1 with the exact man from @Image 2. @Image 2 is the sole authority for his face, red hair, glasses, body proportions and wardrobe. Keep his exact brown vintage three-piece suit, white shirt, bow tie, cuffed trousers and brown dress shoes from @Image 2. VIDEO LOCK: Recreate exactly: same environment, location, background, set dressing, lighting, camera angles, framing, lens perspective, cuts, timing, blocking, movements and interactions. Transfer the original performances to the corresponding replacement characters, including speaking timing, lip movement, facial expressions, gestures, posture, eye lines and physical movement. The two replacement identities must remain completely separate. Never blend, swap or morph their facial features, hair, wardrobe or body characteristics. Preserve each character's reference-image body proportions and silhouette throughout every frame and camera angle. Do NOT transfer the original actors' faces, hair, body characteristics or clothing. Remove ALL source-video overlays: text, captions, subtitles, logos, usernames, watermarks, social-media icons, stickers and interface graphics. Do not introduce new people, props, objects or environmental changes. Ultra-photorealistic cinematic realism, natural skin texture, realistic anatomy, accurate hands, stable facial identity, clean temporal consistency. @Video 1 = environment, cinematography and performance. @Image 1 = Character 1 identity + wardrobe. @Image 2 = Character 2 identity + wardrobe.

Reference videos and images