De un video grabado en tu sala a un keynote de seis cámaras

Lo que prometí en el video

El prompt completo, listo para llenar

Es el mismo que usé, convertido en plantilla. Lo que está entre [corchetes] es tuyo: qué presentas, tus frases y en qué segundo las dices. Va en inglés porque así lo sigue mejor el modelo; tus frases van en tu idioma, entre comillas.

Generate a [17]-second vertical 9:16 video in a single generation: a live technology keynote, filmed like a real broadcast of a product launch, with six camera angles joined by hard cuts. One presenter on stage introduces [WHAT YOU PRESENT: a fitness app], speaking [YOUR LANGUAGE: Spanish (Mexico)] to a full theater. Real-time speed throughout.

REFERENCES — each one has a single job:
- @Image1 and @Image2: the presenter's real face (front and three-quarter). Match it exactly in every shot and from every angle: same features, same [YOUR HAIR: full head of dark curly hair], same [YOUR FACIAL HAIR: light stubble], same age, same healthy skin. Never bald, never a full beard, never grey hair, never older, never a different person. Do not beautify, slim or smooth the face.
- @Image3: the presenter in costume, close: the face wearing the glasses, the black mock-neck sweater, the tiny black lavalier mic clipped at the collar, the stage light.
- @Image4: the complete costume, head to toe, on this stage: black fine-knit mock-neck sweater, light-wash straight blue jeans, plain grey sneakers, the round rimless glasses. Identical in every shot.
- @Image5: the theater seen from the audience: matte black stage, the giant screen behind as a deep navy field with soft beams of light, the cool-white follow-spot pool at center stage, two low black floor monitors at the stage edge, thin haze, silhouetted audience heads in the foreground.
- @Image6: the reverse angle, from upstage behind the presenter, looking out at the seated audience.
- @Image7: the phone and how it is held: a plain black smartphone whose screen is flat, evenly lit chroma-key green, with the black bezel and all four corners visible.
- @Image8: the exact eyeglasses: small, perfectly round, rimless, thin silver bridge and temples. Always on the face, never changing shape, never tinted.
- @Video1: the presenter's performance, voice and timing. Follow it beat by beat: [YOUR ACTIONS IN ORDER: walking on stage, the left hand resting in the left front pocket of the jeans, the right hand taking the phone out of the right front pocket, raising it beside the face, lowering it and sweeping the arm out to present the giant screen]. Lip-sync exactly to the [YOUR LANGUAGE: Spanish] speech in the audio of @Video1. Take only body language, voice and timing from it; ignore its clothing, [YOUR ROOM: white room, framed picture, tiled floor], daylight and static camera. None of these ever appear.

LOOK: an unedited live broadcast of a tech keynote, shot on broadcast cameras with long zoom lenses. Real theatrical lighting: one soft-edged cool-white follow spot from the balcony, a dim blue wash, the navy giant screen rim-lighting the presenter, thin haze. Deep blacks, natural warm skin tones, natural motion blur, mild sensor noise. No cinematic color grade, no lens flares. The presenter speaks to the audience with the eyeline just past the lens, never into it, eyes open with natural blinks.

TIMELINE — hard cuts exactly at [2.3, 5.8, 8.6, 10.6 and 13.8] s; no other cuts, no dissolves:

0–[2.3] s · SHOT 1 · WIDE FROM THE AUDIENCE, framed like @Image5. Slow push-in from the center aisle past the silhouetted heads. The presenter walks in from frame right into the follow-spot pool at center stage, left hand in the jeans pocket, and says, warm and confident: "[YOUR LINE 1]" ([0.4–1.7] s). Low audience murmur.

[2.3–5.8] s · SHOT 2 · MEDIUM SHOT, LOW ANGLE from the front edge of the stage, slightly left, the navy giant screen behind the head; gentle drift to the right. With a knowing half-smile, the right hand gesturing: "[YOUR LINE 2]" ([2.6–5.0] s, playful). [AUDIENCE REACTION: On the last word the audience laughs off-screen; the presenter holds a straight face for a beat.]

[5.8–8.6] s · SHOT 3 · MEDIUM CLOSE-UP, FRONTAL, eye level, framed like @Image3: long lens, shallow depth of field, very slow push-in. "[YOUR LINE 3]" ([5.9–8.1] s), [YOUR GESTURE: the right palm opening upward at chest height].

[8.6–10.6] s · SHOT 4 · REVERSE ANGLE from upstage behind the presenter, framed like @Image6: back of the head and shoulder in the foreground, the silver temple of the glasses catching the light, the seated audience beyond, smiling. Slow lateral move. "[YOUR LINE 4]" ([9.2–9.7] s), [YOUR GESTURE: one finger raised].

[10.6–13.8] s · SHOT 5 · ONE CONTINUOUS MOVE, no cut. It starts as a close-up at hip height: on "[YOUR LINE 5]" ([10.8–11.4] s) the right hand slides into the right front pocket of the jeans and pulls out the phone from @Image7. The camera tilts up and rises with the phone to a medium close-up as the phone is held up beside the face at shoulder height, screen toward the camera. From [11.6] to [13.4] s the screen is fully visible: flat, evenly lit chroma-key green with nothing on it, no reflections, the black bezel and all four rounded corners visible and sharp, fingers only on the edges. A quick glance at the phone, back to the audience with a confident smile, the phone lowers, and softly: "[YOUR LINE 6]" ([12.9–13.4] s).

[13.8–17.0] s · SHOT 6 · THE REVEAL. Wide shot from the audience side, high on a crane, looking toward the stage and the giant screen. On "[YOUR LAST LINE]" ([14.4–14.9] s) the presenter turns to face the giant screen, back to the camera, sweeps the right arm up to present it, then walks out of frame to the right. The camera keeps pushing forward and up, straight into the giant screen, while its navy glow fades to pure black. From [16.2] s the frame is completely black until [17.0] s.

SOUND: only the presenter's own voice from @Video1, the same words with the same timing, no other words and no narrator, with the natural reverb of a large theater; a low audience murmur at the start, [a burst of laughter at 5.0 s], an expectant hush from [11] s, warm applause rising at [14.9] s and fading under the black. No music.

CONTINUITY IN EVERY SHOT: exactly one presenter on stage; the same face, hair, glasses, sweater, lavalier mic, jeans and sneakers; the phone only in the right hand and only from [10.8] s on; five fingers on each hand; the giant screen is navy (black at the end), never green, never with text or images; the audience stays seated.

PROHIBITED: subtitles, captions or any on-screen text; broadcast graphics such as lower thirds, name titles, channel logos, tickers or timecode; logos, brand marks or watermarks on anything, including the phone; anything readable on any screen; text, logos or product images on the giant screen; a headset microphone; a second phone; green light on the face; extra people on stage; morphing faces or hands; slow motion or speed ramps; lens flares, neon or light leaks; the [YOUR ROOM: white room, framed picture or tiled floor] from @Video1.

Así lo llené yo

Mis frases y sus segundos, tal cual van en la TIMELINE. Si tu guion tiene otro ritmo, cambias los segundos y los cortes, no la estructura.

Qué hace cada bloque

Generate a 17-second vertical 9:16 video in a single generation: a live technology keynote, filmed like a real broadcast of a product launch, with six camera angles joined by hard cuts. One presenter on stage introduces a fitness app, speaking Spanish (Mexico) to a full theater. Real-time speed throughout.

REFERENCES — each one has a single job:
- @Image1 and @Image2: the presenter's real face (front and three-quarter). Match it exactly in every shot and from every angle: same features, same full head of dark curly hair, same light stubble, same age, same healthy skin. Never bald, never a full beard, never grey hair, never older, never a different person. Do not beautify, slim or smooth the face.
- @Image3: the presenter in costume, close: the face wearing the glasses, the black mock-neck sweater, the tiny black lavalier mic clipped at the collar, the stage light.
- @Image4: the complete costume, head to toe, on this stage: black fine-knit mock-neck sweater, light-wash straight blue jeans, plain grey sneakers, the round rimless glasses. Identical in every shot.
- @Image5: the theater seen from the audience: matte black stage, the giant screen behind as a deep navy field with soft beams of light, the cool-white follow-spot pool at center stage, two low black floor monitors at the stage edge, thin haze, silhouetted audience heads in the foreground.
- @Image6: the reverse angle, from upstage behind the presenter, looking out at the seated audience.
- @Image7: the phone and how it is held: a plain black smartphone whose screen is flat, evenly lit chroma-key green, with the black bezel and all four corners visible.
- @Image8: the exact eyeglasses: small, perfectly round, rimless, thin silver bridge and temples. Always on the face, never changing shape, never tinted.
- @Video1: the presenter's performance, voice and timing. Follow it beat by beat: walking on stage, every gesture, the left hand resting in the left front pocket of the jeans, the right hand taking the phone out of the right front pocket, raising it beside the face, lowering it and sweeping the arm out to present the giant screen. Lip-sync exactly to the Spanish speech in the audio of @Video1. Take only body language, voice and timing from it; ignore its clothing, white room, framed picture, tiled floor, daylight and static camera. None of these ever appear.

LOOK: an unedited live broadcast of a tech keynote, shot on broadcast cameras with long zoom lenses. Real theatrical lighting: one soft-edged cool-white follow spot from the balcony, a dim blue wash, the navy giant screen rim-lighting the presenter, thin haze. Deep blacks, natural warm skin tones, natural motion blur, mild sensor noise. No cinematic color grade, no lens flares. The presenter speaks to the audience with the eyeline just past the lens, never into it, eyes open with natural blinks.

TIMELINE — hard cuts exactly at 2.3, 5.8, 8.6, 10.6 and 13.8 s; no other cuts, no dissolves:

0–2.3 s · SHOT 1 · WIDE FROM THE AUDIENCE, framed like @Image5. Slow push-in from the center aisle past the silhouetted heads. The presenter walks in from frame right into the follow-spot pool at center stage, left hand in the jeans pocket, and says, warm and confident: "Todos conocemos a ese coach…" (0.4–1.7 s). Low audience murmur.

2.3–5.8 s · SHOT 2 · MEDIUM SHOT, LOW ANGLE from the front edge of the stage, slightly left, the navy giant screen behind the head; gentle drift to the right. With a knowing half-smile, the right hand gesturing: "…ese que no te hará caso, a menos que seas la linda del gimnasio." (2.6–5.0 s, playful). On the last word the audience laughs off-screen; the presenter holds a straight face for a beat.

5.8–8.6 s · SHOT 3 · MEDIUM CLOSE-UP, FRONTAL, eye level, framed like @Image3: long lens, shallow depth of field, very slow push-in. "Imagínate tenerlo…" (5.9–6.7 s), a beat, then "…en la palma de tu mano," (7.3–8.1 s), the right palm opening upward at chest height on "palma".

8.6–10.6 s · SHOT 4 · REVERSE ANGLE from upstage behind the presenter, framed like @Image6: back of the head and shoulder in the foreground, the silver temple of the glasses catching the light, the seated audience beyond, smiling. Slow lateral move. "o mejor aún," (9.2–9.7 s), one finger raised.

10.6–13.8 s · SHOT 5 · ONE CONTINUOUS MOVE, no cut. It starts as a close-up at hip height: on "en tu bolsillo." (10.8–11.4 s) the right hand slides into the right front pocket of the jeans and pulls out the phone from @Image7. The camera tilts up and rises with the phone to a medium close-up as the phone is held up beside the face at shoulder height, screen toward the camera. From 11.6 to 13.4 s the screen is fully visible: flat, evenly lit chroma-key green with nothing on it, no reflections, the black bezel and all four rounded corners visible and sharp, fingers only on the edges. A quick glance at the phone, back to the audience with a confident smile, the phone lowers, and softly: "Les presento…" (12.9–13.4 s).

13.8–17.0 s · SHOT 6 · THE REVEAL. Wide shot from the audience side, high on a crane, looking toward the stage and the giant screen. On "…a Entrena." (14.4–14.9 s) the presenter turns to face the giant screen, back to the camera, sweeps the right arm up to present it, then walks out of frame to the right. The camera keeps pushing forward and up, straight into the giant screen, while its navy glow fades to pure black. From 16.2 s the frame is completely black until 17.0 s.

SOUND: only the presenter's own voice from @Video1, the same words with the same timing, no other words and no narrator, with the natural reverb of a large theater; a low audience murmur at the start, a burst of laughter at 5.0 s, an expectant hush from 11 s, warm applause rising at 14.9 s and fading under the black. No music.

CONTINUITY IN EVERY SHOT: exactly one presenter on stage; the same face, hair, glasses, sweater, lavalier mic, jeans and sneakers; the phone only in the right hand and only from 10.8 s on; five fingers on each hand; the giant screen is navy (black at the end), never green, never with text or images; the audience stays seated.

PROHIBITED: subtitles, captions or any on-screen text; broadcast graphics such as lower thirds, name titles, channel logos, tickers or timecode; logos, brand marks or watermarks on anything, including the phone; anything readable on any screen; text, logos or product images on the giant screen; a headset microphone; a second phone; green light on the face; extra people on stage; morphing faces or hands; slow motion or speed ramps; lens flares, neon or light leaks; the white room, framed picture or tiled floor from @Video1.

Haz las tuyas

Seis imágenes, en este orden

Cada una usa a las anteriores como referencia, así que el orden importa. Antes de pasar a la siguiente, revisa su lista: una imagen que falla contamina todas las que siguen.

1. 

   Los lentes. Van primero

   Motor: Nano Banana 2.1 o GPT Imágenes 2.5 · 1K · 1:1
   Adjunta: nada

   Van primero porque las imágenes con tu cara copian estos lentes. Sale como @Image8.

   Redondos perfectos, sin aro alrededor del cristalPuente y patillas de plata delgadaNingún logo ni grabado

   Ver el prompt de los lentes

   Paso 1 · los lentesCopiar

   Scene: Product photograph on a seamless matte black surface, a soft overhead softbox plus a thin white edge light from behind, shot on a Sony A7 IV with a 90mm macro lens at f/8, three-quarter front view at table height. Square 1:1.
   Subject: A pair of small, perfectly round rimless eyeglasses with the temples open, resting on the surface. Clear lenses about 40 mm across, no tint and no rim around them, held only by a thin silver metal bridge and two thin silver temples fixed to each lens with tiny screws; small clear nose pads on thin silver arms; plain silver temple tips.
   Important details: One long soft reflection runs across each lens; a faint reflection of the glasses on the black surface; a little dust on the surface in the foreground, out of focus.
   Use case: Reference image so an AI video reproduces exactly this eyewear on a presenter in every shot.
   Constraints: No logo, engraving, text or watermark anywhere. No tinted or colored lenses, no thick frames, no tortoiseshell, no gold. Square 1:1.

2. 

   El escenario. Visto desde el público

   Motor: GPT Imágenes 2.5 o Nano Banana 2.1 · 1K · 9:16 (si no lo acepta, 2:3)
   Adjunta: nada

   Es el lugar de todo el video. Sale como @Image5.

   La pantalla gigante sin nada escrito ni dibujadoUn solo foco en el centro del escenarioPúblico de espaldas, gente normal; nada de láseres ni neón

   Ver el prompt del escenario

   Paso 2 · el escenarioCopiar

   Scene: An unedited frame grab from the live broadcast of a technology product keynote, taken from the center aisle about two-thirds of the way back in a darkened 600-seat theater, camera at seated eye level, Sony FX9 with a 24-70mm zoom at about 40mm. Natural broadcast exposure, no cinematic color grading. Vertical 9:16.
   Subject: The stage just before the presenter walks on. A wide stage with a matte black floor. A giant 16:9 LED screen fills the upper back wall; it is on but shows only a deep navy-to-black gradient with a faint, soft cool-blue glow along its lower edge: no text, no logo, no image. A single soft-edged cool-white follow spot draws an oval pool of light on the floor at center stage.
   Important details: The bottom quarter of the frame is the backs of audience heads and shoulders in silhouette, softly out of focus: ordinary people of different ages and haircuts, one head tilted, one person leaning over to whisper. A thin haze makes the follow-spot beam visible from the balcony above. Two low black confidence monitors sit at the front edge of the stage, angled toward the stage, their screens not visible. Small strips of white spike tape mark the floor near the light pool; a black cable runs along the stage lip. The sides of the auditorium fall into darkness. Deep but not crushed blacks, mild sensor noise in the shadows, no added sharpening.
   Use case: Location reference for an AI video. It must read instantly as a tech keynote: stage, giant screen, light pool and audience, with nothing competing.
   Constraints: No people on stage. No readable text, signs, logos or watermarks anywhere, screens included. No lens flares, no light rays other than the one haze beam, no neon, no LED strips, no lasers, no smoke clouds. Vertical 9:16.

3. 

   Tu retrato de keynote. La pieza más importante

   Motor: prueba en Nano Banana 2.1, la final en GPT Imágenes 2.5 · 1K · 9:16
   Adjunta, en orden: tus 3 fotos de cara, los lentes (paso 1) y el escenario (paso 2)

   De aquí salen tu cara, los lentes y el suéter de todas las tomas. Sale como @Image3. Saqué dos versiones y en una los lentes salieron ovalados: me quedé con la de lentes redondos.

   ¿Eres tú? Si la cara no se parece, no sigas: repite este pasoLos lentes son los del paso 1Cuello alto doblado una vez y mangas subidasBoca cerrada, sin sonrisa

   Ver el prompt del retrato

   Paso 3 · tu retratoCopiar

   REFERENCES (attached in this order): Image 1 — IDENTITY (front). Image 2 — IDENTITY (three-quarter). Image 3 — IDENTITY (three-quarter, other side). Image 4 — GLASSES (reproduce exactly). Image 5 — LOCATION (this stage, its giant navy screen and its light).
   Scene: An unedited frame grab from the live broadcast of the keynote in Image 5, shot from the front of the house on a Sony FX9 with a 70-200mm zoom at about 160mm, f/2.8, so the stage and the giant navy screen behind melt into soft dark-blue bokeh. Medium close-up from mid-chest up, the presenter slightly off-center to the right. Natural broadcast color, no cinematic color grading. Vertical 9:16.
   Subject: The person in Images 1-3 — those reference images ARE their face: copy it exactly, feature for feature, hair and facial hair included, at exactly the age and with exactly the same fresh, healthy skin shown in those photos. Do not age them up, do not add wrinkles, pockmarks, pitted or rough skin, exaggerated pores, under-eye bags or blemishes, do not slim or lighten the face, and do not average it toward a generic model. No description of their features is given on purpose; if the face is not recognisably them, the frame is wrong. IGNORE the clothing and the background in Images 1-3; the wardrobe and setting are defined here. The presenter is mid-talk at center stage, body turned a little toward camera left, eyes on the audience just past the lens, mouth closed between two words, calm and sure, a trace of amusement in the eyes but no smile. One hand rises into the bottom edge of the frame, palm up, slightly motion-blurred.
   Wardrobe: a black fine-gauge merino mock-neck sweater (a slim turtleneck folded once at the neck) with the sleeves pushed up to mid-forearm; the eyeglasses from Image 4 exactly (small, perfectly round rimless lenses, thin silver bridge and temples) sitting naturally on the nose, each lens holding a small reflection of the follow spot; a tiny black lavalier microphone clipped at the collar. No watch, no jewelry.
   Light: one soft-edged cool-white follow spot from high front-left (the balcony), the face correctly exposed, the right side of the face falling about one stop darker; a soft blue rim on the hair and shoulders from the giant screen behind; the black knit shows its texture where the spot hits and falls to near-black in shadow.
   Important details: a few curls out of place, a slight sheen on the forehead from the spot, the knit bunching a little at the pushed-up sleeves; the skin keeps its natural, healthy warm tone, never grey, ashy or sallow; fine grain over the whole image, not on the skin; the skin looks clean and healthy with natural texture, never over-sharpened, never waxy.
   Use case: Character reference for an AI video: the face, the glasses and the sweater must be unmistakable and identical in every later shot.
   Constraints: No readable text, signs, logos or watermarks anywhere. No smile, no teeth, no theatrical expression. Vertical 9:16.

4. 

   Cuerpo completo. En el escenario

   Motor: prueba en Nano Banana 2.1, la final en GPT Imágenes 2.5 · 1K · 9:16
   Adjunta, en orden: tus 3 fotos de cara, tu retrato (paso 3) y el escenario (paso 2)

   El vestuario de pies a cabeza y tu tamaño en el escenario. Sale como @Image4.

   La misma ropa y los mismos lentes que el retratoJeans claros y tenis grises, sin logosOcupas como un tercio del alto y el escenario es el del paso 2

   Ver el prompt del cuerpo completo

   Paso 4 · cuerpo completoCopiar

   REFERENCES (attached in this order): Image 1 — IDENTITY (front). Image 2 — IDENTITY (three-quarter). Image 3 — IDENTITY (three-quarter, other side). Image 4 — SESSION ANCHOR (the presenter, the glasses, the sweater and the light exactly as here). Image 5 — LOCATION (this exact stage, screen, light pool and audience, from the same camera position).
   Scene: The same live keynote broadcast as Image 4, from the wide camera at Image 5's position in the center aisle (Sony FX9, 24-70mm at about 40mm). The presenter stands full-body inside the follow-spot pool at center stage, about one third of the frame height, the giant navy screen filling the space above, the backs of audience heads in silhouette across the bottom. Natural broadcast color, no cinematic color grading. Vertical 9:16.
   Subject: The person in Images 1-3 — those reference images ARE their face: copy it exactly, feature for feature, hair and facial hair included, at exactly the age and with exactly the same fresh, healthy skin shown in those photos. Do not age them up, do not add wrinkles, pockmarks, pitted or rough skin, exaggerated pores, under-eye bags or blemishes, do not slim or lighten the face, and do not average it toward a generic model. No description of their features is given on purpose; if the face is not recognisably them, the frame is wrong. IGNORE the clothing and the background in Images 1-3; the wardrobe and setting are defined here. A relaxed stance with the weight on one leg, the left hand in the front pocket of the jeans, the right hand open at waist height mid-gesture, looking out at the audience, mouth closed.
   Wardrobe: exactly as Image 4 (black fine-gauge mock-neck merino sweater with the sleeves pushed up, the small round rimless glasses, a tiny black lavalier mic), plus light-wash straight-leg blue jeans with a natural fade on the thighs and a slight break over the shoes, and plain grey suede-and-mesh running sneakers with white midsoles. No visible logos on any garment or shoe.
   Light: the cool-white follow spot from high front-left; the presenter's soft shadow falls on the black floor to the right; a thin blue rim from the screen on the shoulders and hair.
   Important details: white spike-tape marks near the feet, the two low confidence monitors at the stage lip, the haze beam above; the sweater wrinkles at the waist where the hand goes into the pocket; mild sensor noise in the dark areas; the skin keeps its natural, healthy warm tone, never grey, ashy or sallow; fine grain over the whole image, not on the skin.
   Use case: Full-body wardrobe and stage-scale reference for an AI video.
   Constraints: No readable text, signs, logos or watermarks anywhere, screens included. No extra people on stage. Vertical 9:16.

5. 

   El contraplano. Desde atrás de ti

   Motor: GPT Imágenes 2.5 o Nano Banana 2.1 · 1K · 9:16
   Adjunta, en orden: el cuerpo completo (paso 4) y el escenario (paso 2)

   La toma desde el escenario hacia el público. Sale como @Image6.

   No se te ve la cara: sólo nuca, oreja y la patilla de los lentesEl público se ve real: edades y ropa normalesNadie mira a cámara ni levanta el celularSin micrófono de diadema: a mí se me coló y pasó al video

   Ver el prompt del contraplano

   Paso 5 · el contraplanoCopiar

   REFERENCES (attached in this order): Image 1 — SESSION ANCHOR (the presenter's wardrobe, glasses and light). Image 2 — LOCATION (the same theater and stage, seen from the audience).
   Scene: The reverse angle of the same live keynote: an unedited frame grab from the broadcast camera standing upstage behind the presenter, looking out at the audience, Sony FX9 with a 24-70mm zoom at about 35mm, at shoulder height. Natural broadcast color, no cinematic color grading. Vertical 9:16.
   Subject: In the left foreground, the presenter seen from behind at three-quarter back: the back of the curly head, one ear, the thin silver temple of the round glasses, the shoulder and back of the black mock-neck sweater, edge-lit by the follow spot and softly out of focus. Beyond, the dark auditorium: rows of seated audience rising toward the back, faces dimly lit by spill from the stage, ordinary people of different ages and clothes, attentive, a few starting to laugh; nobody looks into the lens.
   Important details: the follow spot itself is a small, bright soft source in the balcony at the back; a thin haze hangs over the first rows; a broadcast camera on a tripod in the center aisle with a small red tally light; the stage edge and one low black confidence monitor at the bottom of the frame; aisle step lights as tiny warm dots; mild sensor noise.
   Use case: Reverse-angle location reference for an AI video of the same keynote.
   Constraints: The presenter's face is never shown. No readable text, signs, logos or watermarks anywhere. No phones held up and no screens in the audience. Vertical 9:16.

6. 

   El teléfono en verde. Para cambiar la pantalla después

   Motor: GPT Imágenes 2.5 · 1K · 9:16
   Adjunta: tu retrato (paso 3), para el tono de tu piel, la manga y la luz

   Cómo sostienes el teléfono y su pantalla en verde. Sale como @Image7.

   Verde plano de orilla a orilla, sin brillosSe ven las cuatro esquinas y el bisel negroNingún dedo encima de la pantalla y ningún logo en el teléfono

   Ver el prompt del teléfono

   Paso 6 · el teléfono en verdeCopiar

   REFERENCES (attached in this order): Image 1 — SESSION ANCHOR (the presenter's skin tone, the black sweater sleeve and the stage light).
   Scene: An insert shot of the same live keynote: an unedited frame grab from the broadcast camera at the front of the house, Sony FX9 with a 70-200mm zoom at 200mm, f/2.8, at shoulder height. Natural broadcast color. Vertical 9:16.
   Subject: The presenter's right hand holds a plain, modern black smartphone upright at shoulder height, screen facing the camera, turned about 10 degrees so its edge catches the light. The screen shows only a flat, perfectly even, matte chroma-key green (#00B140) from edge to edge: no interface, no icons, no text, no camera cutout, no reflection, no glare. A thin, even black bezel is visible on all four sides, with rounded corners, and all four corners are fully in frame and sharp. The fingers wrap around the side edges and the bottom, never over the screen. The pushed-up black merino sleeve shows at the wrist.
   Light: the cool-white follow spot from front-left on the hand and on the phone's edges; the green screen casts only a faint glow on the fingertips.
   Important details: behind the phone, out of focus, the edge of the presenter's shoulder in the black sweater and the soft navy bokeh of the giant screen; a thin highlight along the phone's metal edge; natural skin texture on the knuckles; the skin keeps its natural, healthy warm tone.
   Use case: Reference for an AI video and for screen replacement in post: the green screen must be flat, evenly lit and fully visible so it can be tracked and replaced.
   Constraints: No logo or brand mark on the phone, no readable text anywhere, no watermarks. Vertical 9:16.

Dónde y cuánto

Seedance 2.5, en Kie

Yo lo corrí en Kie, una plataforma que da acceso a varios modelos de video con un solo saldo de créditos. Ahí Seedance 2.5 aceptó mis fotos con cara. En Dreamina las fotos de caras reales se bloquean.

Con video de referencia te cobran los segundos que subes más los que genera: (17 + 17) × 95 = 3,230 créditos. Sin video la tarifa por segundo es más alta, pero sólo cuenta lo generado. Según la página de precios de Kie, un crédito cuesta unos US$0.005.

{
  "model": "bytedance/seedance-2-5",
  "input": {
    "prompt": "[EL PROMPT DE ARRIBA, YA LLENO]",
    "reference_image_urls": ["[URL DE LA FOTO 1]", "[… HASTA LA 8, EN ORDEN]"],
    "reference_video_urls": ["[URL DE TU VIDEO DE 720p]"],
    "resolution": "720p",
    "aspect_ratio": "9:16",
    "duration": 17,
    "generate_audio": true
  }
}

La post · opcional

La pantalla verde es para poner tu app después

Seedance no sabe cómo se ve tu app, y si se lo pides, inventa letras. Por eso el teléfono sale en verde plano: después se sigue y se cambia por la pantalla real.

Lo que dio Seedance

Después de la post

1. Sigue las cuatro esquinas. Contra el bisel negro

   Con cualquier editor que siga cuatro esquinas (corner pin o seguimiento planar), como After Effects o DaVinci Resolve. Por eso el prompt pide las cuatro esquinas siempre a la vista.

2. Recorta los dedos con el verde. Para que queden encima

   El verde plano sirve de llave: los dedos que pasan por la orilla se quedan delante de tu pantalla.

3. Pega tu pantalla real. Con la luz del plano

   Bájale un poco el brillo y dale el mismo desenfoque del cuadro. Una pantalla perfecta y nítida se ve pegada.