Jak generovat videa pomocí AI

Pomůžeš mi s přípravou promptu pro generování videa v modelu Veo3. Budeš vycházet z guidu, jak psát prompty na https://github.com/snubroot/Veo-3-Prompting-Guide. Chci vytvořit fotorealistické video o … Model bude vycházet ze startovní fotografie. Délka videa je přesně 8 vteřin. Nevymýšlej si. Pokud potřebuješ další informace, ptej se.

{ "version": „veo-3.1“, "duration": 8, "character_references": [ { "id": „PHOTO_OF_BLONDE_WOMAN“, "role": „main_subject“, "consistency": „high“ } ], "constraints": { "character_lock": „maintain exact facial features from PHOTO_OF_BLONDE_WOMAN reference - no morphing, no redesign, no drift in face structure, nose shape, lip fullness, or eye spacing“ } "scene": { "setting": „busy Japanese market street in Tokyo“, "time_of_day": „daytime“, "description": „Crowded market with Japanese shoppers, colorful stalls (fruit, vegetables, spices, fresh fish), Japanese price tags.“ }, "subject": { "role": „female travel vlogger“, "look": „blonde woman with specific facial features: oval face shape, medium-width nose with defined bridge, fuller lips (especially lower lip), soft jawline, light blue-green eyes, natural arched eyebrows, high cheekbones, European features matching reference photo exactly“ }, "action": „walking through market while holding camera in front of her at arm’s length, talking to audience“, "props": „small vlogging camera or smartphone visible in her hand in lower part of frame“ }, "camera": { "type": „selfie/vlog camera held by subject“, "movement": „camera position stays attached to subject’s hand; ONLY small natural handheld sway; DO NOT dolly or pull back“, "framing": „tight-medium selfie shot (head and shoulders), background market visible behind her“, "style": „travel vlog, natural handheld“ }, "audio_prompt": { "speech": { "language": „cs-CZ“, "text": „Čau, právě jsem v Tokiu na tradičním trhu, je tu neskutečný výběr – ryby, koření, čerstvé ovoce!“, "voice_style": „young female, energetic, cheerful, vlogger tone“, "lipsync": „tight“, "start_delay_seconds": 0.6 }, "ambience": „busy Japanese market ambience, Japanese voices, vendor calls, rustling bags“ }, "beats": [ { "time": „0.0-0.6s“, "description": „Selfie angle. Vlogger holds camera in front of her, arm slightly bent. Mouth CLOSED, friendly expression. Background shows moving market behind her as she walks forward.“ }, { "time": „0.6-3.5s“, "description": „She starts speaking: „Čau, právě jsem v Tokiu na tradičním trhu…“ Lipsync must match Czech speech. Camera stays fixed to her hand, only small handheld sway. Market passes behind her.“ }, { "time": „3.5-6.0s“, "description": „Continues: „…je tu neskutečný výběr – ryby, koření, čerstvé ovoce!“ Background briefly reveals fish stall / colorful produce. Still selfie angle, NO zooming out, NO pulling back.“ }, { "time": „6.0-8.0s“, "description": „She finishes speaking, smiles at camera, maybe quick small gesture with free hand. Keep camera at SAME distance and angle, still selfie style, only subtle handheld motion. Background continues to move as she walks.“ } ], "output": { "aspect_ratio": „16:9“, "quality": „cinematic“, "fps": 24 }, "constraints": { "start_with_closed_mouth": true, "face_consistency": „match PHOTO_OF_BLONDE_WOMAN“, "camera_distance": „constant, arm’s length selfie; do not dolly or pull back at any time“, "location_consistency": „same Tokyo market street“ } }

Create a 15-second realistic cinematic video in 16:9. Two-scene structure. No subtitles or on-screen text of any kind. Overall setting: - Stylish European café in Prague’s Karlín district. - Warm late-afternoon light, wooden tables, modern interior, slightly blurred background with a few other guests. - Two women sit at a small café table facing each other at a slight angle. - Brunette on the left, redhead on the right. - Brunette drinks cappuccino from a ceramic cup. Redhead drinks café latté from a tall glass. - Both women are around 35, attractive, well-dressed in stylish but different outfits and color palettes. - Natural café ambience: quiet background chatter and clinking dishes, but no intelligible background speech. CHARACTER RULES: - Only the brunette speaks full lines. - The redhead does NOT speak any full words; she only reacts with subtle facial expressions and a small nod. - No one laughs during the entire video. CAMERA STYLE: - Slightly dynamic throughout: gentle forward drift and subtle shoulder-level motion, like a natural handheld camera. - Depth of field: faces in clear focus, background softly blurred. SCENE 1 — FIRST SENTENCE + NOD REACTION (about 7 seconds) Framing: - Medium shot of both women at the table, slightly closer to the brunette, but both are clearly visible. - Camera slowly, gently moves closer to them. Actions: 1. At the beginning, both women briefly take a sip of their drinks. - The brunette takes a sip of her cappuccino and lowers the cup back to the saucer. - The redhead takes a sip of her latté and sets the glass down. 2. After putting her cup down, the brunette turns her head and looks directly at her friend (the redhead). - While maintaining eye contact with the redhead, the brunette clearly says in Czech, with accurate lip-sync: „Můj manžel zestárl!“ - Her facial expression is dry, slightly resigned, not amused. 3. Right after this first sentence: - The redhead reacts with a small, subtle nod, as if silently agreeing or saying “hm, fakt” - The redhead does not say any word out loud; this is a purely non-verbal reaction. Behavior constraints: - During this scene, only the brunette’s lips move in a way that looks like clear speech. - The redhead’s reaction is limited to a small nod and slight change in facial expression, with no articulated words. SCENE 2 — FULL PUNCHLINE + FINAL REACTION (about 8 seconds) Framing: - New shot from a slightly closer angle, still a medium shot focused more on the brunette’s upper body and face, but the redhead remains partially visible. - The camera continues a gentle inward movement, emphasizing the intimacy of the conversation. Actions: 1. At the start of this scene: - The redhead briefly glances at the brunette, then looks back at her own drink. - The brunette gently touches or adjusts her cappuccino cup, then raises her eyes to look at the redhead again. 2. While maintaining eye contact with the redhead, the brunette delivers the second part of the joke in Czech, with accurate lip-sync: „Nejdřív si prohlíží jídelní lístky a až potom servírku.“ - Tone: dry, slightly ironic, not exaggerated, no smile. - medium shot of the speaking woman 3. Immediately after she finishes speaking: - The redhead looks down into her latté glass, as if processing the joke with a slightly tired or resigned expression. - She can make a very subtle reaction, like a tiny exhale through the nose, a slight raise of one eyebrow, or a faint tightening of the lips, but she does NOT smile or laugh. - She does not respond verbally; no additional lines of dialogue. Behavior constraints: - Only the brunette moves her lips in a clearly speaking way for both Czech sentences. - The redhead never delivers any spoken line; she communicates only with nods, changes in gaze, and subtle facial expressions. - Neither woman laughs or breaks into a big smile; keep the mood dry and slightly cynical. GLOBAL CONSTRAINTS: - No subtitles, no captions, no on-screen text at all. - Maintain the same character appearances, outfits, hair, and café environment in both scenes for perfect continuity. - Keep the audio mix natural: Czech dialogue clearly audible, café ambience quiet and unobtrusive underneath.