🎁 PROMPT: STUDIO QUALITY TANPA STUDIO (GRATIS)

Makasih udah nonton & komen! Ini prompt yang gw pake di video.

═══════════════════════════════

BAHAN YANG LO BUTUHIN

1. Video ngomong ke kamera, MAKSIMAL 10 detik, 1 angle doang

2. 1 gambar background yang lo suka (opsional — kalau gak ada, background asli lo dipake tetep bisa)

3. Akun Google Flow (fitur Omni Flash), buka di labs.google/fx/tools/flow

═══════════════════════════════

LANGKAH-LANGKAH

1. Buka Google Flow → New project → pilih Video

2. Upload video lo

3. Upload gambar background (kalau ada)

4. Paste prompt di bawah, ganti bagian timestamp sesuai durasi video lo

5. Kirim, tunggu → jadi

═══════════════════════════════

PROMPT GENERAL (TEMPLATE)

Ganti angka detik di "Camera plan" sesuai panjang video lo. Contoh: video 10 detik → bagi 5 (0-2 / 2-4 / 4-6 / 6-8 / 8-10).

Keep the speaker exactly the same as in the source video. Their face, facial structure, skin details, hairstyle, clothing, body shape, age, and overall identity must remain unchanged from the first frame to the last. The person must be instantly recognizable as the original subject in every single frame.

Retain the original audio exactly as recorded. Do not alter the voice, dialogue, speaking rhythm, pauses, lip synchronization, expressions, or natural delivery. The performance must match the source footage precisely.

Use the provided reference image only as inspiration for the setting, mood, and visual atmosphere. Recreate a similar environment while keeping it realistic and cohesive. The new background must feel naturally integrated with the speaker, as if the footage was originally recorded in that location. Match the perspective and eye-line of the source footage so the speaker sits correctly within the space.

Illuminate the subject with soft, professional front-facing lighting. The face must remain bright, clear, and easy to see at all times, with balanced exposure and natural-looking shadows. Background lighting should match the direction and colour temperature of the light falling on the speaker.

Camera movement is not allowed. Instead, create the impression that multiple cameras were filming the scene simultaneously. Switch between viewpoints using clean, instantaneous cuts only. Avoid zooms, pans, morphs, crossfades, motion interpolation, or any transition effects.

Camera plan:

- At {X}s to {X}s:

Begin with a standard front-facing shot. The subject speaks naturally while maintaining the exact original performance.

- At {X}s to {X}s:

Instantly cut to a left-side perspective. The speaker continues talking naturally without any changes to timing, delivery, or expression.

- At {X}s to {X}s:

Cut to a camera positioned slightly above eye level, looking down at the speaker from a subtle elevated angle.

- At {X}s to {X}s:

Cut to a higher overhead three-quarter viewpoint from above. Keep the subject's proportions, appearance, and behaviour completely consistent.

- At {X}s to {X}s:

Cut back to the original front-facing composition, matching the opening shot as closely as possible.

Requirements:

- The speaker's identity must remain perfectly consistent.

- Facial features must not be altered or regenerated.

- Clothing, hairstyle, and body proportions must stay unchanged.

- Audio must remain identical to the source recording.

- Dialogue, pacing, and lip sync must remain identical to the source recording.

- The environment must reflect the style of the provided background reference image.

- Maintain clear, flattering front lighting throughout the video.

- Camera angle changes must occur only through hard cuts.

- No transition effects of any kind.

- Hold each angle for at least 2 seconds — never cut mid-word.

- Keep the speaker's framing and head size consistent across all angles.

- The final result should look as though several professional cameras captured the same uninterrupted moment from different viewpoints.

- Prioritize identity consistency and realism over cinematic effects or stylization.

═══════════════════════════════

CONTOH JADI — KASUS PERSIS DI VIDEO

Yang ini tinggal copy semua, gak usah diedit.

Keep the speaker exactly the same as in the source video. Their face, facial structure, skin details, hairstyle, clothing, body shape, age, and overall identity must remain unchanged from the first frame to the last. The person must be instantly recognizable as the original subject in every single frame.

Retain the original audio exactly as recorded. Do not alter the voice, dialogue, speaking rhythm, pauses, lip synchronization, expressions, or natural delivery. The performance must match the source footage precisely.

Use the provided reference image only as inspiration for the setting, mood, and visual atmosphere. Recreate a similar environment while keeping it realistic and cohesive. The new background must feel naturally integrated with the speaker, as if the footage was originally recorded in that location. Match the perspective and eye-line of the source footage so the speaker sits correctly within the space.

Illuminate the subject with soft, professional front-facing lighting. The face must remain bright, clear, and easy to see at all times, with balanced exposure and natural-looking shadows. Background lighting should match the direction and colour temperature of the light falling on the speaker.

Camera movement is not allowed. Instead, create the impression that multiple cameras were filming the scene simultaneously. Switch between viewpoints using clean, instantaneous cuts only. Avoid zooms, pans, morphs, crossfades, motion interpolation, or any transition effects.

Camera plan:

- At 0s to 3.76s:

Begin with a standard front-facing shot. The subject speaks naturally while maintaining the exact original performance.

- At 3.76s to 6.86s:

Instantly cut to a left-side perspective. The speaker continues talking naturally without any changes to timing, delivery, or expression.

- At 6.86s to 9.66s:

Cut to a camera positioned slightly above eye level, looking down at the speaker from a subtle elevated angle.

Requirements:

- The speaker's identity must remain perfectly consistent.

- Facial features must not be altered or regenerated.

- Clothing, hairstyle, and body proportions must stay unchanged.

- Audio must remain identical to the source recording.

- Dialogue, pacing, and lip sync must remain identical to the source recording.

- The environment must reflect the style of the provided background reference image.

- Maintain clear, flattering front lighting throughout the video.

- Camera angle changes must occur only through hard cuts.

- No transition effects of any kind.

- Hold each angle for at least 2 seconds — never cut mid-word.

- Keep the speaker's framing and head size consistent across all angles.

- The final result should look as though several professional cameras captured the same uninterrupted moment from different viewpoints.

- Prioritize identity consistency and realism over cinematic effects or stylization.

═══════════════════════════════

MAU GANTI ANGLE LAIN?

- Angle dari samping kanan → "Instantly cut to a right-side perspective."

- Angle deket banget (close-up) → "Cut to a close-up shot, framed from the shoulders up."

- Angle dari atas nunduk → "Cut to a high overhead three-quarter viewpoint looking down."

- Angle jarak jauh (wide) → "Cut to a wide shot showing the full body and more of the background."

═══════════════════════════════

TIPS

Kalau video lo pake gerak tangan aktif banget, kadang generate-nya gagal — tinggal generate ulang aja, biasanya lolos di percobaan berikutnya. Video sumber jangan lebih dari 10 detik, kalau lebih potong dulu.

═══════════════════════════════

Segini doang triknya. Kalau mau belajar trik AI kayak gini terus, follow