[Guide] How We Made “You Look Real” Video with GenX Pro App

The story

A man stands at a mirror in a dark hotel suite, tying his bow tie. He is completely absorbed in himself.

A woman comes out of the shadows behind him, close enough to touch. She watches his reflection and says, quietly:

"You look real."

His hands stop.

He turns his head to her. Flat, unhurried, without inflection:

"I didn't ask."

He turns back to the mirror. Her face falls. Neither of them moves again.

What's The Generation Sequence for This?

Step 1 — The man

MODEL: GenX Pro -> Image → Nano Banana Pro.

Photorealistic cinematic portrait of a strikingly handsome man in his late
thirties, front-facing, head and shoulders, looking directly at camera. Dark
hair slicked back, sharp clean jawline, light evening stubble, cool grey-green
eyes, a fine scar through the left eyebrow. Wearing a black tuxedo with silk
peak lapels, crisp white dress shirt, black bow tie tied. Expression composed
and self-assured, mouth closed, a faint trace of arrogance. Warm practical
lamp light from camera left, soft rim light along the jaw, deep rich shadow
on the right side. Background: dark wood panelling and warm bokeh from a hotel
suite at night, heavily out of focus. 85mm lens, shallow depth of field,
high-end cinematic colour grade, deep blacks, subtle film grain. Real skin
texture. Expensive, magnetic, editorial portrait quality.

Why the scar. It's your continuity anchor. Across two video generations you need to know whether the face drifted, and "does this look like the same man" is slow and unreliable. "Is the scar still through his left eyebrow" takes half a second. Give your character one small, specific, checkable feature.

Step 2 — The woman

MODEL: GenX Pro -> Image → Nano Banana Pro.

Photorealistic cinematic portrait of a strikingly beautiful woman in her early
thirties, front-facing, head and shoulders, looking directly at camera. Dark
auburn hair in soft waves swept over one shoulder, luminous pale skin, striking
dark eyes, defined brows, deep red lip. Wearing a black bias-cut silk evening
gown with fine straps, diamond drop earrings. Expression cool and composed,
mouth closed. Warm practical lamp light from camera left, soft rim light along
the cheekbone and collarbone, deep rich shadow. Background: dark wood panelling
and warm bokeh from a hotel suite at night, heavily out of focus. 85mm lens,
shallow depth of field, high-end cinematic colour grade, deep blacks, subtle
film grain. Real skin texture. Glamorous, expensive, magnetic.

Same lamp direction, same background, same lens. That's deliberate — it lets them share a room later without looking pasted together.

Step 3 — Scene one

MODEL: GenX Pro -> Video → Gemini Omni Flash.

Settings: 9:16.

Attach both portraits as Reference, the man first. No start frame — this shot is built from nothing.

In the prompt, type @ where you see @man / @woman and pick the right image from the list.

Interior, a dark wood panelled hotel suite at night, warm lamplight. The camera
faces a tall mirror straight on. A man @man in a black tuxedo stands close to
the glass, tying his black bow tie, his face filling the reflection. He works
the tie slowly, unhurried, entirely absorbed in himself. He is alone for a long
beat. Then, far behind him and deep in shadow, a woman @woman in a black silk
evening gown walks slowly out of the darkness toward him and stops just behind
his shoulder, close enough to touch him. She is softly out of focus. She
watches his reflection. She says: "You look real." His hands stop. He goes
completely still, eyes fixed on his own reflection. Camera locked off, almost
imperceptible slow push toward the mirror. Ambient room tone, faint silk
rustle. No music. Anamorphic, 85mm, very shallow depth of field, deep blacks,
high-end cinematic color grade, subtle film grain. No subtitles. He does not
speak.

Trim the first 3 seconds of what comes back. She spends them walking in and it drags. You want to open with her already close.

Step 4 — Grab the last frame

Pause your trimmed scene one on the final frame and save it as an image.

This is the most useful move in the whole build. Omni Flash can't extend a clip, but it accepts a start frame — and a start frame carries the room, the lighting, the wardrobe, both faces and their exact positions into the next shot. Skip it and scene two is a different room with different people in different clothes.

Step 5 — Scene two

MODEL: GenX Pro -> Video → Gemini Omni Flash.

Settings: 9:16.

Attach the still as Start frame. Attach both portraits as Reference, man first.

Start frame: the attached still. One continuous unbroken shot, no cuts. [0-2s]
A @man man in a black tuxedo stands completely still, hands lowered, looking
straight into the lens. A @woman woman in a black silk evening gown stands just
behind his left shoulder, her hand resting on it. She looks at him, never at
the camera. Neither moves. A faint muffled party somewhere far below — soft
strings, low crowd murmur, distant glassware, heard through a closed door.
Close room tone. [2-3s] He turns his head toward her. One quick movement. His
face stays sharp and unchanged throughout. His expression is flat and
unreadable. [3-4s] He says quietly, almost without inflection: "I didn't ask."
[4-5s] He turns his head back to the lens and stares ahead, impassive. Her
expression falls — the warmth goes out of her face, her lips part slightly. Her
hand slides off his shoulder. [5-10s] Neither of them moves again. She keeps
watching him. Long still hold. Near-silence — only room tone and the distant
party far below. Consider micro-detail, expression and timing to create a rich,
detailed, entirely natural scene. Dark wood paneled hotel suite at night,
single warm table lamp. Camera locked off. Anamorphic, 85mm, shallow depth of
field, deep blacks, high-end cinematic color grade, subtle film grain. No
mirror, no reflections, nothing in the foreground. No other dialogue. No music.
No subtitles, no captions, no on-screen text.

Note the dialogue is generated with the video — voice and mouth made together, which is why lip sync isn't a problem here.

Step 6 — Captions

MODEL: GenX Pro -> Chat → Beast Agent.

1. Upload the video to the chat
2. Pick a caption style using the caption style chip in the prompt bar
3. Ask the Beast Agent to add captions

Something like:

Add captions to this video in the style I selected.

Beast transcribes the audio, builds the styled subtitles and burns them into the video. The captioned file lands in your Cloud Library.

Two things worth knowing. There are only seven words across fifteen seconds, so check both lines came through correctly before you accept it. And put the captions centre, not bottom — the bottom of the frame is where Instagram puts the username and caption.

Putting the two clips together

Cut scene one to scene two with a short crossfade — and place it where neither of them is moving.

Two Omni Flash clips of the same still setup can't be hard-cut. Scene two regenerates the actors in very slightly different positions, so a straight cut jumps even though both are standing still. A crossfade hides it completely.

Sound

Add a music bed that enters after the second line, not before.

Barely audible electronic drone, low and slightly detuned, with a faint pulse
underneath. No melody, no percussion, no vocals. Feels synthetic and slightly
wrong. Cold and unresolved. 15 seconds.

Bring it in about a second after "I didn't ask." The line lands in silence, then the music tells you what it meant. Before the line it just fills space.

Make it your own

Log in to GenX Pro or create a free account, then open the App and try these prompts.