[Guide] How We Made “Not Really Here” Video with GenX Pro App

The story

A black coupe swings into an empty gas station after the rain. The car door opens, and a heel shows up. She walks the pumps in a gold sequinned dress — dressed for somewhere she clearly hasn't come from.

Inside, coffee pours. She sets the cup on the counter, looks at the attendant and says, quietly:

"I'm not really here."

He is in awe, says nothing, just stares. She leaves and drives off.

What's The Generation Sequence for This?

Step 1 — The woman

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

Find an image online of a woman you like and attach it as a styling reference for a better outcome.

Two-panel beauty campaign reference sheet on a seamless neutral grey backdrop
(#7f7f7f). LEFT PANEL: full-body front view, confident stance, weight on one
hip, one hand loose at her side. RIGHT PANEL: large head-and-shoulders
close-up, facing camera. Clear vertical divide between panels.

Woman, mid twenties, strikingly glamorous. Long dark brown hair falling well
past her shoulders, glossy and voluminous with soft waves and high shine.
Luminous glowing skin, large dark eyes with defined smoky eye makeup and long
lashes, sculpted brows, full lips in a glossy nude. Direct confident gaze,
faint knowing half-smile.

Wardrobe: short metallic gold sequinned mini dress with sharp tailored
shoulders, catching light across every sequin. Long bare legs. Black patent
stiletto ankle boots. Black wool coat worn open over the top. Fine gold chain
bracelet on her left wrist, small gold hoop earrings. Dressed for a party she
has just left.

Bright high-key beauty lighting from front-upper, broad soft key with strong
wraparound fill, glowing skin with fine pore detail, bright catch-lights in
both eyes, hair backlit for shine and separation, hard glitter highlights
across the sequins. Soft contact shadow at the feet. Editorial beauty campaign
photography, 5500K. No text, no logos.

Use the attached photographs as reference for hair styling, skin luminosity,
makeup and overall glamour treatment only. Do not copy either face.

Step 2 — The attendant

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

Two-panel character reference sheet on a seamless neutral grey backdrop
(#7f7f7f). LEFT PANEL: full-body front view, standing, arms at his sides.
RIGHT PANEL: large head-and-shoulders close-up, facing camera. Clear vertical
divide.

Man, mid thirties, average build, slightly tired. Short dark hair, three-day
stubble, brown eyes, plain unremarkable face. Neutral expression.

Wardrobe: plain grey short-sleeve work shirt, a dark blue apron over it, black
trousers. No badges, no name tag, no logos.

Even soft studio key from front-upper, wraparound fill, smooth gradient on the
grey, soft contact shadow at the feet. Pore-level skin realism, visible stubble
texture, living eyes with catch-lights. 5500K. No text, no logos.

He is deliberately ordinary. She is the event; he is the room reacting to her.

Step 3 — The forecourt

MODEL: GenX Pro -> Image → FLUX 1.1 Pro Ultra.

Rural petrol station forecourt at night in the rain, plain unmarked fuel pumps
with blank dark display panels and a blank white price board, smooth unlettered
shop facade with warm-lit windows, cracked patched asphalt with wide puddles, a
plain steel bin, yellow bollards, coiled hoses, faded painted bay lines.
Documentary night photography. Cold white canopy downlights, warm sodium
streetlight on the road beyond, warm interior glow through the shop glass. Shot
on Canon 5D Mark IV, 35mm f/1.8, high ISO grain, visible rain. Three-quarter
angle.

Three things make this read as a photograph rather than a render:

Never write "photorealistic" or "cinematic" — both produce CG. Write "documentary night photography" and name a camera body, an aperture and high ISO grain.

Describe mess. Cracked patched asphalt, a bin, bollards, coiled hoses, faded lines. Clean and empty always looks fake.

Three or four colour temperatures, never one. Cold canopy, warm sodium, warm shop interior. A single blue grade is most of what makes an image look rendered.

Flux will invent garbled signage on every pass. Ignore it — this is a style reference, not a keyframe.

Step 4 — The shop, entrance

MODEL: GenX Pro -> Image → FLUX 1.1 Pro Ultra.

Interior of a small rural petrol station shop at night, service counter with a
coffee machine, crowded shelves of plain packaged snacks, scuffed grey linoleum
with worn patches, a mop and bucket beside a drinks fridge, glass entrance
doors at the back looking out onto a dark wet forecourt with canopy lights
beyond. Empty, no people. Documentary night photography. Green-white
fluorescent ceiling panels, cold blue night through the glass. Shot on Canon 5D
Mark IV, 28mm f/2, high ISO grain. Three-quarter angle.

Step 5 — The shop, counter

MODEL: GenX Pro -> Image → FLUX 1.1 Pro Ultra.

Service counter in a small rural petrol station shop at night, coffee machine
on the counter end with a stack of plain paper cups, a card reader, a plain
till, cabinet of packets behind, crowded shelves of packaged snacks to either
side, scuffed grey linoleum floor. Empty, no people. Documentary night
photography. Green-white fluorescent ceiling panels overhead, cold blue night
through glass doors off to one side. Shot on Canon 5D Mark IV, 28mm f/2, high
ISO grain. Three-quarter angle.

Step 6 — The car

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

Three-view vehicle reference sheet on a seamless neutral grey backdrop
(#7f7f7f), all three views on one baseline at equal scale. FRONT view, strict
SIDE profile, REAR view.

Low-slung two-door sports coupe in satin black, sharp modern angular body
lines, black alloy wheels, dark tinted glass, twin round tail lights, clean
unmarked bodywork with no number plate.

Broad soft studio key from front-upper with long graduated highlight bands
running along the flanks and roofline so the black form reads clearly against
the grey, gentle wraparound fill, soft contact shadow under the tyres. Sharp
panel gaps and crisp reflection edges. Neutral 5500K. Plain unbadged panels,
blank grille, no lettering anywhere.

Step 7 — The film

MODEL: GenX Pro -> Video → Seedance 2.5 Reference.

Settings: 9:16 · 720p · duration 21 seconds · audio on.

Attach your six images in this exact order — Seedance addresses them by index, so the order is the wiring:

SCENE CONTEXT Night, after rain. A woman drives into an empty petrol station,
buys a coffee, says one line to the attendant and leaves. He is still staring
when she is gone.

ACTIVE REFERENCES @Image1 — woman in a short gold sequinned dress and open
black coat, long bare legs, black patent stiletto ankle boots, fine gold chain
bracelet on her left wrist. 100% matches the reference, same person throughout.
@Image2 — man in a grey work shirt and dark blue apron, three-day stubble. 100%
matches the reference, same person throughout. @Image6 — satin black two-door
sports coupe, twin round tail lights, unmarked. 100% matches the reference.

LOCATION MAP @Image3 forecourt and @Image4 / @Image5 shop interior are STYLE
REFERENCES ONLY, not fixed keyframes; the model extends the world and the
subjects move through real space, never pinned 1:1. The lit shop building
stands at screen-left, the pump island at screen-centre under the canopy, the
road entrance at screen-right. Inside, the service counter runs across
screen-right with the glass entrance doors in the background.

FIRST FRAME / BLOCKING Non-empty opening frame: @Image6 is already moving,
approaching camera down the forecourt with headlights lit, small in frame.
Nothing is static.

FORMAT MODE Timed multishot, explicit HARD CUTs at the stated seconds; cuts
only at those points, the camera does not cut on its own. 21-second clock.

OPTICS Fixed shooting distance per shot — crop size changes only on the cut.
SEG 1 EWS 84°. SEG 2 ECU 12°. SEG 3 MS 47°. SEG 4 WS 63°. SEG 5 ECU 12°.
SEG 6 ECU 18°. SEG 7 MS 47°. SEG 8 MCU 29°. SEG 9 MCU 29°. SEG 10 WS 63°.
SEG 11 EWS 84°. No drift mid-segment.

CAMERA Locked off in every segment except SEG 3, which tracks smoothly
alongside her at eye level three metres out. Camera on the dark side of the
forecourt outside, on the shadow side of the counter inside. 180-degree shutter
motion blur throughout.

ACTION
0.0s to 3.0s — @Image6 drives toward camera down the forecourt at 20 km/h,
headlights dragging streaks across standing water, and stops close to camera.
The engine cuts. 3.0s HARD CUT.
3.0s to 4.5s — ground level: @Image1's foot in a black patent stiletto boot
lowers from the open car door and plants into a shallow pool, the surface
breaking into rings. 4.5s HARD CUT.
4.5s to 7.5s — @Image1 walks past the fuel pumps in profile at 4 km/h, gold
sequins catching the canopy strips, looking straight ahead. 7.5s HARD CUT.
7.5s to 9.0s — from deep inside the shop looking at the glass doors: @Image1
pushes through and walks in toward camera. @Image2 stands at screen-left.
9.0s HARD CUT.
9.0s to 10.5s — dark coffee streams from a machine spout into a plain white
paper cup, steam rising. 10.5s HARD CUT.
10.5s to 11.5s — @Image1's hand closes around the cup, gold chain bracelet
sliding down her wrist, and lifts it clear. 11.5s HARD CUT.
11.5s to 13.0s — two-shot at the counter: @Image1 sets the cup down and slides
it toward @Image2, who looks up and stops. 13.0s HARD CUT.
13.0s to 15.0s — @Image1 looks directly at him and says quietly, without
inflection: "I'm not really here." She holds the look. 15.0s HARD CUT.
15.0s to 17.0s — @Image2 stares, completely still, lips parting very slightly,
one slow blink, eyes never moving. 17.0s HARD CUT.
17.0s to 18.5s — wide down the aisle: @Image1 walks away from camera, through
the glass doors and out. @Image2 stays where he is, watching. 18.5s HARD CUT.
18.5s to 21.0s — exterior from behind @Image6: it pulls away from the pump
island, accelerates and recedes down the empty road, tail lights shrinking.
Standing water holds their red reflection after it has gone.

PERFORMANCE @Image1 is unhurried and entirely self-contained throughout — she
never looks at camera and never looks back. Her line is completely level: no
smile, no challenge, no invitation, stated the way someone reads out a time.
@Image2 is restraint only: amazement held still, no body movement, and he never
speaks. Pore-level realism on both faces, visible stubble texture, living eyes
with catch-lights.

PHYSICS The car settles on its suspension as it stops. Water displaces in real
rings under the boot and sprays from the tyres. The coat carries weight and
swings behind her stride. The cup settles with a small liquid shift. Correct
contact shadows on wet asphalt and on the counter. Nothing floats.

LIGHTING Exterior: cold white fluorescent canopy strips directly overhead as
the key, hard pools on soaked asphalt, warm sodium streetlight on the road
beyond, standing water holding mirror reflections. WB 5600K. Interior:
green-white fluorescent ceiling panels overhead as the key, flat and
unflattering, blue-black night beyond the glass doors, small hard catch-lights
in both sets of eyes. WB 4200K.

COLOR GRADE 60:30:10. Exterior: dominant cold blue-white canopy light on black
wet asphalt, secondary warm sodium on the road, accent the gold sequins and the
red tail lights. Interior: dominant green-cyan fluorescent cast, secondary
blue-black night through the glass, accent her gold sequins reading warm
against the green. Color tied to source and surface, not a flat list.

AUDIO NO MUSIC. SFX ONLY — diegetic sound throughout. Tyres through standing
water, an engine cutting, a heel entering water, patent heels on wet asphalt
and on linoleum, a glass door opening and swinging shut, a coffee machine, a
cup settling on a hard counter, a low fluorescent hum, an engine receding,
wind. Dialogue (English, audio only): @Image1: "I'm not really here."

STYLE 8K photorealistic — no 3D render, no game engine. Documentary night
cinematography, deep blacks, fine film grain.

OUTPUT SETTINGS 9:16 vertical, real-time throughout, total duration 21 seconds.

POSITIVE LOCKS Cuts only at the stated seconds; the camera does not cut on its
own. @Image1 speaks once; @Image2 stays silent for the whole take. @Image1's
gold sequinned dress and black patent boots stay identical in every segment,
gold chain bracelet on her left wrist. @Image6 stays satin black and unmarked
with no number plate. Every surface stays unbranded — no logos, no signage, no
readable text anywhere. Standing water holds mirror reflections in every
exterior segment. Contact shadows read clearly. Both sets of eyes stay natural.
No subtitles, no captions, no on-screen text.

Step 8 — The music

MODEL: GenX Pro -> Audio → ElevenLabs Music.

Cinematic spy-thriller sting. Opens instantly on a huge orchestral hit — full
brass blast, timpani and low strings slamming together on the very first beat.
No intro, no fade-in, no build-up. Straight from that hit into a tense surf
guitar tremolo riff with muted brass stabs and a walking upright bass. E minor,
dark, cool and dramatic. Instrumental only, no vocals.

Step 9 — Captions

MODEL: GenX Pro -> Chat → Beast Agent.

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

Add captions to this video in the style I selected.

There are only four words of dialogue in twenty-one seconds, so check the line came through before you accept it. Put the captions centre, not bottom — Instagram owns the bottom of the frame.

Make it your own

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