Case study · Curved beachfront residential tower with glass balconies
One breath on the balcony
A tower render and a blurred mood photo, turned into a sequence of editorial frames, from bright midday to a sunset flare, plus native vertical stories.
- First session
- 30 min
- Runs
- 15
- Key visuals
- 15
- Credits
- 1,800
In short
This campaign started from one CGI render of a curved beachfront tower and one soft focus mood photo of a woman turning into the light. The brief asked VantageKV to take the light, mood and action from the reference and put her on a balcony. In about 30 minutes of focused work it made 15 key visuals, from bright midday to a sunset flare, including vertical 9:16 story frames rendered at that shape, for 1,800 credits.
What went in
The first thing typed
“Take the light, mood and action from the reference. put her in one of the balconies.”
Word for word, typos and all. Claude turned it into a full art direction from the render; the building itself was never described.
- Building
- Curved beachfront residential tower with glass balconies
- Look
- Editorial
The session
Every run, in the words that made it.
1 min
“Take the light, mood and action from the reference, photography style, blueness.. put her in one of the balconies.”
Started in 9:16, for stories, and rendered natively at that shape rather than cropped from a landscape. The curtain across the lens came from the reference's soft foreground.
4 min
“photography style, blueness.. put her in one of the balconies. sunset”
In landscape the frame went bright and airy. Lovely, but closer to a brochure than to the reference.
15 min
(the same brief, Golden hour look added)
Closer, eyes closed, the tower curving behind her. The building is the backdrop, and it is exactly the building.
23 min
“Take the light, mood, photography style and action from the reference.”
Asking for the reference's photography, not just its mood, brought the soft focus and the foreground blur in.
29 min
(the same brief, another take)
By now every take is usable. The choice becomes editorial, not technical.
Three days later
“the same brief, Sunset picked on the time of day strip”
The hero, from a revisit: one tap on the time of day strip and the sun sits on the horizon behind her.
Before a credit was spent
The art direction, in five lines anyone can check.
Claude writes a full art direction for the image model, and the Studio shows it back as five plain lines. Any line can be switched before rendering. These are the five lines behind the final key visual.
- Camera
- close long lens, shoulder height, tilted
- People
- one woman turning into the light
- Light
- sun on horizon, strong backlight, flare
- Motion
- hair and dress smeared by slow shutter
- Colour
- warm gold against cool muted shadow
The campaign
One session. Everywhere the launch runs.
One breath on the balcony
Register your interest
Site hoarding, 21:9
Stories, 9:16, rendered natively
Feed post, 1:1
The residence
One breath on the balcony
Brochure spread, 3:2
Layouts are mockups built from this session’s frames, cropped to fit. In the Studio every format, 21:9, 9:16, 1:1 and 3:2, is rendered at its own shape, so nothing important is cut off.
What made it work
01
A mood photo can be a whole brief.
The reference here is barely a photograph, a blur of a face and some light, and it set the camera, the lens and the feeling for every frame.
02
Make the vertical at the vertical.
The 9:16 story frames were rendered at that shape, so nothing important falls off the edge.
03
A sequence beats a single image.
Fifteen frames of one moment gives a launch its carousel, its stories and its hoarding from half an hour.
Questions
What people ask about shots like this.
Can a blurry mood photo be used as a reference?
Yes. The reference here is barely a photograph, a blur of a face and some light, and it still set the camera, the long lens, the soft focus and the feeling for every frame. VantageKV takes a reference’s photography, never its people.
Can AI make 9:16 stories from a render without cropping?
Yes. Pick 9:16 and the key visual is rendered at that shape, so the person and the building are composed for a phone screen rather than cut out of a landscape.
How do you make AI people look real, not like stock models?
Every VantageKV render carries a rule for real skin texture, natural expressions and unposed moments, so people read as photographed residents rather than retouched models.
How long did it take?
About 30 minutes of focused work for 15 frames, with a final sunset version made later by picking Sunset on the time of day strip.
Try it on your own render. Free.
Next case study
All case studies