Images for a business goal

One person — endlessly many frames

Ordinary generation draws a new face every time. We lock one in and shoot any series with it: new scene, clothing, season, same person. Six months later you can shoot a sequel and no one spots the seam.

Try it free → Order a series No sign-up: describe a frame, it is ready in a minute. Like it? We lock the face and shoot a trial series.

Try the selection on a live example →This is what the client sees: mark the faces that work, add a note and send with one button.

How it works

Four steps. The first two happen once, then it is only shooting.

Step 1

Choose the look

We show face options. You pick the one that becomes the brand's face.

See how the selection works →

Step 2

Lock the face

We train the model on the chosen face. Ninety minutes and it is fixed for good.

◆ face locked

Step 3

Shoot series

You name the scenes, we deliver frames. Overnight the farm makes about 460 finished.

Step 4

You select

We send a grid page with a recognizability score under each frame. You call the numbers.

Not stockOn stock every photo has its own face. You cannot build a brand series from them.

Not a one-off AIAn ordinary generator does not remember a person: tomorrow it is someone else.

Not a photoshootA shoot means a model, a location, lighting and a full day per batch.

100%

frames passed a machine face-recognizability check

460 pcs

finished frames per night shift

71 s

per frame with dedicated face refinement

∞

series with one character after training

Why it is cheaper than a shoot

The difference is not the price of one image but of the second, tenth and hundredth. A photoshoot starts over every time; a character is created once.

Task	Photoshoot	Stock	Brand character
First batch of frames	a shoot day	minutes	90 min + one night
Second batch a month later	another day and budget	the faces differ	just one night
The same person	if the model is free	no	always
Change season, city, clothing	a new shoot	whatever exists	one line in the brief
Image rights	a model contract	a limited licence	yours, unrestricted
Add a single frame	impossible	impossible	71 seconds

Brand character

One woman in six scenes. We did not cherry-pick similar frames — it is one series in a row. The number under each frame is a machine match of the face to the character reference.

morning at camp · recognizability 0.74

Product photography

Products without a studio, lighting or logistics. The same item is reshot in a new palette and angle in a minute — for a specific promo, season or channel.

Interiors and atmosphere

A space that does not exist yet, or is awkward to shoot with guests around. Straight verticals, correct light, free room for a headline.

Live generator — try it yourself

The same technology we use for characters and product cards — open right here on this page.

1Pick a style and describe the frameOne sentence: who, where, in what light.

2Press “Generate”In about a minute the frame appears on the right, next to the example.

3Need the same person across a series?Write to us — we lock the face and shoot a trial series before any commitment.

How we work

The character stays yours. We do not use it for other clients.

A talkWhat the frames are for: feed, cards, ads, covers. Everything else follows from that.
Look optionsWe show several faces. You choose the one that becomes the brand's face.
A trial seriesWe shoot a small batch before agreeing on anything. You look, you decide.
Work in seriesThen you send scenes and take frames. Selection by link, the originals are yours.

Measurements

Shot on our own hardware; the numbers are not marketing.

Frame with a character and face refinement	71 s
Product frame	39 s
Raw frames per shift	1376
Finished after selection	460
Passed the recognizability check	100 %
Source resolution	up to 1824 px

A SmittMediaGroup project — AI for business and staff training · Also: Ready-made sites — demo · Neuro-art galleries

Recognizability is measured by machine: each frame's face is turned into a vector and compared with the character reference. It is not an eyeball guess.

Frequently asked questions

Short answers to what people ask most.

How is this different from an ordinary neural network?

An ordinary generator draws a new person every time: tomorrow the face is different. We lock the face once, and after that every frame shows the same person.

Do I need a photo shoot and a model?

No. The character is created once, then you name the scenes and we deliver frames. No studio, no model, no day on location.

How long does it take?

Locking the face takes about ninety minutes, once. One frame with the character and face refinement takes 71 seconds. A night shift produces about 460 finished frames.

Can I add more frames six months later?

Yes. The face is fixed for good, so a sequel can be shot at any time and will not differ from the first batch.

How do I know it is the same person?

Recognizability is measured by machine: each frame's face is turned into a vector and compared with the character reference. It is not an eyeball guess.

What does it cost and where do I start?

The first generation in the live generator is free, no sign-up. A trial series comes before any commitment: you look first, then you decide.

Capacity is available

Tell us about your task

Write a couple of lines about what the frames are for. I reply personally, usually within an hour. We shoot a trial series before any commitments.

Telegram: @SmittMG

Email: windowsplanet [at] gmail [dot] com