Lens Cap Labs
A child in a yellow raincoat running along a misty lake pier at golden hour, seen from behind

Lens Cap Labs

Photos you
never took.

A roll of film held thirty-six exposures. Frame thirty-seven is the one you did not get: the light that had already gone, the day you left the camera at home, the face you wish you had turned toward a little sooner.

Join the waitlist

Local AI

The computer on your desk is enough.

No queue, no quota, no upload. The model runs where the pictures are, and it works when the network does not.

Read more

Privacy

Some pictures are not for a machine you have never met.

Your library is not training data for someone else. Nothing leaves, and what is made is marked as made.

Read more

Personalization

Not the average of everyone. You.

Ten thousand of your own decisions are a better teacher than the internet. Software should learn from them.

Read more

Ownership

A tool you own, not one you rent.

It lives on your disk, works without permission, and answers to no meter. Make a thousand, keep one, and never ask anyone.

Read more

Show, don't tell

A picture is worth a thousand words. Do you really want to type a thousand of them?

That is what generating an image has come to mean: a paragraph of adjectives, a wait, a result that is almost right, and another paragraph to fix the part that was not. You type and hope. Then you type again.

A photograph should not be argued into existence. Put the pieces where they belong. Paint over what should change and leave alone what was right. Slide until it looks like what you meant. The picture is the instruction, and the frustration goes away, because you are no longer describing what you want. You are doing it.

A vast terraced city carved into a canyon wall at dusk, thousands of lit windows, bridges and a monorail spanning the chasm, a river of mist and blue bioluminescent plants below, a ringed gas giant with two moons filling the sky, airships, waterfalls, and a lone figure on a stone balcony
Try describing this. Which towers are lit and which are dark, where the second moon sits against the rings, how far the train has come along the viaduct, where the mist ends and the river begins, what the figure on the balcony is looking at. Or just move the pieces until it looks right.

I

Fixing is the first thing. Making is the second.

Once something has learned your way of seeing, it can do the ordinary work the way you would: straighten the horizon you would have straightened, warm the skin you would have warmed, leave alone what you would have left alone.

And it can go one step further. It can make the picture that was not taken, of people and places it knows, in light you have shot a hundred times. Not an image of someone. An image of them, the way you see them.

A man in a grey sweater at a cafe window, flat overcast light, straight out of camera
As shot
The same portrait, warmed and graded in the photographer's style
In your style
The same man standing on a beach at sunset, a photo that was never taken
Never taken

II

Keep the subject. Change everything else.

A good photograph is mostly about what you refuse to change. The dog stays the dog: the tilt of the head, the wiry coat, the way he sits. The porch can become a mountain in snow, and nothing that mattered is lost.

A wire-haired terrier sitting on farmhouse porch steps in morning light
Porch, as shot
The same terrier in the same pose on a snowy mountain trail with pine trees behind
Same dog, new place

III

Things have a style too.

A potter photographs the same forms every season. A shop shoots the same shelf every week. The look is already settled, the light is already chosen, and the only question is whether the next hundred pictures will match the last hundred without a hundred hours of work.

Three handmade stoneware vases on linen against a plaster wall in soft daylight A tall green stoneware vase with dried grasses and a terracotta bowl, photographed in the same style

IV

Nothing leaves.

The pictures of your children, your clients, your home and your work are not training data for someone else's model. They are not uploaded to be studied. They stay on the machine they are on, and everything that learns from them learns there, where you can see it, keep it, or delete it.

What is made is marked as made. A photograph you took and a photograph you did not should never be confused, by you or by anyone, and nothing generated is ever learned from.

A darkroom with prints drying on a line under a red safelight
The darkroom never asked where the negatives came from. Neither should this.

The app

Frame37

Frame37 is the first thing we are making. It is a local AI photo app for Mac, for your own library: a place to look at what you have, a way to fix it as you would have, and a way to make the frames you missed. It learns on your computer and runs on your computer.

We are building it with the people who will use it, and we would rather show it than describe it. Leave an address and we will tell you when there is something to see.

Be the first to know.

Only news about Frame37. No sharing your address. Unsubscribe any time.

About

Lens Cap Labs

Lens Cap Labs is a small company in Plano, Texas, founded in 2026 by Dr. Suraj Raghuraman. Suraj previously built and led the AI engine team at Topaz Labs, where he was responsible for the on-device inference engine behind its products and the company's partnerships across GPU and silicon vendors. He holds a PhD in computer science from the University of Texas at Dallas.

The name is the thing you take off before the picture.