For Zeev Farbman,
$1.99 was only the beginning

At four in the morning, a university career can begin to look like a peculiar bargain. Zeev Farbman and Yaron Inger had been working late in their Jerusalem lab. Walking home, Inger offered a calculation: put these hours into something outside academia and they might already be millionaires. Farbman remembered the conversation. Until then, he had imagined a future with tenure.

The ambition that followed was large, but the first price tag was small. Facetune began at $1.99. A few dollars bought an ordinary phone owner access to retouching that had previously required considerably more patience and software. The researchers had found an audience well beyond their colleagues. The audience, it turned out, had quite a few pictures to fix.

More than a decade later, Farbman is spending his time on a different sort of image problem. He leads LTX, the AI business separating from Lightricks, and wants other companies to build on its models. The phone app helped finance the research. Now the researcher is reconsidering the shape of the company it created.

A childhood with engineering in the background

Farbman was born in Zaporizhia, in what is now Ukraine, and grew up in Yakutia. His parents were engineers who operated a satellite ground station. In 1992, the family moved to Netanya. One detail of the arrival stayed with him: people in Israel wore their shirts untucked. A change of country can announce itself through the smallest breaches of etiquette.

Computers entered his life early. He recalls using BASIC at ten and enjoying the business of making attractive pixels. At Hebrew University of Jerusalem, that interest became sustained study: computer science through a doctorate, with research in image processing and computational photography. There was a direct connection between the child arranging pixels and the adult asking how a computer could improve a photograph.

His intended destination was academia. Nir Pochter, another future co-founder, shared that ambition. When Farbman later described their younger selves, university professorship sounded like a prize worth pursuing. That matters to the story: the company emerged from people who already had something they wanted to do. Entrepreneurship had to compete with the lab.

A pocket-sized research problem

Early prototypes explored low-light photography. The work interested the researchers; getting its value across to a wider audience proved harder. A reshaping tool offered a more immediate connection. People could see what it did, understand why they might use it, and make the change themselves. Scientific ingenuity had found a gesture that fitted a fingertip.

In 2013, Farbman, Pochter, Inger, Amit Goldstein, and Itai Tsiddon founded Lightricks. Facetune arrived that year. Adobe and Autodesk offered examples of what a creative-software business could become, while smartphones offered a new place to build one. Facetune was conceived as an entry into that larger market, rather than the full extent of the founders’ plans.

“At the beginning, we were throwing darts.”

Zeev Farbman, on early Facetune pricing

Pricing required its own experiments. The team started at $1.99, raised the price, and eventually found a balance around four dollars. Pochter argued that effortless customer acquisition might indicate a price set too low. Alongside the mathematics of images came the less elegant mathematics of a business: what would people pay, and what would it cost to reach them?

The surroundings were suitably practical. Farbman has recalled a co-founder pursuing journalists because there was no marketing budget. Office space came from renovating an abandoned student dorm. Even another table counted as a meaningful expense. The mobile-software business began with a very physical problem: finding somewhere affordable to put the people making it.

The boss becomes the material

Image-editing software also provided an occupational hazard. A photograph of Farbman asleep on an office sofa circulated online, followed by altered versions that placed him in increasingly improbable circumstances. His own company’s field of expertise had made the boss exceptionally easy to repurpose. Usually a chief executive has to commission this much imaginative attention.

There is a useful change of scale in that episode. The public biography of a technology founder tends toward financing rounds and product launches. Here, the person making image tools became an image other people could play with. The joke needed no explanation of computational photography. The result was immediately legible, which was part of what had made the original app work.

Outside work, Farbman practices Brazilian Jiu-Jitsu and enjoys Japanese food. His reading has included Neal Stephenson’s The Baroque Cycle, which he cited while discussing scientists who changed accepted ways of seeing. His description of the early company emphasized room for research and respect among colleagues. Those interests offer a more specific picture than the usual founder vocabulary of appetite and ambition.

When the successful tool becomes ordinary

The business expanded into other creative applications, including Enlight and Videoleap. In July 2019, Lightricks announced a $135 million financing round at a $1 billion valuation. Its seven apps then had more than 180 million downloads and nearly three million paying subscribers. A small purchase had become a continuing relationship with millions of customers.

In 2021, another financing round valued the company at $1.8 billion. The stated direction extended beyond editing: Lightricks wanted to support creators through more of their work, including ways to earn from it. The lab’s research had become part of an organization dealing with products, distribution, acquisitions, and customers in different markets.

Success did not remove the possibility that the underlying product category would change. By December 2023, Farbman was discussing photo editing becoming a commodity. Generative AI could make formerly specialized operations widely available. A company that had simplified a difficult task now faced technology capable of simplifying it further.

That left two kinds of work on the same desk: serving people who already used the apps and finding uses for a technology that could change what those apps were worth. A paying customer still expects the familiar tool to work on Monday. A new model does not wait politely until the product roadmap has a vacancy.

Two businesses, different clocks

In February 2026, Farbman announced two business units, Facetune and LTX. He described the profitable consumer business as the source of LTX’s funding. The arrangement connected two different commercial rhythms: a mature application with customers and revenue, and research that required substantial investment before its eventual market could be settled.

June brought a more difficult account of the transition. Plans for separate companies accompanied 75 job cuts. Farbman would focus on LTX, while Asaf Porat would take over Facetune. The restructuring also involved hiring for different technical skills. The change had a cost for employees, as well as a strategic rationale for management.

The distinction between operating separately and completing a legal split matters here. In August, completion was expected by year end, subject to tax approval. LTX already presented Farbman as its CEO; Lightricks’ public description continued to call the separation a process in progress. The new chapter had begun while its corporate paperwork was still catching up.

Let someone else build the next layer

Farbman’s answer to the AI market centers on giving developers more control. LTX releases model weights so other teams can adapt the technology and run it in their own environments, alongside a commercial licensing business. A studio might want a particular visual style. An enterprise might want its material kept within its own infrastructure. Access to a remote service does not satisfy every requirement.

In a conversation with Eden Shochat, he explored the relationship between a model and the products surrounding it. The gaming-engine analogy is helpful: a capable underlying engine supports many applications, while the tools around it determine how much useful work people can accomplish. His public arguments increasingly concern what others can build, rather than only what LTX can demonstrate.

He also remains connected to the research itself. The January 2026 LTX-2 technical report lists Farbman among its authors. The model joins video and audio generation in one system, with separate streams linked to keep them synchronized. The executive title has not removed his name from the paper. The work still involves the particulars of how images get made.

What happens after this frame?

LTX’s ambitions now extend into robotics and simulation. Farbman argues that a system able to predict subsequent frames could provide useful foundations for machines acting in the physical world. Startups could adapt a foundation model to their own data and applications, rather than each paying for the entire underlying training effort.

“We want to build the expensive part.”

Zeev Farbman, on foundation models for other companies

This remains an ambition with demanding tests ahead. A robot must connect predictions to actions and information from sensors. LTX describes world models as a way to reason about physical space and what follows from a change within it. Farbman is placing the business at that junction, where a moving picture becomes useful for something beyond watching.

The scale has changed since the walk home from the lab. So has the question. The young researchers wondered what their effort might produce outside the university. Farbman now wants to supply research that lets other people build their own businesses. Facetune gave someone a way to change a picture. He is betting his next chapter on what happens in the next frame.