Look at a finished effects shot and the labor has vanished. The spaceship has mass. The impossible creature has skin. A sunny landscape that never existed has weather. Somewhere beneath those convincing surfaces is usually a less cinematic sight: boxes connected by noodles on a computer screen. That node graph is where Foundry lives.
The London company makes the production tools that turn separately rendered elements into a final image, paint detail onto digital objects, light enormous CG scenes, organize storyboards and carry imagery onto an LED stage. Nuke is the flagship compositor. Mari handles high-resolution 3D painting. Katana manages lighting and look development across many shots. Flix keeps story development from becoming a landfill of filenames. Nuke Stage brings familiar color and compositing logic onto virtual-production sets.
Its customer list reads like the credits you stay seated to watch: Industrial Light & Magic, DNEG, Pixar, Weta FX, Walt Disney Animation Studios, Framestore, Skydance Animation and others. Foundry says its software has played a part in every Academy Award winner for visual effects since 2007. The boast sounds glamorous. The work is mostly about preventing a late color change from becoming a small organizational fire.
The invisible factory
One movie, thousands of tiny dependencies
A modern shot may combine live action, computer-generated characters, simulated smoke, digital crowds, matte paintings, lens effects and last-minute notes. Each element can arrive from another department, another vendor or another country. The compositor must make the pieces look photographed together, then preserve enough structure to revise the result tomorrow.
Nuke approaches that job as a graph. An artist connects operations rather than flattening every decision into a stack of layers. The graph can be inspected, scripted in Python, adapted to a studio pipeline and reused. More than 200 nodes cover keying, tracking, color, transforms and image manipulation. Deep compositing stores depth information with image samples, reducing the need to re-render CG when the order of objects changes. It is not the friendliest first afternoon in software. It is designed for the five-hundredth shot.
Where Foundry sits in the production line
The suite is narrower than Adobe Creative Cloud and less interested in general-purpose 3D than Autodesk or SideFX. Its distinction is depth at specific, expensive junctions. Katana can defer loading objects an artist is not working on. Mari can paint assets too dense for ordinary texture tools. Flix sits between artists and editorial, quietly naming, versioning and retaining revisions. The products also speak the dialects of large studios: OpenUSD, EXR, ACES, OCIO, render delegates and APIs.
“The real product is not a prettier button. It is a handoff that does not require a rescue meeting.”YesPress analysis
Born inside the studio
Foundry’s best trick was knowing what to buy
Foundry did not invent its entire portfolio in a Soho conference room. Bruno Nicoletti and Simon Robinson founded the company in 1996 to make plug-ins for Flame and Inferno, the elite finishing systems of the day. Furnace, its motion-based image-processing suite, later earned an Academy Scientific and Technical Award. But plug-ins imposed a ceiling. In Foundry’s own account, the company wanted a proprietary platform on which it could build more ambitious tools.
That platform was already working nights in Los Angeles. Nuke began as Digital Domain’s in-house compositor in the 1990s and helped on films including Titanic. Foundry took over its development, sales and support in 2007. The customer that had incubated the software became a buyer; a private studio tool became a commercial standard.
The pattern repeated. Mari emerged from Weta Digital’s need to paint detailed assets. Katana came from Sony Pictures Imageworks, where it organized lighting and look development. Flix also came from Sony, created to tame the iterative mess between story artists and editorial. Foundry’s method was less romantic than the garage myth and more useful: find software proven under production pressure, productize it, support it and sell it to peers facing the same pain.
Ownership changed almost as often as the software did. Carlyle bought Foundry in 2011, then sold it to Hg in 2015 at a £200 million enterprise value after revenue had grown more than 160 percent. Roper Technologies acquired it in April 2019 for a reported £410 million. Roper’s model is to own durable specialist software businesses, and Foundry continued under its own name. What it did not gain was public financial transparency: Roper does not break out Foundry’s revenue.
The invoice
Industry standard is a business model
Foundry sells annual subscriptions, rentals, render entitlements, support and larger negotiated studio deals. Current US list prices put Nuke at $3,839 a year, NukeX at $5,219 and Nuke Studio at $6,379 before tax. Mari lists a $1,289 annual team subscription and a cheaper individual plan. Katana, Flix and Nuke Stage lead customers toward a conversation rather than a checkout button.
Sticker price, selected annual US subscriptions
Public prices viewed September 2026, before tax. Enterprise, Katana, Flix and Nuke Stage pricing is quote-led.
The price makes more sense when measured against the alternative: custom engineering, retraining, broken integrations and artists waiting for renders. Familiarity becomes a moat when hundreds of people, years of scripts and a library of shots assume the same behavior. The lock-in is partly contractual, but mostly operational.
That is also why licensing changes sting. Foundry plans to move active maintenance customers for Nuke Family, Mari and Katana to subscription when they renew from January 2027. It says the first subscription will match the maintenance fee and annual increases will not exceed 5 percent. Perpetual licenses remain usable, but customers who leave maintenance lose new versions, fixes and support. A predictable recurring model for Foundry is a fresh dependency calculation for buyers.
The retreat and the pivot
Modo ended. The pipeline won.
The first visible crack in the broader portfolio was Modo. Foundry had merged with its maker, Luxology, in 2012, expanding into modeling and animation. In November 2024, Foundry announced that active development would end after Modo 17.1. Customers could obtain a ten-year license, but Foundry warned that future operating-system compatibility could not be guaranteed and recommended migration.
The official explanation was focus: invest in core media and entertainment products and new solutions. It is an unusually plain admission for a software portfolio. A product can be admired, useful and still lose the internal contest for engineering attention. Foundry’s center of gravity had returned to the connected studio pipeline.
The replacement bets sit on either side of production. Nuke Stage extends Foundry onto LED stages, using USD, EXR, familiar node logic and shared color management rather than asking a VFX crew to think like a game developer. It records camera, lens, timecode, scene and color metadata so post-production can begin with the choices made on set instead of reconstructing them.
AI is arriving through the other door. CopyCat, introduced in Nuke in 2021, lets artists train image-to-image effects from examples. In February 2026, Foundry acquired Griptape, an orchestration company built to connect multiple models and agents. By July it had launched an Enterprise tier and named Griptape co-founder Kyle Roche chief AI officer. The pitch is controlled infrastructure, not a magic prompt box: studios choose models, keep workflows secure and fit the results into existing systems.
The useful lesson
Automate the repetition, preserve the judgment
Nuke 16 offers the cleanest example. Compositing was historically organized one script per shot. A change repeated across a sequence meant copying nodes, recreating settings and inviting inconsistency. Graph Scope Variables let artists store values that apply across shots, while Link nodes and multishot structures keep related work together. Skydance Animation used the approach on Spellbound.
“Multishot brought us a lot of efficiency because it enabled us to handle more than one shot with the same script.”Alberto Juarez, compositing supervisor at Skydance Animation
There is a condition attached. Juarez said shots sharing a script need the same layer breakdown. Preparation is key. Multishot is powerful when inputs are consistent; it becomes another knot when every shot is a special case. Nuke Stage likewise works best in productions already committed to disciplined color, metadata and open formats. Katana rewards teams with enough shots to justify templates and pipeline engineering. For a solo creator making a short social clip, cheaper and simpler tools will often win.
Copy this
Productize tools that have survived real operating pressure. Studio-born software arrives with an unusually honest requirements document.
Copy this too
Automate naming, versioning and repetition before reaching for generative spectacle. Administrative friction compounds across every shot.
Do the prep
Standardized inputs make multishot work. Without common layers and conventions, shared automation magnifies exceptions.
Know the boundary
Deep pipeline software pays off at scale. Small, irregular projects may not recover the license, training and integration cost.
Foundry’s culture statements fit the business: Customer First, Ambitious and Stronger Together. The important evidence is not the wall copy. Product teams say they spend time with studios; tools expose Python and open-format integrations; retired software receives an exit license rather than simply disappearing. The company can still frustrate customers, particularly when price and licensing change. But its best work begins with a production nuisance specific enough to sound boring.
That may be Foundry’s most durable advantage. Spectacle changes every season. The need to track a revision, carry a color decision, load only the useful part of a scene and hand work to the next person does not. The final shot should feel miraculous. The software behind it should feel dependable.