Rebuilding Photoshop's Engine, 35 Years In.
As you read in our last post, we're rebuilding Photoshop's engine from the inside out. Before I get into where that's headed, I want to back up and tell you how we got here.
Photoshop 1.0 was released on February 19, 1990. That week, the number one song in the country was "Opposites Attract" by Paula Abdul. Computers ran on 8-MHz processors. There was no JPEG format yet. Photoshop 1.0 shipped on a single 800K floppy disk, with one undo.
A lot has changed since then. I've been lucky enough to watch, and work on, a good chunk of it.
Performance, a recurring thread
Some of my earliest projects at Adobe were about performance: moving Photoshop from 68K chips to PowerPC on the Mac, then later helping it take advantage of MMX on x86, Intel's first real push into SIMD-style instructions for things like image processing. Since then we've added multithreading, more low-level CPU intrinsics, optimized bottlenecks, and gone through the 32-bit to 64-bit transition and the move from Carbon to Cocoa on the Mac. Then came GPU acceleration: first for canvas display, then for individual filters, and eventually for compositing itself, the core of what Photoshop does.
I've always liked watching things run at their best, whether it's a machine, a piece of software, or an athlete. So the question of how fast Photoshop could really go has stuck with me for a long time.
A different era of computing
Here's the thing that's nagged at me for over a decade, though: Photoshop was designed for a very different era of computing. I first felt this sharply when the iPhone showed up. I asked one of our engineers back then if we could get Photoshop running on it. He was honest with me: sure, maybe, but it would run terribly. Photoshop was built around desktop-class CPUs. It just wasn't built for a mobile world.
Around 2012 our teams started exploring multiple approaches and learned valuable lessons that informed the architecture we're building today. Some of that work found its way into our mobile products later. But we also ran into the real limits of GPU-only imaging engines: they struggle with the enormous files Photoshop is known for. And we learned just how high the bar is to earn the name "Photoshop.”
Fast forward to 2016 and the iPad Pro. Hardware had caught up, and we'd learned a lot from bringing Lightroom to mobile. This time, instead of rebuilding from zero, we ported the actual Photoshop codebase to mobile and web. That's how we got Photoshop for iOS, Android, and web, and it solved a real problem: true interoperability with desktop Photoshop and full PSD fidelity. We didn't have to rebuild every feature from scratch. But we ran straight back into the same wall: the compositing architecture itself.
November 2022
Which brings me to November 2022.
During one of our team-wide garage sprints, a small group of engineers put together the first prototype of what we now call the NextGen Engine. This NextGen Engine is a modern imaging architecture that powers how Photoshop renders, composites, and processes images, creating the foundation for faster performance and future innovation. When I saw it, I was floored, not just by how fast it was, but by the fact that it handled genuinely huge files without breaking a sweat. That combination is what had eluded us for a decade.
Two things made this possible, and I'll let the engineers who built them explain in more detail in future posts. The short version is that recent advances in both hardware and software finally opened a door that simply wasn't there before. NextGen Engine could not have existed even a year earlier.
Since that sprint, I've spent my time incubating this project: exploring where it could go, pushing on the internals, and yes, navigating plenty of internal debate along the way. New ideas take time to earn trust, and this one was no exception. But it's been one of the most rewarding stretches of my career.
Bringing it home to desktop
Now we're climbing what might be the tallest mountain yet: bringing this new engine into Photoshop desktop. This part is a genuine team effort. The NextGen Engine team built the new architecture, and the Photoshop desktop engineering teams know every corner of the existing codebase inside and out. Integrating the two is deep, demanding work on both sides, and it's happening because all teams are in it together.
This is where the earlier lessons really matter. NextGen Engine is a ground-up, GPU-first engine that doesn't compromise on speed. But we're not rewriting all of Photoshop. That would be nearly impossible, and honestly, unnecessary. Instead we're marrying a brand-new engine with the tools and capabilities that already exist. The best of both worlds.
And this pairing opens doors we couldn't open before. Take filters: with this engine, our goal is to make filters non-destructive and real-time, no longer trapped behind modal dialogs. You'll be able to attach them to any layer, or let them float freely in the layer stack, without needing Smart Objects or Smart Filters just to get non-destructive behavior. All of this runs through a floating-point HDR pipeline with premultiplied alpha, so transparent edges blend cleanly with no visible seams, using a single implementation per feature instead of separate versions for 8-bit, 16-bit, and 32-bit. That last part matters more than it sounds: today that split means real gaps in feature parity across bit depths, plus a lot of duplicated engineering and testing. One implementation fixes that.
There's more we think this unlocks, and honestly, we're still finding out.
Let's build this together
We have heard you: you want better performance, smoother workflows, higher-quality output, and more modern compositing. This project is our answer to that, and I'd like this to be a two-way conversation. What would you want a new imaging engine to unlock for you? What's been holding your workflows back? Tell us in the comments, I read them.
In the coming months, members of the team will be writing deeper technical posts on specific pieces of this work. I think you'll find it interesting.
This won't be an easy climb. But we are all committed to rebuilding Photoshop from the inside out, and I'm genuinely excited for what's next.
David Howe
Director of Engineering, Photoshop
AdChoices