I didn’t exactly have Terry Crews dancing on stage on my Adobe MAX 2026 bingo card. Then again, I didn’t expect to see half the things Adobe demonstrated that evening, either. If you’re a bit confused here, let me give you a real quick recap. MAX is Adobe’s flagship creativity event, where it brings together creatives from all over the world to showcase its latest innovations, share ideas, and give us a glimpse of what’s next.

One of the highlights of the conference is Adobe Sneaks, an event where Adobe’s researchers and engineers share some of the wildest technologies they’ve been experimenting with. If you follow laptop manufacturers closely, you might be familiar with the concept laptops Lenovo shows off at major tech conferences. We’re talking transparent displays, rollable screens, and all sorts of futuristic designs that look like something straight out of a sci-fi movie.

Adobe Sneaks is essentially the software equivalent of that. This year’s Sneaks had the entire crowd going wild, and for the first time ever, two of the ten projects showcased at MAX 2026 are already available for anyone to try through Adobe Labs Playground: Project Deep Brush and Project Hands On.

Project Deep Brush lets you paint on 2D images as if they were 3D

A whole new dimension to painting

Painting over an image sounds simple enough, right? You pick a brush, choose a color, and start painting. Imagine you’re trying to paint something on a curved surface, like a vase, or a piece of fabric with a bunch of folds, though. Suddenly, things get a little more complicated.

The problem here is that traditional brushes don’t really understand what you’re painting on. As far as they’re concerned, everything is just a flat surface. Well, Adobe’s Project Deep Brush lets you essentially turn any 2D image into a canvas that behaves like a 3D surface, allowing you to paint over objects while your brushstrokes naturally follow their curves, contours, and lighting.

During the live demonstration at Adobe MAX, they took an image of a wrinkled piece of fabric and painted over it using a bright red brush. Rather than appearing as a flat stroke of color, the paint followed the fabric’s folds and creases. They also demonstrated how you could use the tool to paint graffiti onto an angled concrete wall, with the lettering naturally following the surface’s perspective.

A more fun example was adding realistic-looking tattoos to photographs of people. The engineer demonstrated this by placing tattoo designs on his face and neck, with the designs wrapping around the contours of his skin.

Project Hands On lets you edit photos like you’re playing with paper dolls

Who needs prompts when you’ve got hands?

AI has made editing images significantly easier than it used to be. You upload an image, enter a prompt describing what you’d like to change, and patiently wait. While it works well enough in most instances, what if you want to make a very specific adjustment without having to figure out how to describe it in words? How do you explain that you’d like someone’s hand to be positioned just a little higher, or their arm bent at a slightly different angle?

Wouldn’t it be much easier if you could just grab their arm and move it yourself? Well, that’s essentially what Adobe’s Project Hands On lets you do. Instead of relying on text prompts, you can manipulate elements directly within an image using interactive controls. During the live demonstration, Adobe showed how you could reposition someone’s arms, reshape a perfume bottle, and even stretch a couch simply by dragging controls on the image.

The funniest moment came when Terry Crews asked the demonstrator whether the tool could give him bigger muscles. As luck would have it, the very next example Adobe had lined up involved doing exactly that: making someone’s biceps bigger. Talk about perfect timing!

Project Rinse Repeat makes repetitive edits a thing of the past

Been there, edited that

One of my favorite experimental projects from Sneaks was something called Project Rinse Repeat. This lets you essentially copy and paste edits you’ve made to one part of an image onto another, without having to repeat the entire process manually.

For instance, imagine you’ve spent ages perfecting the lighting and shadows on one object in Photoshop, only to realize you need to make similar adjustments to several other objects in the same image. Instead of going through the same steps over and over again, Project Rinse Repeat can capture the edits you’ve already made and intelligently apply them elsewhere. Interestingly, the tool doesn’t just simply copy and paste pixels. It recreates the edits using Photoshop’s native tools, since it intelligently understands the context of the image and adapts the edits accordingly.

You can also go back and fine-tune the results afterward. I personally think this is just the way AI should be used in creative tools. In no way is it doing the actual creative thinking for you. Instead, it’s simply taking care of the tedious, repetitive parts of the process, leaving you with more time to focus on the things that actually require your creativity.

Project Link Up automatically adjusts vector objects to match their surroundings

A perfect fit, without the fiddling

If you’ve ever worked with vectors in Adobe Illustrator, you probably know that adding a new object to an existing illustration isn’t always as simple as dragging and dropping it onto the canvas. You have to make sure its size, perspective, and style match the rest of the artwork, which can take quite a bit of manual tweaking.

Well, Adobe’s Project Link Up uses AI to make this process a whole lot easier. During the live demonstration, Adobe showed how you could drag an illustrated car into a parking space and have it automatically adjust to the scene’s perspective. They also demonstrated dragging a crab across a beach and an alligator into a swimming pool, with the tool generating realistic movement effects and water splashes.

The most impressive part is that everything remains fully editable vector artwork, including the AI-generated effects. Instead of getting a flattened image, you can continue editing every element in Illustrator!

Project Sight Span and Screen Test bring fictional worlds to life

Art comes to life (and starts talking back)

At one point during Sneaks, Terry Crews joked about how he’d always wanted to walk into a painting. Well, as it turns out, that’s pretty much what Adobe’s Project Sight Span lets you do. The project transforms a 2D image into a navigable 3D environment, and allows you to explore it from different angles and perspectives.

During the demonstration, Adobe used Johannes Vermeer’s famous painting The Milkmaid to show this off. Rather than being limited to the original view, the demonstrator could actually move around the room, explore areas that weren’t visible in the painting, and even generate new images from entirely different perspectives.

Another fun project was called Screen Test. During the demonstration, Adobe researcher Gabi Duncombe introduced us to Blork, a fictional alien character she’d created. Rather than simply typing prompts to give Blork a personality, she essentially hopped on a FaceTime-like call with her.

Gabi described Blork as a nervous, anxious alien, and Blork actually started asking follow-up questions to better understand her own personality. For instance, she asked whether she tended to hide or stand out around people, and how she moved when she was nervous.

Adobe then showed side-by-side comparisons of videos generated with and without Blork’s personality profile, and you could clearly tell the difference. Without it, Blork appeared relatively calm and confident. With it, you could clearly see her nervous personality come through in the way she moved and reacted to her surroundings.

These three experiments could change how we edit videos

Lights, camera, AI-ction!

A lot of Adobe’s focus throughout MAX was on editing videos using AI. Sneaks was no exception, and three of its ten experimental projects focused on making video creation and editing easier.

Project Scene Sweep lets you make changes to an entire video without having to manually edit every individual shot. For instance, you could change someone’s outfit, hairstyle, or even replace an object in the background, and the tool would intelligently apply those changes across every shot where they appear.

The video Adobe used to demonstrate this was hilarious, though. It followed a woman walking through a building, introduced a random man, and somehow ended on a romantic note. I couldn’t quite figure out what the storyline was supposed to be, and apparently, neither could the researcher demonstrating it. Regardless, Adobe showed how it could change the woman’s outfit, give the man a completely different hairstyle, and even replace a painting in the background with artwork of Terry Crews, all while maintaining consistency throughout the video.

Then there’s Project Live Set, which takes things a step further by letting you make changes to AI-generated videos while they’re still being generated. During the demonstration, Adobe turned a cat into the star of an action movie, complete with boxing scenes, and later showed how the researcher could change his own outfit into a tuxedo and turn a pen into a rose in a live video feed.

Finally, Project On Cue tackles another problem with AI-generated videos: timing. Instead of simply describing what you want to happen, you can control exactly when it happens. Adobe demonstrated this with a video of a Golden Retriever jumping on command, using an editable curve to control precisely when and how often the dog jumped. They also showed how the same technology could synchronize movements and flashing lights to music.

What I appreciated about each of these projects was that AI wasn’t doing the creative work for you. Instead, it was handling the tedious parts while leaving the creative decisions entirely up to you.

Project Sound Bender makes editing audio feel like editing an image

Sounds like a drag (in a good way)

We’ve covered images and videos, but Adobe had an interesting experiment for audio editing, too. It’s called Project Sound Bender, and instead of making you fiddle around with a bunch of settings to get a sound just right, it lets you experiment with sounds using an interactive canvas.

During the live demonstration, Adobe showed how you could generate sound effects using text prompts, then explore different variations by simply moving around the canvas. The researcher demonstrated this using an animated character brewing a magical potion, experimenting with different bubbling sounds before moving on to a more dramatic effect.

There’s been a lot of conversation following Adobe MAX about what the company’s latest AI advancements mean for human creativity. After watching Sneaks, though, I noticed a pattern. Most of these projects weren’t about replacing human creativity, but rather giving creators more control while taking care of the tedious work. This is exactly how I’d like to see AI used in creative software, and I’m really hoping we get to try most of these projects soon!