Adobe MAX 2025 Sneaks: Your First Look at the Latest Next-Gen Tools

Hosted by Adobe’s Paul Trani and comedian Jessica Williams, the annual MAX Sneaks session mixed comedy and Williams' constant astonishment with early-stage tech demos, showing just how far Adobe’s research teams are pushing generative AI.

“It’s like getting your mind blown repeatedly for two hours,” said Shrinking star, Daily Show alumnus, and all-round goddess of this parish Jessica Williams at the end of the Adobe MAX 2025 Sneaks session. And she's not far wrong. There was some stunning stuff on show on the MAX 2025 Sneaks stage as Williams' constant surprise made abundantly clear.

There was perhaps less video-related stuff than we have seen before, but the tone was set early on by Adobe's Mohit Goel demoing Project Motion Map, who showed how to animate static artwork with text-based prompts rather than using timelines or keyframes. "Make the eyes blink," he typed, and the eyes blinked. "Rotate the logo," he commanded, and the logo rotated.

"What the hell are you guys clapping? What just happened?" yelled an astonished Williams, setting the tone for the evening. "Walk me back like I'm a Shark."

Project Frame Forward

"Imagine if video editing was as simple as editing an image," said Trani introducing Project Frame Forward. And that's pretty much the elevator pitch for Project Frame Forward. Basically, it lets users edit one frame of a video in Photoshop and then has AI propagate the same change across the whole clip.

Check it out at 1:15:25 in the video below. Jui-hsien Wang gives the demo and it's a bit of a jaw-dropper. A speeding car trailing smoke filmed from a fast-moving drone shot is perfectly deleted from an entire sequence just by Wang highlighting the car in one frame. Frame Forward can add too. One click adds a non-existent puddle in front of a cat with its reflection already on the water. Another click extends the new content throughout the sequence.

It's impressive stuff and isn't accomplished using masks or tracking, rather the new Photoshop Trace Erase next-gen remove tool that was also introduced in the Sneaks presentation.

Audio Sneaks

Project Sound Stager is an AI sound-design assistant that analyses video scenes, adds appropriate effects, and offers full multitrack and chat-based control as it goes. Again, the demo was impressive. A cute animation of a busy mother octopus getting her kids ready for school gets analysed, split into scenes, and foley effects are added.

Each effect is detailed in a full multi-track view of all the generated sound effects that can be individually edited, alternatives selected, and more.

There was another good audio one earlier in the shape of Project Clean Take, where misspoken dialog was cleaned up. Uptalk inflections were removed, the word 'fifth' was changed to 'fourth', and much more. This is potentially going to save productions a lot of money, as we all know that reshoots are expensive.

"How much time is that gonna save for actors?" said an astonished Williams. "I'm gonna learn half my lines now, and I'm gonna call you and ask you to fill in the rest."

You can watch a replay of the whole event below. (Sorry for the small window size, but you can click through to watch directly on YouTube. We resize the picture and we lose the preview image for some reason. Coding, eh? Ed.)