Every year at its Max conference, Adobe trots out a batch of unfinished ideas it calls "sneaks" — concept tools that may or may not ever become products. This year's Adobe AI experiments, reported by The Verge's Jess Weatherbed, do two things that sound like science fiction until you watch the demo: they let you step inside a photograph, and they let you sit down and interview a character who doesn't exist.

Both are prototypes. Neither has a price tag, a release date, or an obvious home inside Adobe's existing apps. But the demos tell you a lot about where photo editing and AI video are heading, and how quickly.

What happened

Adobe showed two headlining sneaks. One is a new way to look at pictures. The other is a new way to build a personality.

Project Sight Span turns a flat photo into a room you can walk around

Sight Span takes any flat image and rebuilds it as a 3D space you can navigate and reposition. Think of it like a film set: on screen you only ever see what the camera framed, but the crew builds the whole room, including the wall behind your back. Sight Span generates that missing world — objects, backgrounds, the stuff just outside the original frame — so it still looks like a natural part of the picture when you turn around.

Here's the clever part. While you're exploring, the scene looks rough and pixelated, like a blurry rehearsal. The moment you pick the angle you want, the resolution snaps back to full quality. So the tool is ugly exactly when it doesn't matter, and sharp exactly when it does.

It works with photographs, paintings and illustrations, and it preserves the original artistic style. Repositioned details — tattoos, facial features — stay surprisingly recognizable and true to the reference. Adobe isn't the first to try this; Apple has a Spatial Reframing feature, and Sight Span is being described as that idea on steroids, with more aggressive repositioning and the ability to change the shooting point entirely.

Project Screen Test lets you interview your AI character

Screen Test tackles a different problem. AI-generated characters usually feel hollow because you write them a prompt, and a prompt is a terrible place to store a personality.

Instead of typing a description, you talk to the character, like a FaceTime call with someone you're casting. You start by describing a few base traits — "anxious, shy, awkward" — and giving them a voice. You can pick a preset, describe how it sounds in a prompt, or upload sample audio for the tool to reference.

Then you keep chatting. You fill in backstory, temperament, goals, flaws, fears, even movement quirks. A Personality tab lets you go back and manually adjust anything the conversation generated. When you're happy, you describe an environment or scenario and Screen Test spits out an animated screen test — an audition tape for your character.

Adobe says the results felt more identifiable and consistent than the soulless animated AI characters you see everywhere. That's the whole point: traditional animation gives an artist control over every detail of a character, and AI video generation gives you far less.

What it means for you

You don't need to be a motion designer to care about this. Both tools solve everyday problems.

At home: see what's behind the picture

You've got a hundred family photos where the framing cuts off half the room. With something like Sight Span, you can nudge the camera sideways and see a plausible version of what was there. Old holiday snaps stop being frozen rectangles and start feeling like places. It's the difference between a postcard and standing on the beach.

At work: fewer reshoots, faster mockups

If your job involves marketing images, product shots or presentations, the boring part is always the reshoot. New angle, new background, another day of scheduling. Being able to reposition the camera on an existing shot — while the detail stays recognizable — kills a lot of that busywork. You scout the angle first, then render it.

For business: characters that don't drift

Anyone using an AI presenter or mascot knows the pain: yesterday she was warm and confident, today she's a different person with the same face. Screen Test's conversational approach keeps a character consistent — the same flaws, the same fears, the same little movement quirks — across every clip. For a small brand, that's the difference between a mascot and a stranger.

For creators and side projects: faster storyboards

Writers and illustrators can walk around their own scene before committing to a final drawing. Screen Test gives you a character you can actually interrogate: "What would you do if the plan failed?" That's a lot more useful than staring at a blank prompt box. If you want free tools to play with while these prototypes stay locked down, mykreatool.com has a stack of them.

For income: new services you can sell

Freelancers will sell both of these the day they ship. Real-estate agents want listings that feel walkable. Authors want animated character auditions for pitch decks. Etsy sellers want product photos from three angles off one shoot. The skills — knowing how to direct a camera, how to build a personality — transfer directly.

How to try it right now

Honest answer first: neither project is available to the public. Adobe hasn't announced a sign-up, a beta, or a ship date. So the free option today is to watch the demo and get ahead of the crowd.

1. Watch the Sneaks keynote (free). Adobe says a video demonstration of Project Sight Span will run during the Adobe Max Sneaks keynote. That's the clearest look you'll get at how the 3D navigation actually feels.

2. Rehearse Project Screen Test with any chat assistant (free). Write three base traits for a character — say "anxious, shy, awkward." Then hold a conversation as if you're on FaceTime and ask about backstory, goals, flaws and fears. Save the transcript. You now have a personality document you can reuse anywhere.

3. Try the reframing idea by hand (free). Take one photo. Ask a free AI image editor to fill in what's just outside the frame, then crop a new angle. You'll immediately understand both the appeal and the limits.

4. Watch for the beta. Keep an eye on Adobe's Firefly and Max channels. Sneaks that get traction usually graduate into real products; the quiet ones don't.

5. Pick your first use case now. Decide today which photo or which character you'd test first. When access opens, you'll move while everyone else is still reading the announcement.

Upsides and what changes

The big shift is that images stop being flat. A picture becomes a place with a camera you can move, which changes how you shoot, edit and present everything from listings to lesson slides.

The second shift is character consistency. Talking to a character to define it is a genuinely better interface than writing a spec sheet, because people describe personality through conversation. It's also more fun, and fun tools get used.

The third is speed of iteration. Rough preview while you explore, full quality when you commit — that's the right way to work with anything heavy.

Limitations

Let's be blunt. These are experiments, and plenty of Adobe sneaks never become products — there's no release date, no pricing, and no confirmation that either tool will appear inside the Creative Cloud apps you already pay for. Sight Span's 3D scene is genuinely rough while you're moving around inside it: pixelated, ill-defined, and only sharp once you lock an angle, so it's a scouting tool rather than something you'd show a client mid-session. Screen Test is described as a standalone creative exercise, and it's unclear how — or whether — it plugs into Adobe's AI video tools, which means your character might not travel into your actual footage. And both tools raise the usual uncomfortable questions about generating content that never existed: the person in the photo didn't stand where the tool put them, and the character you interviewed isn't a collaborator you can credit.

Conclusion

Adobe's message here is simple: a photograph is no longer the end of the process. It's a starting point you can walk into, and the people inside it can be talked to. Whether these specific prototypes ship or not, that direction is now locked in.

Your one action for today: open your camera roll, pick a single photo with room cut off at the edges, and ask yourself what's behind the frame. Then save that image somewhere obvious. That's your first test case for the day access opens.