gen‑ai.news
← Back
Multimodal

Adobe’s AI experiments let you walk around in photos and interview your characters

Adobe’s AI experiments let you walk around in photos and interview your characters

Every year at Adobe Max, the company sets aside space for its "Sneaks" - experimental projects that may or may not make their way into shipping products. The 2026 edition includes three tools that push into territory that feels meaningfully different from the text-to-image and generative fill features Adobe has already released to the public.

The most unusual of the three is Project Screen Test, which addresses a specific pain point in AI character work: consistency and direction. Rather than writing out detailed text prompts to define how a character should speak, react, or carry themselves, creators can simply talk to the character directly. The conversation shapes the character's personality over time, more like a casting session than a prompt-engineering exercise. For anyone producing AI-driven video content, the ability to refine a character through dialogue rather than technical descriptions could make the process feel more intuitive.

The second experiment turns still photographs into explorable 3D environments. The idea of inferring depth and geometry from a single 2D image is not new - researchers have been working on it for years - but Adobe's framing around creative workflows suggests the tool is aimed at designers and video producers who want to move a virtual camera through a scene that started as a simple photo. The practical applications range from adding cinematic movement to stock imagery to building out background environments without a full 3D modeling pipeline.

The third tool tackles photo editing through a copy-and-paste model, though the specifics of exactly how it works remain limited based on available details. The general premise appears to be reducing the number of steps between seeing an edit you want to make and actually applying it. Taken together, the three projects suggest Adobe is trying to compress complex, multi-step creative processes into more direct interactions - whether that means talking, navigating, or pasting. As with all Sneaks, there is no guarantee these features will reach Adobe's commercial apps, but they tend to be reliable signals of where the company's research priorities are focused.

Read at The Verge (AI) →
Share:X

Enjoy this story? Get the next one in your inbox.

Twice a week: the most important stories in generative image and video AI, distilled into a 2-minute read.

Free. Unsubscribe any time. No spam, ever.

Your next read

What Is MAX Sneaks? Your First Look at Tools That Don’t Exist Yet
Multimodal

What Is MAX Sneaks? Your First Look at Tools That Don’t Exist Yet

Adobe MAX Sneaks is an annual event where Adobe researchers demo experimental technology that has not yet made it into any product. Some of these demos eventually become familiar tools in Premiere, After Effects, or Photoshop - others never ship at all. This year's Sneaks airs Wednesday, November 11, and is free to watch online.

Tom Hanks Played Five Characters in This 2004 Christmas Movie
Multimodal

Tom Hanks Played Five Characters in This 2004 Christmas Movie

The Polar Express, Robert Zemeckis's 2004 performance-capture film, is back in seasonal rotation - and it remains one of the more technically unusual productions of its era. Tom Hanks played five distinct characters across 38 days of work, a casting choice that was rooted in story logic as much as technical curiosity. Here is a closer look at how that decision came together and what it meant for the actors involved.

Introducing the Multimodal Knowledge Agent benchmark
Multimodal

Introducing the Multimodal Knowledge Agent benchmark

Google has introduced the Multimodal Knowledge Agent (MKA) benchmark, a new evaluation framework designed to test how AI agents reason through visually complex, multimodal tasks. Unlike simpler benchmarks, MKA focuses on the kind of layered reasoning required when an agent must draw on both visual and textual information simultaneously. The release signals growing interest in more rigorous, real-world testing standards for multimodal AI systems.