Sponsored
Software

Adobe MAX 2026: Ten experimental AI tools that could change photography and video

Ten Sneaks across photography, video, audio and design, with two available to test in Adobe Labs Playground.

Πολύχρωμο επίσημο γραφικό Adobe MAX 2026 με μεγάλα γεωμετρικά στοιχεία και λευκό MAX
Adobe MAX 2026 visual identity. Image: Adobe.

Summary

  • Adobe introduced ten experimental AI research projects at MAX 2026.
  • Four focus on video, one on sound and five on imaging and design.
  • Deep Brush and Hands On are already available as research previews.
  • Adobe has not announced release dates for the others.
Contents
  1. Four projects rethink video editing
  2. Sound: Project Sound Bender
  3. Five experiments for images, photography and design
  4. Which projects can you try today?
  5. What we think
  6. Frequently asked questions

Adobe has unveiled ten experimental AI tools at MAX 2026, offering a glimpse of how photographers, filmmakers, audio creators and designers might edit their work in the future.

At the November 11 Sneaks presentation, hosted by Terry Crews, Adobe researchers demonstrated technologies built around brushes, gestures and direct manipulation rather than endless text prompts. These are research prototypes, not ten newly released Creative Cloud features.

Two projects, Deep Brush and Hands On, are already accessible as research previews in Adobe Labs Playground. Adobe has not announced whether or when the remaining concepts will reach commercial applications.

Four projects rethink video editing

Project Scene Sweep: A single creative change to a subject, look or location needs to remain consistent across the other shots in the same scene. Scene Sweep identifies characters and environments across cuts and attempts to carry one editing decision through different camera angles. Anyone editing a multi-shot sequence knows why that matters: fixing a single clip rarely finishes the job. Unlike the existing beta Premiere Generative Remove, Scene Sweep is an experimental approach to editing an entire sequence.

Project Screen Test: This experiment focuses on fictional characters. A creator defines a character’s appearance, voice and personality, then talks to the character to explore how it might respond in different situations. The concept is particularly relevant to animation, narrative development and preproduction, where behavior matters as much as visual consistency.

Project Live Set: A creator can change the subject, transform the setting or redirect the camera while a video keeps playing. The idea is closer to directing an evolving scene than submitting a fixed generation request before the action starts.

Project On Cue: Timing takes center stage. Creators draw an editable timing signal, or extract one from video or music, to specify when an event such as a flash of light should happen. For work that needs precisely synchronized sound and visuals, this could be more useful than a prompt that only describes what should happen.

Sound: Project Sound Bender

Project Sound Bender treats audio as something that can be isolated and shaped more directly. Adobe says creators can describe a sound to extract it from a finished mix, experiment with variations on an interactive canvas and match its movement to video. In sound design, that could shorten the distance between locating the right sonic element and creating the final effect.

Five experiments for images, photography and design

Project Deep Brush: This AI brush understands the shape, colors and lighting of the image. Brushstrokes can follow curved surfaces, recolor elements without destroying their natural shadows and paint light into a scene. Rather than asking a model to “change the lighting,” creators can work on a specific area much as they would with a conventional brush.

Project Sight Span: A still image becomes a space that can be explored to find different angles and compositions beyond the original frame. Its concept is related to the Photoshop Camera Angles feature, but Sight Span is presented as navigation through a larger environment. The new viewpoints are synthetic, not photographs actually recorded from those positions.

Project Hands On: Object segmentation and vision-language models generate controls directly over the image. Creators drag and reshape elements in the scene instead of describing each movement in a prompt. This is one of the two experiences Adobe has opened for testing.

Project Rinse Repeat: The system learns edits made to one part of an image and reapplies the process to another, adapting it to the new content. Crucially, it recreates the result with native Photoshop tools, leaving the edits available for further adjustment. That is promising for repetitive retouching where copying pixels cannot solve the problem.

Project Link Up: In design, a drag gesture can carry creative intent. Move an illustrated car into a parking space, for example, and the project can adjust its scale, perspective and style to fit. The output remains editable vector artwork.

Which projects can you try today?

Adobe has opened research previews of Project Deep Brush and Project Hands On through Adobe Labs Playground. It also offers a preview of Project Custom Filters, a separate Photoshop-related experiment outside the ten Sneaks. The remaining projects currently have demonstrations rather than announced general-release dates or prices.

What we think

Rinse Repeat and Scene Sweep stand out as the most immediately relevant ideas for professionals. Whether you are processing dozens of photographs from one assignment or several shots from a sequence, repeating the same creative decision costs time. Carrying an edit across the material while maintaining consistency could make a genuine difference.

The move away from prompts and towards on-image controls is just as compelling. A photographer knows exactly what “a touch more light here” or “move this object slightly to the right” means. When software understands a brushstroke or a gesture, and still lets the artist refine the result, AI starts to feel more like a practical retouching tool.

There is a necessary distinction, though. Sight Span could be excellent for storyboarding and advertising concepts, but a synthetic camera angle must not be passed off as documentary evidence. In professional delivery we also need to know what generates new content, which edits remain editable and how the original capture is preserved. Being able to try two demos is valuable because creators can test that interaction now; it does not mean ten finished features have shipped.

Frequently asked questions

Are all ten tools now part of Photoshop or Premiere?

No. Adobe MAX Sneaks showcase experimental research technologies. Their appearance on stage is not an announcement that the ten features have shipped in commercial applications.

Which experiments are already available to try?

Adobe offers research previews of Project Deep Brush and Project Hands On in Adobe Labs Playground. Availability and access conditions may evolve as the research progresses.

✎Comments

Leave a comment