Adobe MAX 2026 Sneaks: From turning images into navigable worlds to AI-generated characters for videos, check out all announcements
Video projects include Scene Sweep, Screen Test, Live Set and On Cue for multi-shot editing, character development and real-time scene control.
Image tools include Deep Brush, Sight Span, Hands On, Rinse Repeat and Link Up for AI-assisted editing and design.
Sound Bender can isolate sounds from finished mixes and create variations, while Adobe Labs Playground offers previews of select experiments.
Popular creative app maker Adobe previewed a range of research innovations at its MAX Sneaks 2026 event. These experimental AI projects mainly focused on video and image, while one was showcased for sound. When available to a wider audience, these tools will transform how creators use AI to aid their editing workflow. Hosted by Terry Crews, Adobe’s blog post highlighted how these innovations will enable creators to optimise their editing process with natural language and gesture controls in future.
For instance, ‘Screen Test’ will allow creators to build characters for videos and monitor their personality and how they respond in different scenes. The ‘Sight Span’ feature will allow exploring new angles in images and recomposing them for an entirely new perspective. For better context, here are the innovations Adobe announced at Sneaks 2026 across video, sound and image verticals.
Video-based innovations:
Adobe previewed ‘Project Scene Sweep’ which can understand scenes and characters in a sequence to apply consistent edits across camera angles and clips and avoid repeating the same process. Among the most interesting innovations, ‘Project Scene Test’ will let people develop fictional characters with a customised look, personality and voice. The character will also support conversations in natural language.
‘Project Live Set’ is advertised to let creators change the ambience, alter camera angles or even the subject while playback continues in real-time. Adobe’s ‘On Cue’ will allow users to mark timings of a particular activity in a video, such as rain, flashes of light and more, in an AI video. “An editable timing signal can be drawn by hand or extracted from a video or piece of music, making it possible to bring motion and sound together on cue,” Adobe said.
Image-based innovations:
‘Project Deep Brush’, aimed at comprehending the colour, shape and other characteristics of the visual at hand, will assist creators in marking brushstrokes on the surface. “Brushstrokes can follow the contours of a surface, colours can be changed while preserving natural shadows, and light can be painted directly into a scene,” Adobe said.
Another key innovation, ‘Sight Span’, can transform a photo into a “navigable environment” and let users explore new angles or recompose shots of an image. The ‘Project Hands on’, which uses Vision-Language Models and object segmentation, will allow people to create controls for manipulating elements through a virtual hands-on approach.
‘Project Rinse Repeat’ notes the edits made to a part of an image and applies them to other parts. Lastly, ‘Project Link Up’ can automatically optimise the size, perspective and style of elements in the scene. For instance, placing an image of a car in a parking space vector can automatically adjust its size.
Sound-based innovations:
Adobe’s ‘Project Sound Bender’ is aimed at allowing creators to describe a sound, extract it from the output and explore its variations. It will allow them to shape sounds in real time, link its movement to a video or create something fresh with a prompt.
Meanwhile, interested creators can try some of the above-mentioned innovations from the Playground in research preview. These include Project Deep Brush, Project Hands On and a preview of Photoshop’s Project Custom Filters.