Adobe MAX 2026: Adobe Research puts more control in creators’ hands with new AI tools
Designers, marketers, artists, and influencers from around the world gathered at this year’s Adobe MAX to glimpse the newest innovations in creative technology. Among the most exciting announcements were new tools and in-development technology powered by Adobe Research. Highlights included AI tools for adjusting lighting in images and creating music and sound effects right inside a video timeline, and an experimental creative experience that mimics how paint behaves in the real world.
New generative AI tools for images and audio give users more control
Adobe Research is behind several of the new generative AI tools announced at this year’s MAX. These tools give users more control as they generate and refine audio and images inside Adobe products.
Relight, a new Generative AI feature in Photoshop Desktop, lets users edit the lighting conditions of an image by adding up to three interactive lights to a scene. The tool operates as a relight layer, allowing users to test the position, intensity, color, and diffusion of the light before activating the Generative AI function. Relight is available now in Photoshop.
Generative Media Tools in Premiere Pro
Adobe Premiere Desktop brings AI-powered audio generation directly into the editing timeline with its new Generative Media Tool. Creators can generate individual sound effects from text prompts, synchronize them with on-screen actions, and refine their results all without leaving their editing workflow.
One of these capabilities is Generate Music, also available for free in Adobe Firefly. The feature utilizes video footage to create custom music tailored to each creator’s story. Users can write their own prompts or generate one based on their video footage, then listen, revise, and fine-tune the generated tracks to match the mood and pacing of their videos. The commercially safe music can be generated and added directly to the Premiere timeline, making it easier to find the right soundtrack without interrupting the creative process.
Another capability is Generate Soundscape, co-developed by Adobe Research and the Premiere Desktop team. The feature intelligently creates four synchronized, editable audio tracks by analyzing the actions and atmosphere of a video, giving creators greater control over their sound design.
Explore the interactive demos below to hear how these generated sounds come together and experiment with individual audio layers.
Explore the soundscape
Explore Generate Soundscape in action here. Choose from three scenes, each featuring four individually generated audio tracks, and use the mute and solo controls to isolate specific sounds or hear how they come together to create a complete soundscape.
Explore another soundscape
More MAX highlights from Adobe Research
Adobe Research contributed to more than 50 technical features and product innovations showcased at this year’s MAX. Here are some of the highlights:
- EnhanceAudio is a new feature that gives users control over all audio elements in a video, including the ambience, the music, the sound effects, and the dialogue, and lets them rebalance every item as needed. The feature is available in Adobe Firefly and Premiere.
- Vary / Remix, which is inside Firefly Boards, gives users more control of generative edits by letting them name one or two custom dimensions that they want to change, such as mood or style. Only those dimensions change, while everything else remains as close to the source as possible.
- Separate Crosstalk fixes a unique production situation when two speakers talk over each other at the same time on a single recording. Users can now isolate and solve the crosstalk moment rather than having to rerecord.
- The team behind Turntable, a tool that allows users to rotate 2D objects as if they were 3D, has revised the model to support near-continuous rotation, stronger consistency between different views, and better handling of detail. It is available in Adobe Illustrator.
- In Firefly Graph, the team released Color Extraction Node, which divides an input image into subregions and samples pixels from each one to better match human color perception.
- The Adobe Research Music AI team released Beat Detection in the Firefly AI Video Editor. Upload your video and images, add music, and sync everything to the beat.
MAX Sneaks give an early look at groundbreaking new creative tools in development
Every year, Adobe MAX audiences are treated to Sneaks, a peek behind the curtain at experimental technologies that could make their way into Adobe products. This year Adobe Researchers took the stage with celebrity guest, actor, and host of America’s Got Talent, Terry Crews, to share boundary-pushing new tools, from a feature that can edit every shot in a video based on a single text prompt to technology that transforms an image into a digital world you can explore.
Traditional copy and paste allows Adobe Photoshop users to copy pixels, but Project Rinse Repeat copies Photoshop edits and applies them intelligently wherever they’re relevant.
It understands the context of each target, then selects and adapts the appropriate edits rather than blindly repeating the same operations everywhere. It can even recognize patterns across examples and extend them to new designs. All edits use native Photoshop tools, layers, and masks to keep the results non-destructive and fully editable.
In today’s vector tools, dragging one object onto another only determines where it lands. Project Link Up turns the drag gesture into semantic fusion, combining an understanding of both objects with drag path, velocity, and creative controls to shape how they influence each other in real time, all while keeping the result fully editable as vector artwork.
Edit long-form, multi-shot video from a single text prompt or one edited frame, and see the changes applied consistently across each shot, cut, and camera angle. You can recast characters, relocate stories, and restyle entire sequences while maintaining the original performance and pacing. Project Scene Sweep also supports set-to-set editing, allowing a single edit to be propagated across a gallery of related clips.
Project Deep Brush separates images into physically decomposed layers, giving artists new forms of control when painting, retouching, and designing lighting. It feels like applying paint in the real world, respecting shadows, geometry, and lighting. This unlocks new painting experiences, such as controlling where paint lands and making lighting-aware edits.
Existing generative tools are good at replicating what a character looks and sounds like, but not who they are. Project Screen Test brings personality to a character through natural, direct conversation.
Direct a character before a single frame is generated: give it a face, a voice, and a personality, then have a real conversation as it comes alive and auditions for you on the spot. This is casting, directing, and character development reimagined for AI filmmaking.
Edit an image by working on it directly. Project Hands On analyzes what’s in the image and builds controls right on the canvas, such as control points on a mug’s handle, so you can reshape details by dragging instead of rewriting a text prompt.
Today’s video generators let you control “what” appears, but don’t provide much control over “when” things happen. Project On Cue gives you control over “when”: a simple, editable timing signal that sets the exact moment things move, flash, or hit the beat, in both video and audio.
Project Sight Span transforms a single image into a fully interactive, navigable world. Creators can step inside, explore freely, frame shots beyond the original image, and render new images or camera-controlled videos.
Project Live Set turns video into a live, directable canvas. In one continuous experience, creators can generate, edit, restyle, add reference subjects, propagate changes, and redirect the camera while the video is playing.
Project Sound Bender turns audio into something you can shape: pull any sound out of a mix by describing it in a text prompt, then reshape it on an interactive Sound Canvas of semantic variations in real time.
You can catch everything you missed at Adobe MAX 2026 here.
Wondering what else is happening in Adobe Research? Check out our latest news here.