At this year's Adobe Max Conference, Adobe showcased a series of AI experimental tools still in development, offering photographers, videographers, and audio creators a smarter and more intuitive editing experience. These tools are collectively known as "Adobe Sneaks", demonstrating the potential future direction of creative productivity — from one-click light reconfiguration to intelligent voice correction, AI is completely redefining the creative process.

"Frame Forward": Making Video Editing as Simple as Photo Editing

One of the most impressive new features is "Frame Forward" (Project Fast Frame). This tool allows editors to add, remove, or replace content in a video without using complicated masks.

In Adobe's demonstration, they showed removing a woman from the first frame of a video. The AI automatically identified the subject and filled in a natural background — similar to Photoshop's "Content-Aware Fill," but this modification automatically extends throughout the entire video.

Even more surprising is that users can also insert objects by drawing areas on the screen and entering prompt words. The AI will automatically match the lighting and perspective of the scene, for example, adding a puddle to a video that even reflects the reflection of a cat walking over it, showing a high level of physical and semantic understanding.

Project Light Touch: AI Light Magician

Project Light Touch allows users to control photo lighting as freely as adjusting studio lights. Using generative AI, users can redefine the direction, diffusion, and color temperature of light, achieving lighting effects that do not exist in the original image.

For example, you can make a room appear illuminated by unlit lights, or drag a virtual light source to move around a person. Even with "dynamic lighting," you can switch between day and night or create holiday atmospheres.

The precise light control capability of Light Touch enables photographers and designers to achieve near-physical realistic lighting reconfiguration during post-production, greatly increasing creative freedom.

Project Clean Take: Redefining Audio Restoration

In the audio field, Adobe introduced Project Clean Take — an AI tool that can automatically correct speech pronunciation and emotional expression.

Creators don't need to re-record; they can adjust the speaker's tone, rhythm, or emotion through text prompts, such as making the voice sound more natural, cheerful, or curious.

The tool can also automatically separate ambient noise and voice tracks, allowing users to adjust or mute specific sound sources individually, maintaining voice clarity while preserving the background atmosphere.

This means video creators or podcast producers will be more efficient when editing audio, and even directly "edit emotions."

More AI Experiments: From Material Replacement to 3D Visual Reconstruction

Aside from the three major tools mentioned above, Adobe also showcased several innovative projects in the visual field:

  • Project Surface Swap: Instantly change the material and texture of objects or surfaces;

  • Project Turn Style: By rotating objects in an image, you can view and edit them from different angles, like 3D models;

  • Project New Depths: Give flat images depth information, identify object occlusion relationships, and achieve an editing experience similar to 3D space.

These Sneaks demonstrate Adobe's cutting-edge exploration in generative AI, visual understanding, and spatial perception.

Future Outlook: From Experimentation to Standards

Currently, all the above Sneaks are internal preview features, and it is not yet determined whether or when they will be integrated into Adobe Creative Cloud or Firefly product lines. However, historical experience shows that many widely used features today — such as Photoshop's "Remove Distractions" and "Harmonize" — were initially born from Sneaks projects.