The short version
- Project Screen Test enables users to develop consistent AI character personalities through conversational auditions rather than static text prompts.
- Project Sight Span transforms two-dimensional images into navigable three-dimensional environments, generating plausible background details outside the original frame.
- New audio and editing utilities include a visual interface for mixing sound effects and a feature that copies specific retouching adjustments to other subjects in a photo.
Adobe has introduced a suite of experimental artificial intelligence tools designed to expand creative control over video, image, and audio production. Announced during the company’s annual Max conference previews, these prototypes address specific limitations in current generative workflows, particularly regarding character consistency, spatial depth, and repetitive editing tasks. The tools represent a shift toward more interactive development processes, allowing creators to refine outputs through dialogue and direct manipulation rather than relying solely on initial text descriptions.
Project Screen Test focuses on the creation of animated characters for video content. Traditional AI video generation often struggles with maintaining consistent personality traits across different scenes. This new tool attempts to solve that by treating character development like a casting audition. Users begin by defining basic attributes, such as anxiety or shyness, and selecting a voice profile from presets, text descriptions, or uploaded audio samples. The system then allows the creator to converse with the digital persona in real time, similar to a video call interface.
Through this conversational process, users can elaborate on the character’s backstory, temperament, goals, and physical quirks. This interactive method aims to produce a more identifiable and coherent personality than standard prompt-based generation. Once the persona is established, the tool generates animated screen tests placing the character in specific environments or scenarios described by the user. While the results demonstrate improved consistency compared to other AI-generated characters, the feature currently exists as a standalone creative exercise. It remains unclear whether this technology will be integrated into Adobe’s existing video production suite.
Project Sight Span addresses the flat nature of standard photography and illustration by converting them into explorable three-dimensional spaces. The tool generates objects and backgrounds that extend beyond the original image frame, creating a sense of depth that allows users to reposition their viewpoint. This capability goes beyond simple parallax effects; it constructs a navigable environment where users can look behind subjects or change the shooting angle entirely. The system is designed to preserve the artistic style of the source material, whether it is a photograph, painting, or digital illustration.
The visual quality of these generated spaces varies depending on the user’s interaction. When navigating the 3D environment, visuals may appear pixelated or ill-defined in areas not directly focused on by the viewer. However, resolution and clarity are restored when a specific angle is selected for final output. The tool demonstrates an ability to maintain recognizable features, such as facial details and tattoos, even when the viewpoint shifts significantly. This functionality resembles advanced spatial reframing technologies but offers more aggressive repositioning capabilities, though it raises questions about whether the resulting images retain the integrity of traditional photography.
In the audio domain, Adobe showcased Project Sound Bender, a mixing tool that simplifies sound design through a visual interface. Users upload a video clip and describe desired audio effects via text prompts, such as bubbling sci-fi noises. The system generates corresponding sound effects and automatically synchronizes them with the video timeline. A square canvas allows for real-time adjustment of these sounds, with each corner representing a different AI-generated audio variation. Creators can drag their cursor across this canvas to blend or emphasize specific noise profiles, offering intuitive control over complex audio mixes without requiring extensive technical knowledge.
Project Rinse Repeat targets the tedious nature of repetitive photo editing tasks within Photoshop. This feature allows users to copy specific edits applied to one subject and automatically adapt them to other objects or people in the same image. For instance, if a user manually removes blemishes from one person’s face, the tool can apply similar retouching to others in a group shot, adjusting for differences in shape and size. It can also replicate lighting adjustments, such as adding shadows to products in marketing graphics. Crucially, these automated adaptations remain fully editable using native Photoshop tools, ensuring that creators retain granular control over the final output.
These experimental features highlight Adobe’s ongoing effort to bridge the gap between generative AI and professional creative workflows. By introducing interactive elements like conversational character development and visual audio mixing, the company is moving beyond simple prompt-to-output models. The integration of these tools into existing software ecosystems remains uncertain, as they are currently presented as standalone prototypes. Their eventual adoption will depend on how well they balance automation with the precise control required by professional designers and editors.
The broader implications of these tools extend to the future of content creation standards. As AI becomes more capable of generating consistent characters and realistic spatial environments, the definition of original media may continue to evolve. Tools like Sight Span challenge traditional notions of photography by allowing significant alteration of perspective and context after the fact. Similarly, Screen Test’s ability to generate nuanced personalities raises questions about authenticity in digital storytelling. Adobe’s approach suggests a future where AI serves as a collaborative partner in the creative process, rather than just an automated generator.
Further details regarding these projects are expected to be shared during the upcoming keynote presentations at the Adobe Max conference. Until then, the practical application and potential integration paths for these tools remain speculative. The company has not yet announced release dates or pricing structures for any of these features. Observers will be watching to see if these experiments transition from novelty demonstrations to essential components of the creative industry’s technical toolkit.
Sources behind this briefing
Go to the original reporting
- The Verge↗Adobe’s AI experiments let you walk around in photos and interview your characters
