AI design platform Lovart has announced the launch of the Touch Edit touch function, which attracted over 100,000 creators to experience within 48 hours of its public test. The feature centers on "Mark & Edit": on Mac, Cmd+Click and on Windows, Ctrl+Click automatically identify and lock elements such as hair, clothing, and text. Users can input natural language like "change to a vintage cheongsam" or "warm tone side light," and the model instantly completes the integration at 4K resolution without manual masking or parameter adjustment.

In terms of cross-image mixing, Select & Remix supports uploading multiple photos simultaneously. The system automatically numbers key areas, allowing users to drag and reassemble them like "puzzles"—a jacket from image A, background depth from image B, and lighting effects written into image C. One generation can output print-ready results. The platform also opens the Edit Elements extension, splitting the final product into individual layers for fine-tuning typography, textures, and color matching.

image.png

On the technical side, Touch Edit integrates models such as GPT-4o, Flux Pro, and Sora, offering a three-mode workflow: Talk, Tab, and Tune. Talk handles semantic descriptions, Tab provides multiple variants, and Tune allows touch-based refinement of details. Official tests show that batch modifications for 12 tasks such as tags, backgrounds, and tones have been reduced from 90 minutes with traditional tools to 7 minutes, with an average of less than 3 clicks per image.

Lovart's CEO stated that the official version of Touch Edit will be fully released in January next year, with plans to add video frame-level editing and 3D material replacement features. Industry observers believe that once cross-image consistency is validated on a large scale, AI design tools may become a genuine alternative to Photoshop for the first time.