OpenAI's ChatGPT Images 2.5 adds Sketch doodle input

OpenAI's announcement of ChatGPT Images 2.5 comes with a new Sketch feature, marking a shift in how users can interact with the image generation tool. Instead of relying solely on text prompts, users can now provide a visual foundation—a simple doodle—and then layer on text instructions to refine the output. The Verge's test shows a rough cat drawing being transformed into a realistic photo, demonstrating a more accessible and intuitive creation process.

This update also improves the underlying image model. OpenAI claims that ChatGPT Images 2.5 produces images with more natural lighting and richer textures. It is also reportedly better at following editing instructions across multiple interactions, which is a significant upgrade for users who need to make iterative changes. The reduced latency, cut by up to 50% compared with the previous version, Images 2.0, points to a faster and more seamless experience.

The addition of Sketch could potentially make ChatGPT a more versatile tool for visual brainstorming and rapid prototyping. The ability to draw a rough concept and have the AI interpret it into a polished image bridges the gap between a fleeting idea and a concrete visual representation. The long-term impact hinges on how well the tool handles ambiguity in drawings and complex instructions during the editing process, which are common pain points in AI image generation.