ChatGPT's New Sketch Tool Turns Your Doodles into Realistic Pictures

OpenAI just made it much easier to tell an AI what picture you want. With the new Sketch feature inside ChatGPT, you don't need to find the perfect words. Just open a small drawing window, scribble your idea — even a messy doodle — and type something like "make this a realistic photo." A reporter tried it with a sloppy mouse-drawn cat, and ChatGPT turned it into a lifelike image. You can even refine details by commenting directly on a part of the image, like changing the cat's eyes from blue to green.

This is part of a bigger upgrade called ChatGPT Images 2.5. OpenAI says the new version produces images with more natural lighting and richer textures than before. It also claims the tool is more obedient when you ask for changes across multiple rounds, and that generating an image is now up to 50 percent faster. For everyday users, that means less time waiting and more time experimenting with ideas.

Who is this for? Anyone who has ever had a visual idea but couldn't draw, doesn't use Photoshop, or can't afford a designer. You could sketch a new layout for your living room, a logo concept, a gift idea, or even a scene from your book. It's like having a designer-in-training that happily works from your crude sketches and verbal suggestions. The tool is rolling out now for ChatGPT, ChatGPT Work, and Codex users across desktop, mobile, and web.

The main limitation is that it's still AI: results may be quirky, and complex sketches can lead to strange details. But as a way to quickly bring a rough vision to life, this lowers the barrier for creative work dramatically. If you can scribble and talk, you can now produce images that look polished.