You can now edit DALL-E images in ChatGPT across the web, iOS, and Android

ChatGPT has received an update that introduces image editing capabilities. This new feature allows users to refine images generated within the platform.

Users can now refine their DALL-E creations by adding, removing, or adjusting elements within the image. This is done through a selection tool that allows users to choose a specific part of the image and provide a text prompt describing the desired changes. DALL-E will then use this information to generate a new version of the image reflecting the edits.

For instance, if a user generates an image of a cat wearing a hat and then decides they would prefer a dog instead, they can select the cat and provide a new prompt specifying a dog. DALL-E would then generate a new image with a dog wearing the hat.

Here’s how it works:

1. Users click on a generated image to expand it.
2. A paintbrush tool allows users to highlight specific areas for adjustment.
3. Users can then type a new prompt describing the desired changes (e.g., “add bows”).
4. The AI will generate a new version of the image, incorporating the edits.

This image editing functionality is powered by OpenAI’s DALL-E 3 model, unveiled in September 2023. OpenAI’s models, including DALL-E 3, are currently facing legal challenges regarding the use of potentially copyrighted material in their training data.