Editing With DALLE-3

While OpenAI has not yet released their voice mimicking AI engine, they have just released a new feature for DALLE-3. This gives the “designer” the chance to edit their AI-generated images. For this, we see a new menu item for editing a generated image:

We also go into a designer mode:

We can then select an area to edit with the magic wand, and give ChatGPT a command to change something:

In this case, the text is then removed:

And, let’s say we don’t want the padlock at the centre:

And when processed, it is removed:

Conclusions

DALLE-3 suffers greatly from poorly defined text. The new feature will, at least, allow a designer to remove these elements from the image.