The connection between GPT and DALL-E by OpenAI provides users with a new more natural workflow for creating images. It removes the hurdle of learning to prompt an image model and lets even non-designers create good images in a chat session.
We’ve looked at the improvements AI Image generation can bring to design work before. At that time a designer needed to have a solid idea of the expected image and had to construct the right prompt to get the model to provide that image. This is still a valid process and provides the most control, but it might become one used by ‘the professionals’ while mere mortals will just chat with an AI-designer bot to create an image.
In a ChatGPT session using GPT4, part of the subscription package offered by OpenAI, images can be created inline while you communiciate with ChatGPT. You can communicate the image content or changes you want to see the same way you would chat with GPT about a text, summary or other text related instructions. The added interactivity makes for a natural workflow, a bit like talking to a designer, allthough at this point it is a less creative and experienced one.
The option to create specific chatbots (GPTs) and offer them in a ‘bot-store’ will also open the doors for bots that will help creating images for specific subject or fields, like promotions or blogposts, but also diagrams or cartoons. Adding more support for users that are not trained or experienced in the field.
Image recognition has also been integrated into ChatGPT, further simplifing some use cases. You can provide GPT with an image and use it as a starting point. Not only for creating an image but also for producing text based content about the objects and texts in the image.
