OpenAI has opened up the DALL-E neural network API to application developers in a public beta. The company said it is already used by more than 3 million people and the neural network generates more than 4 million images per day.
According to the developers, the flexibility of DALL-E allows users to create and edit original images ranging from artistic to photorealistic. DALL-E follows natural language descriptions. As it develops, the neural network API will continue to incorporate tools to improve image quality, scalability and so on.
OpenAI noted that it is already working closely with several first-time customers who have built DALL-E into their applications and products. For example, Microsoft is using a neural network in its Designer graphic design application. The company is also integrating DALL-E into its Bing and Edge browsers as part of the Image Creator extension, allowing users to create images if they can't find what they're looking for on a search engine.
At the same time, an interface for Cala designers has been introduced. It integrates the entire design process - from product idea to e-commerce strategy to order fulfillment - on a single digital platform. Cala's new DALL-E-based artificial intelligence tools will enable users to generate design ideas based on text descriptions or uploaded reference images.
And photo startup Mixtiles is using a neural network to suggest pictures with childhood memories, for example.
OpenAI introduced DALL-E 2 in April. Since July, the company has accelerated access to the neural network for users on its waiting list. It has been opened up to around one million people. OpenAI also introduced a partial pay-per-use system of credits.
In September, OpenAI introduced Outpainting for DALL-E, which allows it to refine the background of a source image. With the new mode, it is possible to complement known paintings and develop new stories on them.