OpenAI Offers Early Look at DALL-E API, Shows Text to Image Use Case

The DALL-E API won’t be officially announced until later this fall, according to openAI, but today the company shared data about a customer already using the DALL-E API for a specific business use case.

Based in New York City Calaa startup that bills itself as “the world’s first fashion operating system,” provides a digital platform (including a mobile app launched in March) that allows makers to design and manufacture clothing lines, unifying the process from product idea through order fulfillment. With the addition of DALL-E powered text-to-image generating tools, users will be able to generate new visual design ideas from natural text descriptions or uploaded reference images – which the company says are the first of its kind for the fashion industry.

“From the moment we saw DALL-E emerging in the wild, we knew it was a very good fit for our business and our way of working,” said Dylan Pyle, CTO of Cala, adding that the implementation of the DALL-E E API just happened in the last couple of weeks. “We really see this as elevating human designers — we help you turn these ideas into pretty detailed explorations of what you’re trying to make… streamline and make that whole process faster, more effective, and more efficient.”

Luke Miller, DALL-E API product manager at OpenAI, said the research company already has a large developer base using its APIs, so it has reached out to offer the DALL-E API to specific companies.

“It was a bit opportunistic, because we find creative and interesting uses as we test the product,” he said. “Our team was super excited to work with Cala on this very specific use case, making their creative process kind of super powerful and building it into a real business application.”

How Cala uses the DALL-E API

To use the DALL-E powered tools, a user selects from dozens of product templates, such as a hoodie, dress, or jacket, and adds terms like “dark, delicate, and velvet” to an adjectives section and phrases like “ sewn logo patches” in a section for embellishments and features.

Cala then generates six sample product designs. The user can continue to regenerate designs based on the original prompt or continue to modify a particular design further. Creators can also upload their own designs, and DALL-E will return six images with slightly different variations.

Pyle pointed out that Cala sees the DALL-E API as a way to stimulate the creative inspiration process, whether the creator is an experienced designer or not. “We’re really into designing and making it happen, and if we can make it easier to get to that moment of inspiration, that’s great for us.”

Miller added that the DALL-E API allows developers to leverage the DALL-E technology and build custom solutions specific to their applications.

“We want to build a tool that is flexible enough for them to build specifically for their customers’ needs,” he said. “So in this case [it’s about] enabling end-users to come up with ideas and variations, create an image and generate a lot of different versions of it – to tweak and tweak things to their specific needs and let their creativity work with it.”

Guardrails around quick results

The Cala developers don’t allow fully open DALL-E results — instead, they refined where the inputs might take users in each product category, Pyle said.

“Obviously the model that generates the images at the end of the day is DALL-E… but we convert that into prompts that we’ve developed for each product category to drive the DALL-E results in the way we think is the most makes sense,” he said. “We were amazed at how easy it is… you still have to have such creative control over the inputs and interpret the outputs wisely. But with just a little bit of direction, you can achieving really great results, which clicked for our team almost immediately.”

When asked if Cala’s users can use prompts such as popular designer names or logos, Miller replied that there are guardrails around how users can enter DALL-E prompts and that the API follows OpenAI. content policy – which bans content related to various categories, including hate, harassment, violence, sexual and political.

“That’s a valuable asset to go through all these moderation and security questions and pass it along to our developers so it’s built into the experience,” he said. The Cala implementation is “a very specific use case that limits space and relates specifically to the creative process.”

“We are certainly not interested in encouraging or enabling such approaches,” Pyle added. “We try to focus on the kind of design elements that make your ideas unique.”