Businesses including Stitch Fix are already experimenting with DALL-E 2

OpenAI began allowing customers to use images created by DALL-E 2 just a few weeks ago. Despite the technical limitations and lack of volume licensing, some pioneers are already testing the system for various business use cases, waiting for DALL-E 2 to become stable enough to deploy into production.

Stitch Fix says it has been experimenting with DALL-E 2 to visualize its products based on specific characteristics. If a Stitch Fix customer asked for a "high-rise, red, stretchy, skinny jean" during the pilot, DALL-E 2 was used to create images of that item, which a stylist could use to match with a similar product in Stitch Fix's inventory.

DALL-E 2 helps us surface the most informative characteristics of a product in a visual way, ultimately helping us find the perfect item that matches what a client has requested.

The second generation from Stitch Fix's pilot. The prompt was that it was soft, olive green, great color, pockets, patterned, and long cardigan. The image is from Openai.

Some quirks in DALL-E 2 are making early corporate users pause. Eric Silberstein, the VP of data science at e-commerce startup Klaviyo, wrote a post about his mixed feelings about the system as a potential marketing tool.

facial expressions on human models generated by DALL-E 2 tend to be inappropriate and muscles and joints disproportionate. When Silberstein asked DALL-E 2 to create an image of a candle on a wooden table against a gray background, DALL-E 2 sometimes erased the candle's lid and blended it into the desk.

Experiments with DALL-E 2 for visualization. The image is from Openai.

Silberstein wrote that it couldn't be used as is for photos with humans and models. He said he would consider using DALL-E 2 for certain tasks. Silberstein said that DALLE 2 could replace the "old way" for stock photos without humans and illustrations.

When they collaborated with Karen X. Cheng to create a cover for the magazine, they came to the same conclusion. The editors said Cheng prompted them to arrive at the final cover, which was illustrative of DALL-E 2's limitations as an art generator.

Sometimes the artificial intelligence works as a feature. The Draw Ketchup campaign used natural language terms like "ketchup," "ketchup art," "fuzzy" and "ketchup in space." Fans were invited to send their own questions, which the company shared on its social channels.

A part of the recent ad campaign is a depiction of a bottle. The image is from Openai.

With artificial intelligence dominating news and social feeds, we saw a natural opportunity to extend our 'Draw Ketchup' campaign.

DALL-E 2-driven campaigns can work when the topic is artificial intelligence. Several DALL-E 2 business users say they have used the system to generate assets that don't bear signs of artificial intelligence.

Jacob Martin uses DALL-E 2 to create a logo for a project he is working on. For around $30, Martin was able to create a cartoon image of an animal that looked like it was drawn by a human.

The end result isn't ideal, but I'm very happy with it The main exception is pencil sketches, which are mind-blowingly good, but I think the real breakthrough will come when D is released.

The logo was created after a number of attempts. The image is from Openai.

The head of design at Deephaven was one of the DALL-E 2 users. The idea that posts with images get more engagement than those without is what motivated him to test the system.

As a small team of mostly engineers, we don't have the time or budget to commission custom artwork for every single post. Our approach so far has been to spend 10 minutes scrolling through tangentially related but ultimately ill-fitting images from stock photo sites, download something not terrible, and hit publish.

He was able to replace 100 or so posts with DALL-E 2-generated images after spending a weekend and credits. It took a lot of work, but it was worth it.

He said it took a couple of minutes and a few questions to get something he was happy with. Spending more money and time on stock images resulted in a worse result.

There is a startup trying to commercialise DALL-E 2's asset generating capabilities. Unstock.ai promises high-quality images and illustrations on demand for no charge at the moment. Customers enter a prompt (e.g., "Top view of three goldfish in a bowl") and then choose a preferred style to create images, which can be cut and pasted.

Prompt engineering is a concept that uses artificial intelligence to put a task description in text. In general, the results for a prompt like "Film still of a woman drinking coffee, walking to work, telephoto" will be more consistent than "A woman walking."

It is likely a sign of things to come. DALL-E 2's business users were not shared by Openai. It looks like the demand is there. Unofficial workarounds to DALL-E 2 have sprung up across the web, strung together by developers who want to build the system into apps, services, websites and even video games.