It’s only been a few weeks since OpenAI began allowing customers to commercially use images created by DALL-E 2, its remarkably powerful AI image synthesis system. But despite the current technical limitations and lack of volume licensing, not to mention the API, some pioneers say they are already testing the system for various business use cases – waiting for the day when DALL-E 2 becomes sufficiently stable to be deployed in production. .
Stitch Fix, the online service that uses recommendation algorithms to personalize clothes, says it has experimented with DALL-E 2 to visualize its products based on specific characteristics such as color, fabric and style. For example, if a Stitch Fix customer requested “high-waisted, red, stretchy, skinny jeans” during the pilot, DALL-E 2 was leveraged to generate 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 highlight the most informative features of a product visually, ultimately helping stylists find the perfect item that matches what a customer has requested in their written reviews,” a spokesperson told TechCrunch via email.
Of course, DALL-E 2 has quirks, some of which give first-time enterprise users pause. Eric Silberstein, vice president of data science at e-commerce startup Klaviyo, shares his mixed impressions of the system as a potential marketing tool in a blog post.
He notes that facial expressions on human models generated by DALL-E 2 tend to be inappropriate and that muscles and joints are disproportionate, and that the system does not always fully understand instructions. When Silberstein asked DALL-E 2 to create the image of a candle on a wooden table against a gray background, DALL-E 2 sometimes erased the cover of the candle and melted it into the office or added an incongruous edge around the candle.
“For photos with humans and photos of human modeling products, it could not be used as is,” Silberstein wrote. Still, he said he would consider using DALL-E 2 for tasks such as giving starting points for edits and passing ideas to graphic designers. “For stock photos without humans and artwork without specific branding guidelines, DALL·E 2, to my layman’s eye, could reasonably replace ‘the old way’ right now,” Silberstein continued.
Cosmopolitan editors came to a similar conclusion when they teamed up with digital artist Karen X. Cheng to create a cover for the magazine using DALL-E 2. The Cover Arrival finale required a very specific prompt from Cheng, which the writers say is illustrative of the limitation of DALL-E 2 as an art generator.
But the AI weirdness sometimes works – as a feature rather than a bug. For its Draw Ketchup campaign, Heinz asked DALL-E 2 to generate a series of images of ketchup bottles using natural language terms such as “ketchup”, “ketchup art”, “fuzzy ketchup”, ” ketchup in space” and “ketchup renaissance”. The company invited fans to submit their own prompts, which Heinz curated and shared on his social media.
“With AI imagery dominating the news and social feeds, we saw a natural opportunity to expand our ‘Draw Ketchup’ campaign; rooted in the idea that Heinz is synonymous with the word ketchup – to test this theory in the AI space,” Jacqueline Chao, Heinz Senior Brand Manager, said in a press release.
Clearly, DALL-E 2 based campaigns can work when AI is the subject. But several professional DALL-E 2 users say they’ve used the system to generate assets that don’t bear the telltale signs of AI constraints.
Jacob Martin, a software engineer, used DALL-E 2 to create a logo for OctoSQL, an open source project he develops. For around $30 — roughly the cost of logo design services on Fiverr — Martin ended up with a cartoon image of an octopus that looks like a human illustration to the naked eye.
“The end result isn’t ideal, but I’m very happy with it,” Martin wrote in a blog post. “As for DALL-E 2, I believe that at this time it is still in a ‘first iteration’ phase for most elements and purposes – the main exception being the pencil sketches; these are amazingly good… I think the real breakthrough will come when DALL-E 2 is 10 to 100 times cheaper and faster.
One DALL-E 2 user, Don McKenzie, head of design at development startup Deephaven, took the idea one step further. He tested the application of the system to generate thumbnails on the company’s blog, motivated by the idea that posts with images get significantly more engagement than those without.
“As a small team of mostly engineers, we don’t have the time or budget to commission custom artwork for each of our blog posts,” McKenzie wrote in a blog post. “Our approach so far has been to spend 10 minutes scrolling through tangentially related but ultimately mismatched images from stock photography sites, uploading something lousy, highlighting it, and posting.”
After spending a weekend and $45 in credits, McKenzie says he was able to replace about 100 blog posts with DALL-E 2-generated images. It took some experimenting with the prompts to get the best results. , but McKenzie says it was worth it.
“On average, I’d say it took a few minutes and about four to five prompts per blog post to get something I was happy with,” he wrote. “We were spending more money and time on stock images per month, with worse results.”
For companies that don’t have time for brainstorming prompts, there is already a startup trying to commercialize the asset generation capabilities of DALL-E 2. Unstock.ai, built on DALL-E 2 , promises “high-quality images and artwork on demand” — at no cost, for now. Customers enter a prompt (eg, “Top view of three goldfish in a bowl”), then choose a preferred style (vector art, photorealistic, pencil) to create images, which can be cropped and resized.
Unstock.ai essentially automates rapid engineering, an AI concept that seeks to embed a job description into text. The idea is to provide an AI system with step-by-step instructions so that it reliably does the thing that is asked of it; In general, results from a prompt like “Still image of woman drinking coffee, walking to work, telephoto” will be much more consistent than “Woman walking.”
This is probably a harbinger of applications to come. When contacted for comment, OpenAI declined to share DALL-E 2 business user numbers. But anecdote, the demand seems to be there. Unofficial workarounds to DALL-E 2’s lack of APIs have sprung up on the web, chained by developers eager to integrate the system into apps, services, websites, and even video games.