How small and medium-sized businesses can benefit from text-to-image AI

Getting the chance to try out DALL-E 2, OpenAI’s new AI system that can create realistic images from natural language, was quite extraordinary.

There’s no doubt that the system is still in its infancy, but it’s clear that the technology is changing rapidly and we’re already starting to see improved text-to-image models. Google Brain’s Imagen, which can generate photorealistic images of a scene given a text description, and Meta’s Make-A–Scene, which allows users to draw a free-form digital sketch to accompany a text prompt, are two examples. promising.

This technology isn’t just for big tech companies. There are many ways today for small and medium-sized businesses to take advantage of text-to-image technology.

On average, small businesses are recommended to spend 7-8% of their gross revenue on marketing. And yet, many only spend 3 to 5%. Tools like DALL-E 2 can allow entrepreneurs to punch above their weight even if they don’t have the luxury of hiring talent to produce customizable brand graphics.

As the saying goes, a picture is worth a thousand words. As DALL-E 2 and other text-to-image systems become more widely available, it only makes sense that marketers start using them for their campaigns.

Take, for example, a brand that sells fresh seafood. While it makes sense to include attractive images in your emails, there’s no real benefit to doing your own photo shoots over using stock photos. Commercial-use photo catalog subscriptions can easily cost hundreds of dollars a month, which is hard to justify for smaller brands.

Enter DALL-E 2. The model allows marketers to combine the creativity of hiring an in-house graphic designer with potential price savings. Here, for example, is what DALL-E 2 produces when it asks for “a photo of an appetizing salmon with slices of lemon”.

While these images are indeed appetizing, what happens when the salmon supply drops unexpectedly and our supply this week changes from salmon to cod? For small businesses that need to pivot at all times, the ability to generate new images in 30 seconds is invaluable.

Graphic design by DALL-E 2

DALL-E 2 entered the popular consciousness not because of its ability to faithfully reflect the world, but because of its ability to create wonderfully stylized images.

Consider this watercolor of a panda wearing a hat. The image is playful, well structured and does not exist in any photo catalog. That’s why DALL-E 2 is exciting — it opens up new opportunities, especially for small businesses.

The success of tools like DALL-E 2 may discourage designers who worry about AI automating their roles, especially since some experts predict that 99.9% of online content will be generated by AI by 2030. But we think DALL-E 2 will win. t replacing jobs; rather, it will become part of a marketer’s toolkit and skill set.

The output quality of DALL-E 2 varies greatly depending on the prompts fed into it. Generating a good prompt is a very creative process, similar to writing copy itself. If the future includes marketers in hybrid roles using these tools for inspiration, it will increasingly become a domain skill.

DALL-E 2: Repainting

One of the most interesting features offered by DALL-E 2 is automatic inpainting. Think of it as a magic wand that marketers can wave to edit images however they want.

This is best seen through example. Here DALL-E 2 received the first frame and said to insert a corgi in a specific place. The model understood the context of the image well enough to understand that he was being asked to render the dog inside a painting and was able to compose in a style that matched.

This fascinating feature holds a lot of value, especially for brands that want to capture their products in different places or unique scenarios. Imagine placing your product in sets you can’t visit or see what George Washington would have looked like while sipping a Coca-Cola.

DALL-E 2 can be a remarkable game-changer for marketers, where they will be more constrained by creativity than time and budget.

Limits

While there are reasons to be excited about the future, current technology still has limitations. It struggles to generate photorealistic people and often fails to generate coherent text. Additionally, there have been some significant criticisms of DALL-E 2 on the issue of bias. For example, when tasked with generating the “portrait of an intelligent person”, DALL-E Mini – a lower cost open source implementation of DALL-E 2 – generated nine photos of white men in evening.

OpenAI recently released an update aimed at mitigating some of these bias issues, but it remains to be seen if they are truly resolved. In the meantime, marketers using these tools will need to be careful to avoid unintentionally propagating bias.

Wrap

So would we want DALL-E 2, in its current form, to be a tool in our marketing toolbox? Absolutely, and for several purposes: images that can be used as generated, to give starting points for further editing; to help with brainstorming; convey ideas to graphic designers; run effective A/B tests on creatives; and to create fantastic, eye-catching images that today can only be achieved by artists. Despite its limitations, there are also opportunities.

Over the past six months, we’ve seen growing interest in revolutionary text-to-image synthesis technology. As technology continues to rapidly evolve, better models will continue to emerge and brand use cases will only grow.

Robert Huselid and Tom Dinitz are data scientists at claviyoa unified customer platform for email, text and more that empowers online brands to own their data and grow on their own terms.