Image generative AI – gimmick or paradigm change?

Visual 1st Perspectives

September 7, 2022

Image generative AI (text-to-image or, broader, “any prompt to any visual media type”) has suddenly become the tech innovation story de jour in the last 6 weeks, culminating 2 weeks ago when word broke that an AI-generated artwork, Théâtre D'opéra Spatial, took first place in the digital category at the Colorado State Fair.

Emotions have run high, not only in the creative and tech communities, but also in the mass media, which have extensively covered this this prize and subsequent commotion. In the meantime, the hi-res text-to-image creator, Stable Diffusion publicly launched, while lacking the NSFW filters built into most other solutions.

The result? Image generative AI is now linked to porn creation, deepfakes 2.0, copyright infringement, the death of artistry, and much more.

As it is hard to keep up with, in the sidebar I’ve listed several links of recent image generative AI articles.

But, before we get there, this week’s article is a great exposé from our friend Paul Melcher, publisher of Kaptur, describing that image generative AI is not just a gimmick but has real implications and potential for the business world.

***

Image generative AI and the everyday company

By Paul Melcher

If you haven’t heard about Dalle-2 or any of its variations (Imagen, Midjourney, Wombo), you are probably on a very unplugged vacation or have found that rock where, for some reason, some people sleep under. Regardless. Generative AI is here and, while not perfect, shows promises of excellence. While not a surprise- we have been expecting something like this to emerge- the rapidity of its availability is certainly unpredicted. As of today, anyone with a decent internet connection can generate images via a simple text prompt. Goodby photography. Hello, promptography.

But more than a gimmick to impress family on social media, image generative AI has real implications for the business world. And consequences. Every company today has to use visual content to sell. From product shots to marketing, for social media and blogs, newsletters, reports, and presentations, it is impossible to do without. Up until today, this meant two things: hire a photographer and/or purchase stock images. Photographers (or photo studios) to take all the product shots and maybe some lifestyle/marketing images. And stock houses to fulfill any and all other needs. That meant companies like Shutterstock could rely on an ever-increasing demand for their images.

A photo studio with products on a table and professional lighting around it. Prompt by PM.

For product shots, the trend has already been slowly shifting. Computer-generated art, rendering, and photoshop has already helped numerous companies make their product look as appealing as they can. And while there might initially be a real photo to start, in the end, the product is more a recreation than a photograph. It won’t be long before it doesn’t matter. Already, AI can build a full-perfect 3d rendering of any object ( and faces) via a handful of still images. In fashion, clothes can be put on virtual models, eliminating the need for expensive setups of studios, real models, make-up, hair, and stylist.

Pretty soon, photos will precede the product. Companies will ask generative AI to design a new product like a car or a new dress, and production will follow. Instead of the opposite today.

The same evolution is about to hit stock houses. Instead of searching for images in a database of pre-existing images, companies will “build” them via a text/voice prompt. Generative AI trained with the company’s product shots, as well as company branding, colors, and mood, will create on-the-fly, on-demand images. Furthermore, existing images could be enhanced with generative content. For example, a boring corporate studio portrait of an executive could see the addition of a perfectly fitted set of sunglasses while the background is changed from a factory to the Swiss Alps. From Detroit to Davos in one click. These functionalities, available today, will be built as a feature inside a company’s already well-established digital asset management ( DAM).

Further reading:

Petapixel: You Can Now Buy and Sell AI Image Prompts on a Marketplace

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

[we just recruited Stitch Fix Senior Data Science Manager Sophie Searcy to speak in our "Empowering Shopping" panel at Visual 1st]

Techcrunch: Deepfakes for all: Uncensored AI art model prompts ethics questions

Petapixel: AI Image Generators Could Be the Next Frontier of Photo Copyright Theft

Techcrunch: TikTok rolls out a basic in-app, text-to-image AI generator

Techcrunch: Mobile photo editing app creator Lightricks launches text-to-image generator

The New York Times: An A.I.-Generated Picture Won an Art Prize. Artists Aren’t Happy.

Oct. 4-5, 2022

Golden Gate Club,

San Francisco

Buy your ticket NOW so you can join our pre-conference LinkedIn Events networking page!

Check out the latest speaker lineup

Announcing Visual 1st Fireside Chat:

Generative AI:

From Text to Image – and beyond

Javier Ideami

CEO Ideami

Co-Founder The Geniverse

As we all know that image generative AI is still in its infancy, the big question is:

What’s next?

We’re thrilled to have Javier Ideami, CEO of Ideami and co-founder of The Geniverse share his perspectives in a 1-on-1 fireside chat session at our Visual 1st conference.

Javier is a multidisciplinary engineer, researcher, creative director, and artist, and award-winning creator of internationally recognized AI projects.

We’ll discuss what the biggest opportunities are for generative AI.

In addition, we'll discuss the core questions that have kept popping up in the last few weeks:

Human creativity is a key source of economic value in all areas of the graphic arts industries. What does Generative AI do to that paradigm?

Is it just another tool (like Photoshop for instance)?

Or does it mean that a machine takes over what has been a human prerogative since the age of cavemen?

Or does it enhance human creativity in ways we could have never imagined?

And what is the impact on how the economic value of creativity is measured and financially rewarded?

Who is the image generative AI professional of the future? A “synthographer” or "promptographer" – is that a technician or an artist?

Promptified by pm:

A ceo of a big company standing in front of

a mountain landscape in Switzerland wearing sunglasses.

Companies will soon need an in-house creative person/ team to generate new content. Someone whose role will be to translate requests into prompts and keep the AI up to date. Someone with both an artistic and machine learning background. This new DAM management role would combine the skills of a librarian, software engineer, and creative director.

From blank canvas creation to the repurposing of existing content, image generative AI is entering the everyday life of companies today. Not via broad, hype-seeking applications like Dalle 2 or Imagen, whose appeal is inversely proportional to its usefulness, but through deep integration of solution-specific deep tech applications. The real generative ai revolution will hardly be visible. It will happen at a granular, everyday level in every company worldwide. Either via generation of new images or repurposing of older ones thanks to integrated 3rd party technology providers. It has already started.

All images in this article were generated via text to prompt using openAI image generator AI Dalle-2

Paul Melcher is the founder of Kaptur and Managing Director of Melcher System, a consultancy for visual technology firms. He is an entrepreneur, advisor, and consultant with a rich background in visual tech, content licensing, business strategy, and technology with more than 20 years experience in developing world-renowned photo-based companies with already two successful exits.

Platinum Sponsors to date

Gold Sponsors to date

Silver Sponsors to date

Archive & Subscribe | Share your news with us | Connect on LinkedIn

Visual 1st | Suite 48 Analytics