Getty sues Stability AI, Diversity in AI image making & NeRFs | 002

Omar's newsletter about Creative AI and Other technologies* | 002

Hey! Welcome to newsletter number 2! Firstly, I just want to say a massive thanks for subscribing. The plan is to share insights, news, personal experiments and learnings. Especially as the AI hype dies a little and how we harness and find value in AI becomes the main focus for the creative industries.

This week!

• Unstable Diffusion - Getty, Copyright and Why Stable Diffusion

• AI Diversity, Inclusion and Body Bias - Industrialising inequality, or increasing inclusion like never before.

• NeRFs and why you need to know about them

• Future Tech

• AI in VR = the metaverse but, better. (tease for next week)

Unstable Diffusion

This week we've got a few really interesting developments on a legal definition of creative produced by AI. Getty, the stock image agency, is suing Stability AI, the company behind open-source text-to-image AI, Stable Diffusion. Apparently, Getty doesn't want compensation but wants Stability AI to pay a licence fee.

"Getty Images CEO Peters compares the current legal landscape in the generative AI scene to the early days of digital music, where companies like Napster offered popular but illegal services before new deals were struck with license holders like music labels."

The thing that got me is that it feels Getty is claiming copyright over the space between images, latent space. so, while the AI can't reproduce an actual image, it imagines this space and creates its own image. If Getty gets to own that space, what would it mean for the future of image-making and fair use?

And, why the open-source and not closed-source AI like Dall-e and Midjourney? There will be ramifications for all whatever the outcome.

Diversity, Inclusion and Body Bias

The conversations that we're finally having about Diversity and inclusion in the creative industries are slowly leading to results. We see org-structure change and talent investments, the industries slowly move in a better direction. And, AI has the capability to either speed up or undo decades of progress.

As more and more of our feeds are awash with AI content, biases begin to present themselves. Beauty, body, race, gender and age biases are reflections of the databases the AI are trained on. And the databases are homogenous, amorphous reflections of ourselves. But, there's a solution...

The AI isn't a fixed entity, you can ask it to show you anything and any person. AI is a tool that allows any community or individual to see themselves in the world around them. I want to imagine a time when a movie with characters that reflect the self-hood of the viewer is on at the local cinema and it's a norm for everyone to be seen. Where ads include people, shaping narratives around people to truly connect with them and all sorts of other magic.

I've been working on an AI that has a more diverse and inclusive dataset, resulting in the AI producing imagery that is more reflective of the world, rather than the people that make them. It's way harder than I thought it would be so progress may be slow on this one!

So, if you're an AI artist or a creative that wants to visualise a thought or an idea, you can choose to be as inclusive as you want - you get to choose the representation. Be mindful of your intentions, do it with respect, and share your platform with others - always add to the vibe!

Below is another experiment in this space, I wanted to see the lesser-seen soccer casuals of the late 1980s in the UK and The Netherlands. Women, and Non-binary hooligans that loved the game and the Italian fashion of Massimo Osti.

NeRFs and why you need to know about them

NeRF, short for "Neural Radiosity Field," is a brand-new technique for processing 3D images and creating 3D models. Although it differs significantly from structured light and depth scanning techniques, it superficially resembles photogrammetry in that you must take a bunch of pictures of a person or object.

From a visual standpoint, NeRFs are a beautiful aesthetic, creating milky artefacts that make even the most mundane scene look phenomenal, especially when combined with 'drone-like' camera moves, all orchestrated on a computer.

As this technology matures over 2023, I expect consumer-level apps that will allow any brand or creative to shoot stunning 3D product shots that can live anywhere from the virtual space of a metaverse to an Etsy or DePop page.

I'm experimenting in this space, primarily exploring use cases for fashion; fit check content, locations and b-roll content. Assuming that people would enjoy the new visual spectacle of relatively unknown technology. Turns out that actually, all they want is a NeRF of a bin...

My highest performing TikTok is of a wheelie bin behind my flat. I wanted to see if I could Michael Bay a bin and then watch with amazement as thousands watch and hundreds commented about the bin being their ex or The Weeknd's new album.

When AI can NeRFs merge; making things for a metaverse experience is going to be wild. Individual spaces for anyone. If you're a creative or a brand, immersing users and customers in your brand will take on a new meaning!

Future Tech

Dreamstudio Pro

Here are some of the tools that are either available now or are just around the corner. I'm most excited about Dreamstudio pro by Stability AI, the feature list includes text-to-video, music and audio generation AI, making music videos, and storyboards, and creating and sharing effects and AI tools. I imagine Adobe either has something on the way or will buy Stability AI. Really going to be wild this!

ChatGPT Pro

The whirlwind AI that everyone is talking about has gone Pro. For $42 a month, you get priority access and access to new models before anyone else. This is amazing and I wonder how they'll keep the buzz going beyond the 'I asked ChatGPT to X' content bubble. Are you signing up?

Thanks!

*Need a new catchy newsletter name but, don't want to ask the AI to name it. Any suggestions are welcome!