TikTok Adds AI Image Generator

ByteDance is a consumer-apps AI leader for a reason

THUR AUGUST 18TH, 2022 6:50 PM MONTREAL, CANADA

Hey Guys,

This is Data Science Learning Center Premium,

I came across some news this week that I found relatively interesting. I cover AI trends but also have covered the future of social media for quite some time. It’s not entirely common that the two intersect.

So now it appears, TikTok now offers a very basic text-to-image AI generator directly in the app. With all the hype about OpenAI’s DALL-E 2 tool, it looks like text-to-image gen is going to go a bit more mainstream.

I was pretty surprised to see TikTok do this!

What is an AI Greenscreen?

TikTok has rolled a new in-app text-to-image AI generator that lets users type in a prompt and receive an image that can be used as the background in their videos. The effect is called “AI greenscreen” and can be accessed via the short-form video app’s camera screen.

The new effect, as far as I know, was first spotted by The Verge.

The Verge notes: The output of TikTok’s system is pretty basic compared to that of state-of-the-art text-to-image models like Google’s Imagen, OpenAI’s DALL-E 2, or Midjourney’s eponymous software.

Abstract “Swirly” Images - Augmenting the Dreamy Nature of TikTok

So in practice this sort of makes sense for TikTok, as it’s potentially far less controversial.

It creates only rather abstract and swirling images; a strength reflected in the dreamy nature of TikTok’s suggested prompts like “astronaut in the ocean” and “flower galaxy.” Other models, by comparison, can produce both photorealistic imagery and complex and coherent illustrations that look like they were drawn or painted by humans.

The simplified nature of the feature is definately intentional.

That ByteDance has so quickly created such a tool inside its super-app TikTok is actually pretty impressive considering how new DALL-E 2 and other text-to-image pilots are.

While we’ve covered OpenAI’s industry leading tool, there’s actually tons of them. You might even call this the beginning of the no-code AI-as-a-Service internet where things get a bit synthetic, augmented and deep-fake friendly. I don’t always necessarily know if that’s a good thing!

Image credit: Verge, example of AI Greenscreen.

The Synthetic Internet

All to say that, the launch of the new filter comes as text-to-image AI generators are becoming increasingly popular, especially with the launch of OpenAI’s DALL-E 2. You might even call TikTok a post-modern human-AI app, where not always the most “human” content prevails.

I find the product reminds me of Spotify. And it’s no accident, since in the same vein, the video platform (created by ByteDance) recently announced that it will to launch its music streaming app, challenging Spotify, Apple, and other music streaming services. ByteDance is the heir apparent to Meta and a leading AI consumer-apps company.

They are in my opinion the best app maker and app marketer on the planet. TikTok has built an incredibly immersive product, at least if you are under 25. TikTok has more than a billion users, and giving all these individuals the power to create photorealistic images of anything they can imagine would almost certainly produce some troubling results. Douyin itself has nearly that many users as well, so if you think about it, ByteDance’s reach is actually far larger than it’s commonly given credit for.

Douyin, the Chinese version of TikTok, has over 700 million daily active users and is one of the most popular apps in the country.

So the real reach of ByteDance is far far larger than even the usual bragging reveals.

What Kind of Artist is TikTok?

If DALL-E 2 is gothic and dystopian in style, TikTok’s image generation is very postmodern and distinctly impressionist.

The design is therefore fairly attractive and more poppy and spring-like than DALL-E’s more eerie caricatures and storylines.

Have you tried the tool yet?

No Sex on the Beach on TikTok

As testers tried to make TikTok test the boundaries of convention, the Verge said:

Likewise, a request involving nudity — “naked model on beach” — produces thematically appropriate colors, including flesh-tones, sandy oranges, and ocean blues, but nothing that would make a vicar blush.

Oh well, nice try I guess!

TikTok has a number of suggested prompts that appear when you select the effect, including “Snorkeling in a purple ocean,” “Hidden village in the mountains” and “Erupting rainbow volcano.”

It feels like creating design via prompts is already mainstream! Think about it through, The latest cycle of development for text-to-image AI arguably began in 2021 with the original release of DALL-E by OpenAI. Less than two years later and the tech is already in the hands of millions via an app like TikTok.

TikTok is clearly very innovative. Whatever your personal feelings about the app.

Although a bunch of “American” themes and prompts still feels fairly disturbing in visuals:

The new filter puts a text-to-image AI generator in the hands of millions of the most creative young people instantly. That’s a lot more than the slow roll-out to DALL-E 2, which I’m on a list for “early access”.

It really makes you wonder.

Thanks for reading!

If you value these kinds of articles give me a tip, patronage or show community support. I have to pair my writing frequency according to my earnings.

Don’t forget the Substack App is out on iOS if you want to follow me there with notifications, it might be easier. Cheers!

We heard that the Android Substack app is very close to launching.

Have a good weekend!