Almost human (Lisha Li, Rosebud.ai)

Rosebud AI founder on virtual models and the future of synthetic media

Reading time : 4 minutes

Rosebud AI is a San Francisco-based synthetic media company that creates CGI models that are indistinguishable from real humans. The company provides a vast library of photos and videos of a diverse range of virtual models, called Generative Photos.

In the videos, the face and the words it speaks are synthesised by software. The imagery and videos can be used for brand advertising, storytelling and entertainment and are fully customisable, so that you can choose the skin colour or ethnicity of the virtual model to appear in your content or create your own brand ambassador with just a few clicks. The company also offers a tool that puts clothes photographed on mannequins onto virtual models and is working on a new service, called Tokking Heads, that animates virtual or real humans with just text or speech.

With the pandemic dramatically disrupting production processes for the marketing, media and entertainment industries, Rosebud AI’s offering has never been more in demand. Production methods using CGI videos and imagery, sometimes termed ‘deep fake’, are a cost-effect, efficient alternative to the real thing, and with traditional methods presenting so much more risk dueto Covid-19 restrictions, more corporations and brands are turning to CGI for content creation.

We spoke to Rosebud AI’s founder and CEO Lisha Li to learn more about the brave new world of synthetic media.

Explain how Rosebud AI works?

We use virtual production methods and newer techniques from deep learning to accelerate visual content production through photo and video, and specialise in faces.

Through this period, we have had a high demand for use of our technology in marketing, educational and instructional videos as well as for use for customer service, and other creative use cases.

Do you use real or famous faces?

We started off creating virtual faces from scratch, not based on anyone. The value proposition in print was clear; it lay in having something unique. But, when we move to video, some clients want to use an identity that’s established or somebody famous. Or it could be anyone who wants to use their own face or to edit their speech in a text-based way, so that’s something we’re developing right now.

Is regulation in place to combat fake news and to ensure the tech is used in an ethical way?

Obviously these are the news headlines everyone will be familiar with. And frankly, I think regulation probably has a little way to go, it needs to catch up a little faster. It’s also hard to consider how you properly regulate something like this. Do you try to attach it to the source? How do you force people to comply? Those are all worthwhile questions to ask. More broadly, we have a code of ethics. We want to make sure we play a part in educating the public about the technology, so people are informed.

How has the Covid crisis impacted your business?

Covid has really accelerated organic, inbound interest for virtual production. We haven’t done any self-promotion or advertising for brand ourselves, as we’re still focused on finessing the product, but people still find us. We’re just taking the opportunity to figure out what’s the best way to scale our interface. During this time, clients have been happy to use a self-serve model with the app. But it’s not a consumer tool, we want to be careful about how we bring it to market.

Which sectors are driving the most demand for rosebud?

We’ve seen a lot of demand from the e-commerce industry. There’s also demand for faces to service text chat bots in customer service, and for the platform interface to enable the creation of videos. Interest has been fairly international. Small business owners are very interesting – they feel the pain of not having the resources of larger companies, to produce content. Therefore, interest runs counter to media reports of ‘fake news’, or taking jobs away from models. SMEs want to support minorities and diversity to reflect their business and ethos; and don’t have the capacity, especially during Covid, to hire models.

This is the kind of tool that comes into its own; it allows us to bring a business from physical locations to a digital e-commerce setting.

it allows us
to bring a business
from physical locations
to a digital
e-commerce setting.

Our photo editing tool is available online and we’ve just started working with creators with a closed BETA version of our video animation product called Tokking Heads. By creators, I mean people who are making YouTube content, to marketing videos. They give us quick feedback helping us iterate quickly.

We continue to open the beta to more users, including some in broadcast media. Some exciting projects will be released soon.

Lisha Li

Lisha Li

CEO, Rosebud.ai

Lisha Li is the founder and CEO of Rosebud AI. Rosebud creates images and videos programmatically that look indistinguishable from real ones. She was previously a principal at Amplify Partners leading investments in early stage start-ups and worked in data science at Pinterest and Stitch Fix. Lisha holds a PhD in Machine Learning from UC Berkeley. In 2019, she was one of the three winners of L’Oréal USA’s Women in Digital NEXT Generation Awards.
They launched a collection of 25,000 images of AI-generated faces to be used in photo stocks. The objective was to bring more diversity to stock images and propose images that their customers could relate to. The faces they offer are customisable and personalisable. Users can algorithmically place any face onto anybody in their collection.
Leave a comment
Subscribe
Notify of
guest
0 Comments
Inline Feedbacks
View all comments

Latest Posts