Adobe Inc. is investing $3 per minute in video content to fortify its artificial intelligence (AI) text-to-video generator. This move is aimed at narrowing the competition by OpenAI’s Sora.
Adobe has asked its extensive network of photographers and artists to submit videos capturing commonplace activities and human emotions such as walking, joy, and anger. The objective? To amass a trove of assets for AI training purposes, according to the company.
In the past year, Adobe has diligently augmented its suite of software for creative professionals, integrating generative AI functionalities into stalwarts like Photoshop and Illustrator. Its endeavors have birthed tools that translate text into images and illustrations, garnering billions of engagements, Bloomberg first reported.
However, the unveiling of OpenAI’s video-generation marvel, Sora, has reignited concerns among investors, casting shadows of doubt on Adobe’s resilience in the face of disruptive innovations. In response, Adobe has affirmed its commitment to video-generation technology, promising further revelations later this year.
Recommended reading: Canva acquires Affinity to challenge Adobe
What you should know
The requisition entails the submission of over 100 brief clips portraying human actions, emotions, and rudimentary anatomical features, alongside interactions with everyday objects like smartphones and fitness gear. Notably, contributors are cautioned against furnishing copyrighted or offensive content.
Compensation for submissions averages at $2.62 per minute of footage, with potential earnings scaling up to $7.25 per minute.
More insight
The solicitation shows the heavy data requirements underpinning AI models fueling popular content creation tools like ChatGPT. Debates and controversies surrounding data sourcing have raged on. Mira Murati, OpenAI’s Chief Technology Officer, alluded to Sora’s training data origins in a viral Wall Street Journal interview, hinting at sources like YouTube, Facebook, and Instagram.
Adobe, on its part, champions a distinct approach, training its models predominantly on an extensive stock media library tailored for marketers and creative agencies.
Recommended reading: Adobe to pay $1 billion after terminating Figma startup acquisition
