Adobe Offers Payment for Original Video Content to Train AI
Adobe, the software giant, is venturing into artificial intelligence with a text-to-video generator akin to OpenAI’s recent innovation. Documents obtained by Bloomberg reveal Adobe’s initiative to compensate artists and photographers for capturing footage of individuals performing everyday tasks and displaying basic emotions, such as happiness or anger.
The company seeks videos depicting interactions with common objects like smartphones or fitness equipment but strictly prohibits copyrighted material and content featuring nudity or offensive elements.
According to Bloomberg, Adobe is willing to pay up to $120 for original footage, translating to an average of approximately $2.62 per minute, with potential rates reaching as high as $7.25 per minute. This compensation model parallels Adobe’s previous payments to content creators for static images used in AI training, which ranged from six to 16 cents per image.
OpenAI’s introduction of the Sora tool in mid-February has intensified competition in the AI realm. Sora can generate realistic 1080p video clips up to 60 seconds long solely from text prompts, and it can mimic various styles including animation or black and white.
Despite initial impressive demonstrations by Sora, scrutiny has revealed imperfections, prompting concerns about potential disruption to Adobe’s dominant position in creative software. Indeed, Adobe’s stock has experienced a 16% decline since the year’s outset, reflecting investor apprehension over this development.
In response to inquiries, Adobe directed Bloomberg to a previous statement outlining its perspective on video generation features.
