AI Firm Offers to Pay Photographers for RAWs
TransPerfect’s DataForce Community has launched the Ronia Raw Photo Collection, an initiative that offers to compensate photographers for their unedited RAW images. The images will be used to train computer vision models. The move signals a potential industry shift toward transparent and ethically sourced training data with direct payment and attribution for creators.
- The Ronia project offers photographers a direct payment of $1.50 per accepted RAW photo, a model that contrasts with traditional microstock platforms like Shutterstock or Adobe Stock, where creator earnings often amount to small, unpredictable royalties per download. - This initiative emerges against a backdrop of significant legal challenges over AI training data, with companies like OpenAI and Stability AI facing lawsuits from creators and organizations like Getty Images for allegedly scraping millions of images without consent or compensation. - Other major players are also shifting toward direct compensation models; Adobe Stock offers creators an annual bonus for content used to train its Firefly AI, and Shutterstock has a contributor fund that pays artists a percentage of revenue from its DALL-E 2-based image generator. - The move to acquire images with explicit consent addresses growing resistance from the creative community; a 2026 survey by the Association of Photographers found that 58% of members have lost work to AI and nearly 90% are not interested in licensing their work for training purposes. - The demand for unedited RAW files is driven by technical needs, as this format contains more unprocessed sensor data, providing higher quality and more robust information for training computer vision models on tasks like image processing and feature detection. - Building effective commercial computer vision models requires massive datasets, often numbering in the hundreds of thousands or millions of images, to ensure the AI can perform reliably across a wide variety of real-world conditions and edge cases. - The term "ethically sourced data" is becoming a key market differentiator, though it lacks a single standard; it generally implies that data is gathered with explicit consent, respects privacy, is legally compliant, and aims to be inclusive to avoid algorithmic bias.
