PhotoRoom Scales Large AI Models with Google Cloud Supercomputers
With generative AI’s assistance, PhotoRoom aims to provide high-quality images to a broader lineup of businesses. PhotoRoom has partnered with Google Cloud to use the capabilities provided by Google’s A3 supercomputer.*
PhotoRoom has stated that it processes about two billion images every year. The company’s collaboration with Google will allow PhotoRoom to scale large AI models. Moreover, this collaboration has added new generative AI capabilities, which include incorporating a trained diffusion model* into its platform.
*A3 Supercomputer: Engine purposely built to train and serve demanding AI models, combining NVIDIA H100 Tensor Core GPUs and Google’s networking advancements.
*Diffusion Model: Deep generative models that work by adding Gaussian noise to the available training data and reversing the process to recover said data, gradually learning to remove the noise.
Swift Image Creation and Editing
PhotoRoom CEO Matthieu Rouif made a statement that this partnership with Google will enable them to handle an increasing volume of images worldwide without compromising service, quality, or performance. It is claimed that PhotoRoom can now generate AI images in one second, 10 times faster than other generative solutions.
This collaboration is said to be built upon PhotoRoom’s Remove Background, Instant Backgrounds, and Instant Shadows features used to enhance product shots. Their new AI model allows businesses that require building a completely new background to showcase their products in an instant!
Business Application
The AI-powered capabilities brought out by PhotoRoom and Google are aimed at bringing plenty of benefits, especially to small and medium-sized businesses.
PhotoRoom’s technologies can quicken the process of creating product shots and reduce the need to rent a photography studio, helping businesses become more cost-effective. These features may allow professionals and entrepreneurs in the business to create their creative content without time-consuming hassles.
Read more articles like this at Digger Insights.
