Ticker

6/recent/ticker-posts

amazoon

Google used 64 cameras for Portrait Light




Google has published a blog that explains the technology behind Portrait Light, including how to train its robotic learning models.

Google's Portrait Light can make some of your humble images look much better by giving you a way to change the direction and intensity of lighting.

Google launched the AI-based lighting feature in September for Pixel 4a 5G and Pixel 5 phones before giving older Pixel phones access.

The company needed millions of images with additional lighting from different directions and without additional lighting from different directions, so that they could train one of these models to add lighting to an image from a particular direction.

The company used a spherical lighting device with 64 cameras and 331 individually programmable LED light sources to capture the images it needed.

She photographed 70 people in different shades of face colors, facial shapes, races, hairstyles, and even clothing and accessories, highlighting them inside the device through one light at a time.

The company has also trained a model to determine the best illuminated self-image in order to automatically put the lighting, and the published blog contains all technical details.

The Pixel camera automatically features Pixel 4, Pixel 4a, Pixel 4a 5G and Pixel 5 implements Portrait Light in default mode and through people-to-people Night Sight images.

You can find the option to turn the feature on under the settings section when you edit a picture of someone on your older Pixel phones.

You should take the image in Portrait Mode and edit the image through Google Photos if you want to be able to manually reset the lighting position and adjust its brightness.

Google produces some of the best still images on Pixel phones, a result of hard work and lots of machine learning models.

Google is not the first smartphone manufacturer to have introduced Portrait Lighting effects to its customers, but it may have done so using a large number of advanced algorithms and machine learning models.


Post a Comment

0 Comments