It's all about the angle: Your photos, re-composed
It's all about the angle: Your photos, re-composed
Google Research introduced a new method in Google Photos’ Auto frame to change camera viewpoints after a photo is taken. An internal 3D point-map estimation model reconstructs the scene and focal length, then a generative latent diffusion model fills gaps exposed by the new view. It automatically detects faces and wide-angle distortion to suggest ideal framing. The feature already applies automatically to photos containing people, with reframed versions available among Auto frame candidates.
Selection record
AdmittedSum of both 124 ≥ twice the threshold 120
- Source tier
- Official, first-hand; this tier's threshold is 60
- Pre-filter
- passed：Google Photos用生成式AI重构图，实质AI应用
- Why it was chosen
- Combining 3D scene estimation with generative completion explains how the new post-capture viewpoint editing approach works.
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: Google Research · research.google