Best Take

Why It Exists

Group photos often have at least one person blinking, looking away, or making an imperfect expression. Users need an effortless way to create perfect group photos without retakes.

How It Works

Best Take leverages Google's Gemini Nano vision AI to analyze multiple frames captured in burst mode and identify the optimal facial expression for each person in the group. The system detects facial features, micro-expressions, and eye openness across frames, then seamlessly composites the best elements into a single natural-looking photo. All processing happens entirely on-device on supported Pixel phones, ensuring privacy and speed. The feature was launched as part of the Pixel 8 Pro photography suite and uses the Tensor G3 NPU for efficient image processing.

Everyday Use Cases

• Group portraits
• Family photos
• Event photography
• Selfies with friends
• Capturing natural expressions
• Photo retakes eliminated

User Workflow

• User takes a photo of a group in burst mode
• Gemini Nano AI analyzes each frame for facial expressions
• The best expression for each person is identified
• Frames are seamlessly blended into a single image
• User reviews and saves the final photo

AI Processing Flow

Multiple burst frames are captured by the camera. On-device Gemini Nano vision models detect faces and analyze expressions, eye openness, and mouth shapes across frames. The system scores each frame for each person. A neural blending algorithm composites the best facial regions from different frames into a seamless final image. The entire process runs on the Tensor G3 NPU with no cloud dependency.

Inputs / Outputs

Inputs:

• Burst photo frames
• Face detection data
• Expression analysis
• Eye state detection
• Lighting conditions
• Camera stabilization data

Outputs:

• Single composite photo with optimal expressions
• Natural-looking facial blending
• Preserved background consistency
• Enhanced image quality

Known Limitations

• Requires burst mode capture
• Works best with 2 to 10 people
• May not work well with extreme lighting
• Best Take is unavailable on non-Pixel devices
• Some facial features may show minor blending artifacts

Unsupported Scenarios

• Videos or single frame photos
• Large crowds over 15 people
• Low-light conditions beyond Night Sight
• Non-human subjects
• Professional portrait photography

Performance Notes

Frame analysis completes in under 2 seconds on Tensor G3; Blending and compositing in under 1 second; Entire process uses under 5 seconds total; NPU acceleration reduces CPU load by 70 percent

Available On

Shows where this feature is available and how its AI processing works on each platform. Availability may vary by device.

Research Status

Confidence and verification reflect how complete documentation is. Fields may show Not assessed or Not yet verified while research is ongoing - this flags gaps, not product deficiencies.

Research Notes

Available on Pixel 8 Pro from launch in October 2023; Requires Google Tensor G3 or later; Uses Gemini Nano 1 vision model; Introduced at Google I/O 2023; Pixel 10a also supports Best Take with improved accuracy

Details

Capability Mapping

This feature maps to canonical capability family: Writing.

Canonical capability: Text Rewrite.

• Text Rewrite - Restates or rewrites existing text in a new style or tone, including AI-assisted response suggestions.

Grounded in 1 device(s) on Gemini Nano.

Related Features

• Photo Assist - Cross-platform equivalent