Netflix

Netflix's Multi-Ratio Video Resizing for Mobile Feed and Marketing

Original post Multi-Ratio Video Resizing for Mobile Feed and Marketing

Article Summary

Netflix just solved a problem every mobile team faces: how do you turn cinematic 16:9 video into vertical formats without losing the story? Their answer involves neural networks, active speaker detection, and a virtual camera that knows where to look.

With 1.2B social followers and the new Clips vertical feed in their mobile app, Netflix needed to scale video reformatting beyond manual editing. Their engineering team built an automated multi-ratio cropping system that generates 9:16, 4:5, and 1:1 versions from horizontal footage while preserving creative intent.

Key Takeaways

Critical Insight

Netflix automated vertical video creation at scale by treating resizing as a tracking problem, combining face detection, saliency maps, and smooth camera movement to match cinematic quality.

Continue reading

What This Article Covers

Problem

At Netflix, editors used to manually reframe horizontal shots into vertical formats. This took a lot of time and didn't scale well as the catalog grew and marketing demands increased, particularly for scenes with complex multi-character blocking and camera movement.

Solution

The pipeline relies on an internal face detector combined with LipFlap for active speaker detection. It applies motion heuristics that lock the crop in place for small movements, then smooth into dynamic panning when the displacement becomes significant.

Impact

The reframing tool now lives inside Netflix's internal Clips tool, complete with reviewable shot controls. Producers can export the resizing parameters as XML directly into Adobe Premiere, which cuts down on production time and manual effort.

The article reveals how they handle the trickiest scenario: multiple characters talking on opposite sides of the frame.

This page summarizes “Multi-Ratio Video Resizing for Mobile Feed and Marketing” by Netflix, published Oct 26, 2026. All credit for the original work goes to the authors.

Read the original on netflixtechblog.com (opens in new tab)

AI tools for Android

Browse all AI tools →