Skip to content
Feature

Auto reframe: smart 9:16 from gaming streams

Not a dumb center crop. Clippper detects your webcam, gameplay, and chat, then builds dynamic vertical layouts with face tracking and smooth transitions.

How auto reframe works

Five AI systems work together to turn your 16:9 stream into a vertical clip.

Vision AI region detection

GPT-5.4 vision analyzes screenshots from each scene segment to identify stream type, face cam positions, chat overlays, and the main gameplay area.

SAM 3.1 segmentation

Segment Anything Model 3.1 via fal.ai refines bounding boxes for precise webcam and chat regions. No rough guesses — pixel-accurate masks.

10+ layout templates

Screen-only, cam above gameplay, cam with chat, dual-cam, full-screen cam, and more. Each layout defines slot positions and sizes for 9:16 output.

Editorial layout planning

An LLM generates a layout plan for each clip: base layouts per scene, timed events that override layouts, transitions between shots, and per-region effects.

Face tracking crop-pan

For webcam slots, the face track data drives a smooth crop-pan that follows your head movement. The viewport shifts naturally as you move.

Smooth transitions

When the layout switches between scenes, spring-based interpolation blends slot geometries. Chat panels slide in and out. Changes feel intentional, not jarring.

Available layouts

The AI picks from these templates based on what it detects in your stream.

screen_100

Full gameplay

Gameplay fills the entire 9:16 frame

cam_30_screen_70

Cam 30% + gameplay 70%

Small webcam above, gameplay below

cam_50_screen_70

Cam 50% + gameplay 70%

Larger webcam overlapping gameplay

cam_100

Full cam

Full-screen face cam for reaction shots

cam_30_chat_20_screen_70

Cam + chat + gameplay

Webcam and chat above gameplay

Register and get 10 credits per week

No credit card required. Start clipping in minutes.

Get started free

Auto reframe FAQ

How Clippper reframes your gaming stream to vertical.

What does auto reframe mean?
Auto reframe takes a 16:9 gaming stream and converts it to 9:16 vertical video. Instead of a simple center crop, Clippper detects regions (webcam, gameplay, chat) and arranges them in a vertical layout that shows all the important content.
How does Clippper detect regions in my stream?
Three-stage detection: 1) OpenCV detects scene changes via histogram analysis. 2) GPT-5.4 vision identifies stream type and region positions from screenshots. 3) SAM 3.1 via fal.ai refines bounding boxes with pixel-accurate segmentation masks.
What layouts are available?
Over 10 templates including full gameplay, cam above gameplay (30% or 50% cam), full-screen cam, cam with chat overlay, dual-cam, and more. The AI picks the best layout based on what’s detected in each scene.
Does the layout change during a clip?
Yes. The editorial LLM plans layout changes within each clip based on what’s happening. It can switch from gameplay-focused to full-cam for a reaction moment, then back. Transitions use spring-based interpolation for smooth blending.
Does face tracking work for all webcam positions?
Face tracking activates on webcam slots. The system detects your face position per-frame and adjusts the crop to keep you centered. It works regardless of where your webcam is positioned in the original stream.
What about streams without a webcam?
Clippper handles webcam-less streams too. It selects gameplay-only layouts (like screen_100) and focuses on finding the most important part of the game screen to center in the vertical crop.