Effects SDK

The SDK blurs, replaces, or fully removes the background in a live webcam feed while developers swap in a custom image or even an animated video at runtime and set the blur strength on the fly. It also returns a transparent frame with a segmentation mask, so teams can apply their own post-processing on top. According to the company, the model is tuned for real-world rooms, meaning poor lighting and cluttered walls, not studio setups. Real messy rooms. That's the point.

Auto-framing keeps the speaker in view by combining face tracking with motion detection and a digital zoom. It follows the person at a predefined zoom level, so the subject stays centered even if they shift in their seat. No pan-tilt camera needed. For meeting apps and recording tools, that means less fiddling and a steadier shot without a physical pan-tilt camera.

This effect softens skin in real time, smoothing texture and reducing the look of acne and eye bags while keeping the result natural, which matters because an over-smoothed face reads as fake and can make a call feel less trustworthy. In short, it's face beautification that doesn't look plastic. It adds a gentle softness to the lighting, which is a common request from creators and anyone who's self-conscious on camera. The processing is fast enough to run on live video rather than as a post-step.

AI denoise strips digital noise from grainy webcam feeds, the kind you get from a budget camera in a dim room. The result is a cleaner, sharper image without the blurring that a simple low-pass filter would cause. Big difference. It pairs well with background effects, since noisy footage makes clean segmentation harder.

Color grading adjusts colors, exposure, contrast, saturation, and temperature in one pass, pushing a plain webcam feed toward a more cinematic look. The point isn't creative styling so much as fixing whatever the camera and lighting got wrong. Simple as that. For meetings and streams, it makes the difference between washed-out and professional.

Effects SDK ships on Windows, macOS, iOS, Android, Linux, and the web, including WebRTC flows. It supports DirectX, OpenGL, and Metal, and it can use OpenVINO, WinML, and CoreML to run on the GPU, which is what keeps the effects smooth on lower-powered devices like phones and budget laptops that would choke on a CPU-only pipeline. That breadth matters if you build once and ship to desktop, mobile, and browser without rewriting the video layer.