Advanced AI solutions that boost embedded camera image quality in low-light, high-contrast, and zoom use cases.
Built on deep imaging know-how, our solutions are optimized for seamless deployment in both RAW and YUV processing architectures.
Our award-winning AI Video Low-Light Noise Reduction solution delivers clean, sharp video in challenging lighting conditions for smartphones, surveillance, security, and embedded vision cameras.
Designed for real-time performance, the solution removes noise while preserving detail and motion fidelity — even in dynamic scenes and moving-camera applications.
Advanced global motion handling eliminates ghosting and temporal artifacts, enabling stable, natural-looking video in ultra-low-light environments.
The solution also improves low-light object detection and recognition.
Key Features
• Real-time AI video denoise
• Optimized for global motion and moving cameras
• No ghosting or temporal artifacts
• Ultra-low-light enhancement
• Improved low-light object recognition
• Available pre-ISP and post-ISP for flexible integration
Visidon’s AI-based super resolution technology delivers fast, high-quality video upscaling optimized for embedded NPU hardware — dramatically improving output quality and enabling higher zoom levels compared to basic digital zoom.
For video conferencing and handheld devices, we offer a dedicated Video Face Super Resolution that keeps faces sharp and natural even when filming people from a distance — whether in a meeting room, at a concert, or while vlogging.
We also provide multi-frame super resolution for still imaging, where an AI-enhanced upscaled image is created from multiple low-resolution inputs — ideal for producing clear, detailed zoom images.
Case studyA solution for correcting and optimizing the dynamic range of photos and videos and enriching the color tone of the output. Combines multiple differently exposed shots and automatically handles moving objects to avoid ghosting or artefacts. Brings hidden details in shadows and highlights visible.
Visidon offers two optimized RAW processing suites — RAW Performance and RAW Power — designed to match your hardware architecture and performance targets. Each suite delivers a complete computational photography pipeline for still imaging, including multi-frame noise reduction, HDR, SuperPhoto Zoom, and Night mode.
RAW Performance targets efficiency-focused designs and runs without a dedicated NPU, while RAW Power is built for NPU-equipped platforms demanding maximum image quality. Both suites are ready for integration across mobile, action camera, and drone platforms.
Visidon’s depth and computational photography features bring DSLR-quality imaging to embedded devices — without specialized optics or large apertures.
Our bokeh engine delivers realistic shallow depth-of-field portraits with accurate edge matting around hair, glasses, and person boundaries. It supports both stereo cameras (AI depth matching for a true depth map) and mono cameras (monocular depth estimation with AI matting).
Macro focus enhancement combines multiple focus layers into a single sharp image, preserving fine textures and extending the capability of budget optics.
The portfolio also includes depth estimation and portrait effects as standalone modules for mobile and embedded platforms.
Visidon’s optical correction suite digitally eliminates lens artifacts, delivering flawless image quality as a foundation for further processing such as HDR, Night, and Zoom. Works across all hardware tiers and can be applied as post-processing for flexible deployment.
Lens distortion correction corrects wide-angle geometric distortion while preserving straight lines in architecture and landscapes. A face-aware model maintains natural facial proportions for group photos and selfies.
Video face distortion correction extends this to real-time video — ideal for video conferencing and handheld recording.
Purple edge correction eliminates chromatic aberration fringing without over-filtering, preserving original colors and improving the output of affordable lenses.
Understanding Face DistortionComputes a depth map from single or dual camera inputs. Enables computational bokeh and other stylization effect creation.
Case Study - AI enabled Smart ScreenIdentifies faces in view during video calls. Keeps the primary individual in focus and at the center of the frame with face detection and automated adjustment of zoom level or focuses on several meeting room participants and groups them together.