DeepField-SR1/SR2
It is an AI-powered Super Resolution technology that upscales videos to high resolution and improves video quality by using temporal and spatial information.
DeepField
It is an AI-powered Super Resolution technology that upscales videos to high resolution and improves video quality by using temporal and spatial information.
Eliminate encoding
noise and artifacts
Video SR consists of video processing algorithms such as noise removal, image quality enhancement, and upscaling.
For each algorithm, Neural Network model is trained on a large number of video For each algorithm, Neural Network model is trained on a large number of video
It makes advantage of the temporal information from multiple video frames to provide a more realistic and better visual quality.
Product features
Remove encoding noise and artifact
The trained Neural Network for the noise and artifacts removing algorithm removes encoding artifacts (blocking) that occur during video lossy compression and preserves details.
Enhancing video quality and upscaling video resolution
The trained AI model enhances clarity and adds delicate textures to produce high quality videos.
Use the temporal information from video
The motion blur that typically arises in single-frame-based upscaling has been significantly minimized by combining temporal information from many frames.
Choosing an artificial intelligence Model
We provide two separate AI models that are tailored to serve various video service needs.
Performance focused modelDeepField-SR1
Applied for a live streaming with the highest level of performance
Quality focused model DeepField-SR2
Applied for better enhancement quality and upscaling resolution
How we deliver the product
SDK
Provided as an SDK optimized for GPU/FPGA accelerator cards and supports the FFmpeg framework, making it simple to integrate into your media workflow.
Silicon IP
DeepField products are also available as Semiconductor IP that can be integrated into media processing FPGAs or ASICs for servers for optimal performance.
UI Service
Offers file-based video upscaling cloud service for content providers
Supported Devices / Supported Platforms
A light-weighted AI model implemented into software and semiconductor IP that is designed for supported devices
On premises
Nvidia GPU Card, Xilinx ALVEO FPGA Card
Cloud
AWS EC2 instances, MS Azure NP instances
Inquiry For more information,
please contact us.