GPU Software for high speed image processing
We are developing high quality innovative software for fast image and video processing on GPU. Taking the most from NVIDIA’s CUDA technology, we have managed to create high performance debayer, denoiser, resizer, super fast JPEG and JPEG2000 codecs, and much more. In addition, multiple tools for realtime image processing on GPU are on the pre-release stage.
The main idea behind our products is very impressive: we can win twice by offering significant improvement both in image quality and processing time. Usually people slow down the software by implementing more complicated image processing algorithms, so this is a trade-off between image quality and performance. If you utilize GPU instead of CPU for image processing applications, you will be able to be much faster than CPU and to get better image quality at the same time.
Image Processing SDK for NVIDIA GPUs for Windows/Linux/ARM
We've built a core engine for high performance ISP on NVIDIA GPU. Image Processing SDK is our main product which is available for Windows/Linux/ARM and is compatilble with all types of NVIDIA GPUs: mobile Jetson, laptop/desktop GeForce, professional Quadro and Tesla server hardware. It was created with an idea of ISP for camera applications in mind, so you can find a set of highly optimized software modules which are offering outstanding performance and quality.
FastVCR software for machine vision cameras with GPU-based ISP
We've developed FastVCR software for XIMEA USB3 and PCIe machine vision cameras to offer real time processing with high image quality. We are ready to port that application for your particular camera to build a customized solution according to agreed specification and task to be solved. Integration of the camera application with various AI applications is available as well. We can process frames from various machine vision cameras with USB3, MIPI CSI2, CameraLink, Coax, GigE, PCIe interfaces on Windows/Linux/ARM. Different versions of that software are compatible with the products of the following camera manufacturers: Emergent Vision Technologies, IO Industries, Basler, XIMEA, Lucid Vision Labs, FLIR/DALSA, Imperx, Daheng Imaging, Mindvision, Hikrobot, etc.
High performance GPU-based codecs for image encoding and decoding
We've implemented our own super fast codecs for image encoding and decoding on CUDA: JPEG, JPEG2000, JPEG XS, Raw JPEG, etc. These are highly optimized software solutions on CUDA which are much faster in comparison with corresponding CPU-based codecs. We are ready to integrate these codecs into your software.
MXF Player and DCP Player
FFmpeg solutions on GPU
FFmpeg software already has both J2K codec and Remap filter, but they were implemented on the CPU, so they are very slow. To offer high performance processing for realtime applications we've built both FFmpeg J2K codec and FFmpeg Remap filter on GPU.
GPU RAW Processor
This is a high speed solution for a task of RAW to RGB conversion for machine vision cameras, digital cinema, aerial and volumetric imaging applications. The main idea is to do all processing on GPU and with that approach we could be 30-50 times faster in comparison with Adobe Lightroom / Camera Raw or Raw Therapee software.