Fast Image and Video Processing on GPU
Updated:
Fastvideo develops GPU-accelerated image and video processing software for industrial and machine-vision cameras, medical imaging, broadcasting and aerial imaging. Built on NVIDIA CUDA, its SDK runs the full imaging pipeline — RAW acquisition, demosaicing, denoising, HDR tone mapping, and JPEG/JPEG2000 encoding — entirely on the GPU, including NVIDIA Jetson. This delivers gigapixels per second across a wide range of applications, even at ultra-high resolution (1500+ FPS JPEG encoding at 24 MPix).
Fastvideo at a glance
- Speed: 1500+ FPS JPEG encoding at 24 MPix on the GPU
- Acceleration: 10–30× faster than CPU-based solutions, at the same or better quality
- Coverage: full ISP pipeline — RAW acquisition to compressed output, from NVIDIA Jetson to server GPUs
- Standards: JPEG, JPEG2000 and DNG support with standard-compliant codecs
- Company: B2B GPU imaging software, developed since 2009
*OS Windows 10, CUDA-12.6, timings include GPU computations only, image processing at just one GPU for 24-bit images. JPEG encoding with compression ratio ~10:1 and subsampling 4:2:0. JPEG2000 lossy encoding with compression ratio ~12:1, subsampling 4:4:4, cb 32×32. Tone mapping with ALTMapper algorithm. Denoising algorithm is bilateral with min window.
What we offer
- High performance ISP software
- Low latency imaging solutions
- Excellent image quality at real-time processing
- Reduced TCO for final applications
- Compatibility with third-party SDKs on CPU and GPU
- Ease of maintenance, updates and upgrades
- Availability for full range of NVIDIA GPUs (mobile, laptop, desktop, server)
Cutting-Edge technologies for imaging applications
By applying the proven NVIDIA CUDA parallel processing technology, we have managed to achieve extremely high performance of our algorithms on GPU. Our crown jewels are, to the best of our knowledge, the fastest JPEG codec, JPEG2000 codec, resizer, demosaicing and denoising software, and Image & Video Processing SDK. Thanks to ingenious implementation and optimization, our software offers very high performance, and execution on the GPU makes CPU available for other tasks while providing a substantial boost in processing speed.
Our approach is based on our values and promises. We focus on high performance image processing technology which is based on NVIDIA CUDA. What do you want to do? Why is it needed? We want to know how your project is really going. We believe in two-way communication process and that's why we've created our workflow to communicate with our customers worldwide. This is detailed info about the way how we are working with our customers.
- We get inquiry from a customer.
- We send to the customer our DEMO SDK, manual and questionnaire. We ask to send back filled questionnaire to get more info about the company and about the task to be solved.
- We get signed NDA and full specification from the customer.
- Based on the full specification, we prepare a quotation and send it to the customer together with an SLA for evaluation.
- Customer is developing his own application with Fastvideo SDK.
- We help our customers to build their applications by sharing our knowledge.
- We send final SLA and the Invoice to be signed and paid. After that we deliver SDK according to specification.
- Maintenance and support, new releases.
We offer our products and services through our in-house sales team or via our partners. We are working with OEMs, System Builders, Integrators, Distributors, general Resellers or Agents. If you are interested in our products, please send us your request.
Wide Range of Applications
Fast compression and real-time image processing on GPU is a popular and growing demand in modern industries. Indeed, high framerate cameras, real-time video surveillance systems, heavy-loaded web services and medical equipment often require zero tolerance to delays due to image processing. And this is where the image compression or image processing speed can become a serious bottleneck.
Our video and image processing technologies do provide the required performance! Comparing with classic CPU solutions, our GPU algorithms demonstrate 10-30 times performance speed up retaining the same or even better quality of the output and minimizing the host processor workload.
Made by Experts and for Experts
We have vast experience in image processing as well as in mathematics and system programming. And we love what we do!
Our qualification and enthusiasm have allowed us to develop the first high speed JPEG codec utilizing GPU computation power; it is also one of the fastest on the market. And GPU Debayer software implementing NVIDIA CUDA for image demosaicing delivers a significant performance gain compared with traditional CPU-based approaches.
You can check our latest project - Fast CinemaDNG Processor software. This is high performance end-user application for post production with all features from our GPU Image & Video Processing SDK.
Based on the technologies that we have developed, we provide our high performance Image & Video Processing SDK and custom software design for CUDA-powered image and video processing. This is B2B business and we offer our CUDA software for integration with third-party solutions.
Depending on your needs, we can just provide the technology “as is” or develop a turnkey solution for your very specific needs. Further technical support and consulting are also welcome. In addition we can help integrating our solutions to existing infrastructure. Please don’t hesitate to contact us, we will be happy to assist.
Frequently asked questions
What does Fastvideo do?
Fastvideo develops GPU-accelerated image and video processing software for industrial and machine-vision cameras, medical imaging, broadcasting and aerial imaging. Built on NVIDIA CUDA, its SDK runs the full imaging pipeline — RAW acquisition, demosaicing, denoising, HDR tone mapping, and JPEG/JPEG2000 encoding/decoding — on the GPU, including NVIDIA Jetson.
What is the Fastvideo SDK?
The Fastvideo SDK is a B2B GPU image and video processing library for NVIDIA CUDA. It provides demosaicing, denoising, color grading, HDR tone mapping and JPEG/JPEG2000 codecs as building blocks you integrate into your own application, with the whole pipeline executed on the GPU. See the Fastvideo GPU Image & Video Processing SDK page for details.
Which GPUs does Fastvideo software run on?
Any CUDA-capable NVIDIA GPU — from Jetson embedded modules to laptop, desktop GeForce/RTX and data-center server cards.
How fast is GPU image processing with Fastvideo?
On the GPU our algorithms run about 10–30× faster than classic multithreaded CPU implementations, while keeping the same or better image quality. For example, JPEG encoding reaches 1500+ FPS for 24 MPix images. Exact numbers depend on your GPU, resolution and settings — see the benchmark methodology and test environment.
Which image formats and codecs are supported?
RAW/Bayer, JPEG (ITU-T T.81 / ISO IEC 10918) and JPEG2000 (ITU-T T.800 / ISO IEC 15444), plus demosaicing, denoising, resizing and HDR tone mapping. The JPEG and JPEG2000 codecs are standard-compliant, so their output opens in any compatible viewer.
How can I evaluate Fastvideo software?
Request a demo SDK for Windows, Linux or ARM (Jetson/L4T) and benchmark it on your own hardware. Use the contact form below to get the demo SDK, manual and questionnaire.
Testimonials
“Fastvideo SDK is lightning fast… we can now process a 28,800 image sequence set in under 10 minutes, instead of 10 hours.”
As we use machine vision cameras, we capture RAW frames in DNG format. This data has to be post processed first before it can be input into any 3D reconstruction software. The images needed to have debayering applied, colour corrected, exposure corrected, noise removed, sharpen applied, camera profile added, etc. We first tested a pipeline using Adobe Camera Raw. This was buggy. We used After Effects, this was too slow. We settled on Lightroom for a while, this was slow but reliable. To process 28,000 images took over 10 hours in Lightroom and the UI cumbersome for sequence data. We nearly got around to Nuke but something else came along…
..then we found out about Fastvideo SDK. Fastvideo SDK is lightning fast. It can process sequence data on the fly, on the GPU in real-time in ms, rather than in minutes per frame. This means we can now process a 28,800 image sequence set in under 10 minutes, instead of 10 hours. Literally a game changer. We would also like to link to www.fastcinemadng.com which is where you can download and test the Fast CinemaDNG application.
We are developing SW / HW technologies to accelerate machine vision and embedded imaging systems. When customers ask us to build full image processing pipeline for the system with NVIDIA GPU, with Fastvideo SDK we are able to get very high performance and image quality for the whole system. Color and raw image processing on NVIDIA CUDA is the key feature for our embedded multi-camera solutions.
We are really impressed with image quality and processing performance of Fastvideo SDK whether it works on Jetson or professional GPU, both for real-time and offline mode.
We also find very useful JPEG and JPEG2000 codecs on NVIDIA GPU. The performance is much faster in comparison with CPU-based implementations of JPEG and J2K algorithms for image encoding and decoding.
XIMEA is manufacturing the smallest industrial cameras with the latest image sensors which provide exceptionally high data rates. To ensure high speed, these cameras are equipped with fast interfaces like USB3 or even PCI Express.
Realtime image processing for such low latency cameras is a difficult problem which was successfully solved with GPU-based Fastvideo SDK. We especially appreciate the quality of their debayering, denoising, color grading, tonemapping, and performance of JPEG and JPEG2000 codecs.
FastVCR software with GPU-based image processing and image sensor control offers real time, reliable and high quality solution for XIMEA cameras.
About the author
Fyodor Serzhenko, PhD, is the founder and CEO of Fastvideo. He earned his PhD at the Moscow Institute of Physics and Technology (MIPT) in 1993. Since 2009 he has led the development of Fastvideo’s GPU-accelerated image codecs and ISP modules, including the Fastvideo SDK. Connect on LinkedIn.
Why you can trust these results
Fastvideo has built GPU-accelerated image processing software since 2009. Our codecs comply with the standards they implement — JPEG (ITU-T T.81 / ISO IEC 10918) and JPEG2000 (ITU-T T.800 / ISO IEC 15444) — and our tools are open source on GitHub. Every performance figure we publish is reproducible: see our benchmark methodology, download the Fastvideo SDK benchmark report (PDF), and measure it on your own GPU.
