BitSave
SPECIAL MERIT - Cloud Infrastructures - Delivery
iSize Technologies
The BitSave AI-powered upscaling encoding platform by iSize is a first-of-a-kind, fully AI-powered encoding platform that takes advantage of AI and deep learning to enable up to 70 percent bitrate saving versus non-AI competing codec enhancement solutions.
It is important to note that BitSave is 100 percent codec independent, which means that it is not beholden to any standard or format and can be applied to any application, platform, or workflow that has to move data quickly and efficiently. The proprietary AI technology that drives BitSave substantially increases the efficiency and performance of all the latest codec standards including AVC/H.264, HEVC/H.265, and VP9, thus ensuring seamless integration with existing media workflows.
Speaking of standards, BitSave does this without breaking any standards because it is an add-on solution for an existing codec infrastructure rather than a replacement. Being codec-independent means that BitSave can reduce video delivery system bitrate requirements without waiting for the often-lengthy process of new codec standards to be developed and ratified. BitSave has already been proven to accelerate encoding by up to 500 percent, primarily because of its intelligent approach to pre-coding pixels before they enter an encoder, thus providing the encoder with a smaller, easier, and faster-to-process footprint right from the start.
When people stream video today, they basically choose a streaming recipe that has been pre-cooked according to the platform they want to stream over. That almost always means there is a massive loss of quality, and if you’re in an area with poor Wi-Fi, as many people around the world still are, a 4K stream will quickly get switched to a much lower quality stream.
So iSize accepted that challenge. The feeling was not to just optimise for low-levels metrics like signal-to-noise ratio. They aimed higher. Much higher, into the scientific realm of perceptual metrics.
BitSave is essentially a “pre-coder” that performs perceptual optimisation of the content - as well as finding the best possible resolution for the content according to the desired bitrate stream - before the pixels reach the encoder.
That means that BitSave is a straight pixel-to-pixel engine that ingests and produces pixels that are pre-processed to optimise whatever comes next, no matter what encoder is being used or whatever standard is being worked in.
This approach has several advantages. As mentioned earlier, BitSave doesn’t break anything. BitSave is also unconstrained by standards because it produces a pixel output, which means it can do things like use advanced neural networks and add multiple quality and low-loss functions that just can’t be achieved otherwise.
By essentially reverse engineering the metrics, BitSave takes advantage of a massive amount of research that has already been done in the scientific community regarding the perceptual assessment of content. BitSave can take an input stream and produce an output that is fully optimised according to the accepted standards of the perceptual quality metrics academic community, which delivers a much better perceptual quality, but by using far fewer bits than required by standard encoders.
