Graphics developers routinely face a trade-off when using ray tracing to create realistic, immersive experiences. They either allow for the hours (and hours) it takes to render high-quality images or use denoising methods to produce less than high-quality images, but do it a lot faster.
Intel® Open Image Denoise can remove this dilemma.
It’s an open source library of high-performance, high-quality, machine learning-based denoising filters for images rendered with ray tracing. As part of the Intel® oneAPI Rendering Toolkit, this library is suitable for both interactive and final-frame rendering, runs on almost any Intel® CPU, and significantly reduces rendering times.
It’s also completely free.
Join graphics software engineer Attila Afra for a detailed session that includes:
- The importance and impact of denoising when rendering images with ray tracing
- How Intel Open Image Denoise achieves high-quality denoising using machine learning (hint: it filters out the Monte Carlo noise)
- How to use the simple, flexible C and C++ API to integrate denoising into existing renderers in hours or even minutes
- Training the denoising neural networks yourself to achieve even higher quality for your renderer or implement custom filters