Intel has announced the creation of a new graphics research function within the Accelerated Computing Systems and Graphics Group (AXG). Heading up the new division will be relatively recent hire Anton Kaplanyan, Vice President, Accelerated Computing Systems and Graphics Group. Kaplanyan adds to his existing lengthy title the role of Chief Technology Officer and Director of the Graphics Research Organization.
In a blog post about the expanded research which will go on under his stewardship at Intel Graphics, Kaplanyan outlines several research thrusts; into creating a new open and reproducibly reference scene for graphics researchers, into new levels of photorealism using technologies such as path tracing, into machine learning graphics enhancements like XeSS, and into graphics technologies which will apply to Web 3.0 and/or the metaverse.
Kaplanyan reckons we need richer graphics not just for games but for future immersive experiences, which were discussed by Raja Koduri back in December. That means advanced graphics for remote work and presence, photoreal simulation, and 3D user-"photographed" virtual environments. To make this happen, we don't just need more powerful graphics cards in PCs, but powerful graphics acceleration across the whole gamut of devices from PCs, down through tablets and smartphones, to your home smart devices. Intel says it is readying for this task with the aim to "move forward the entire field, the whole ecosystem."
Talking about a vision is one thing, but Kaplanyan has already been working on something concrete in line with his research organization's pledge to follow an open and collaborative approach.
When Kaplanyan worked at Crytek, he published the iconic Sponza scene, which "became a default scene across gaming, visualization, film and other fields of research and development." Having such a standard allows graphics researchers to test their hardware and software innovations with reproducibility and comparability. In 2022 we will get an updated scene depicting the Sponza Atrium in Dubrovnik, Croatia. It features physically-based materials with 4K textures and high-resolution geometry, photogrammetrically matched to the stunning real-world location.
Sponza's upgrade is a near-term challenge-focused initiative, but one that still needs attention due to many devices still being shipped with rather weedy graphics acceleration.
Looking a little further into the future of graphics, Kaplanyan provides a broad sketch of research topics his Graphics Research Organization will be looking at in the coming months. There are three main prongs on this fork of activity for now. With the advancements in graphics hardware opening up teal-time raytraced graphics capabilities, Kaplanyan says Intel will be working on challenges to do with path tracing. He asserts that the industry isn't yet at a place where real-time raytracing is a practical solution for most.
Kaplanyan also mentions Machine Learning which he characterized as a "great approximation tool for challenging problems," including graphics. Work on XeSS is just the beginning of Intel's use for ML, says the graphics research boss.
Lastly, Kaplanyan sees the next boom in graphics being within 3D user-generated content. He shares his thoughts about making such content being as easy as video production – now a hobby, calling, or career for thousands of YouTubers.
Other than the sharing of the new enhanced Sponza scene, a lot of what Intel Graphics says it is working on sounds rather like an Nvidia GeForce catch-up project. As a relatively new venture that is somewhat understandable, and we must be patient. However, updates from Intel's Graphics Research Organization in, say, a year's time from now will hopefully show a bit more evidence of inspirational rather than me-too thinking.
Maybe they need another new one as the current Xe is not looking great?
Anyway, overall, I don't disagree. All things related to Xe's launch could have been managed a bit better, I guess.
Much like the group from NVIDIA that gave us FXAA.