The Future is finally here!* (almost)
Deep Learning Super Sampling (DLSS) Ray Reconstruction might make its way into Blender (at least, it's in the works). And the potential is, without exaggeration, astonishing.
Unlike "traditional" denoisers, DLSS has a different way of reconstructing an image, and largely ignores scene lighting or polygonal complexity. Because of this, it's used in games to "upscale" a low-res native render output to a much higher-res target, like 4K or more. It's very effective and, ironically, despite being an upscaler, nowadays is almost synonymous with the "best quality" rendering preset in games. Simply because more and more devs get lazy and spend less effort on optimization, especially anti-aliasing techniques, essentially "offloading" this task to a dedicated (and, unfortunately heavily proprietary) neuralnet upscaler like DLSS.
Like me, you might have wondered whether this very approach could be used in a traditional 3DCG Creation app like Blender.
Well, wonder no more! Someone took an unfinished alpha build of the Blender codebase with a DLSS integration, and compiled it into an actual test build.
Prepare to be amazed.
Remember "Brigade"?
Looking at this I can't help but speculate: if the AI boom hadn't shifted GPU manufacturers' focus away from consumer graphics cards, we might have seen entirely ray-traced games within one or two more GPU generations. Remember the Brigade demo for Siggraph from 2012?
I remember.
Or this one, from 2013?
And then, somehow, it all... faded away.
I expected an official announcement or a proper follow-up, but it was largely forgotten, following the release of Nvidia's RTX series. Which perplexes me to no end, to this day.
Well, 14 years later we may finally get the real-time ray-traced or even path-traced future we've been waiting for. It's far from perfect, and struggles with transparent shaders because it's not a "real" denoiser but more of a fancy post-processing effect. This makes sense, since it's designed for games, where transparency is handled very differently, and the geometric state of the scene can be properly interpreted by DLSS for spatial reprojection. BTW, For a deeper dive into real-time rendering techniques in games, see my post on Unity vs offline CGI.
Still, it could be very useful in many other cases, especially indoor scenes with complex, multi-bounce Global Illumination. I hope the video brings a lot of attention to this unfinished feature, to convince the Blender devs to prioritize it, and get Nvidia to support it officially.
Yes, it would end up as yet another vendor-locked feature, but I'd be the biggest hypocrite in the world, if I didn't admit that in this case I just don't care. Nvidia is a monopoly already. The vast majority of 3D workstations are built with their GPUs. So it's only logical to let the users make the most of their hardware.
I want this!