@wedusk101 Profile picture

Navonil Mukherjee (Neil)

@wedusk101

Graphics research @Qualcomm. Ray tracing fanatic, neural rendering afficionado, math nerd, Crysis cultist. Personal opinions only. RT != endorsements. (He/Him)

Joined July 2014
Similar User
Jakub Boksansky photo

@boksajak

voxel raytracer photo

@voxel_tracer

Tobias A. Franke (@thefranke@graphics.social) photo

@thefranke

Guillaume Boissé photo

@guitio2002

Dale Weiler photo

@actualGraphite

Anton Schreiner photo

@antonschrein

dydx photo

@ouyangyaobin

Ingo Wald photo

@IngoWald

Alexandre Lamure photo

@AlexandreLamure

Gabriel Sassone photo

@GabrielSassone

Konstantin Khomyakov photo

@kkhomyakov3d

X photo

@PixelClearr

Pete Morley photo

@Biblo_Shwag

Nikos Papadopoulos photo

@4rknova

Gijs Richard Kaerts photo

@BelgianRenderer

Pinned

Renders from my unidirectional path tracer, Cake. Been working on it on and off for the past couple of years. Supports Embree, OIDN, OpenImageIO and other cool stuff. Sky is using the Preetham model. I would like to thank @Peter_shirley, @Atrix256 and @KostasAAA among others.

Tweet Image 1
Tweet Image 2

Navonil Mukherjee (Neil) Reposted

What does game dev optimization look like? When I was in college, one of my professors told the class: Do not optimize code because the compiler is smarter than you. He was wrong. As an example, here is the timeline of a frame from Uncharted 2 (my old gdc talk).

Tweet Image 1

Navonil Mukherjee (Neil) Reposted

Following over 1.5 years of hard work (w/@njroussel& Rami Tabbara), we just released a brand-new version of Dr.Jit (v1.0), my lab's differentiable rendering compiler along with an updated Mitsuba (v3.6). The list of changes is insanely long—here is what we're most excited about🧵

Tweet Image 1

Navonil Mukherjee (Neil) Reposted

MPM continues to be the GOAT multi-phase physics solver, and I always like to point out / brag that it was introduced into computer graphics by @DisneyAnimation

I've been doing simulation and vfx studies for years and I must say, Bifrost MPM is just mind-blowing. The simulation quality and performance are on another level. and Yes, I think it's better than Houdini. Made in @AdskMaya and Bifrost, rendered with @arnoldrenderer



Navonil Mukherjee (Neil) Reposted

2014 LOW LEVEL OPTIMIZATIONS FOR GCN michaldrobot.com/wp-content/upl…


Navonil Mukherjee (Neil) Reposted

Classic 3D videogame shadow techniques 30fps.net/pages/videogam…


Navonil Mukherjee (Neil) Reposted

Khronos has announced the launch of the new Slang™ Initiative. This initiative will oversee and advance the open-source Slang shading language and compiler, building on 15 years of research, development, and deployment experience by @nvidia Learn more: khr.io/17f

Tweet Image 1

Navonil Mukherjee (Neil) Reposted

This Siggraph 2020 Last of Us II optimisation talk is a great example of what improves the frame performance can sometimes appear counterintuitive if seen in isolation. s3.amazonaws.com/nd.images/rese…, youtube.com/watch?v=CvS6rX… 1/2


Navonil Mukherjee (Neil) Reposted

tiny_bvh.h version 0.9.5 is out: github.com/jbikker/tinybvh New in this version: CWBVH traversal! This is an advanced data structure with some matching GPU code: There's inline PTX, architecture-specific paths and more. Let me know how it runs on your GPU. :)


Navonil Mukherjee (Neil) Reposted

25000 allocations per key press in Chrome :( I once helped a game project optimizing their allocations. Had around 1000 allocations per frame initially and we got it down to around 10 allocations per frame. Stalls were gone. Also did the same for load times. Over 2x faster load.

This guy worked on Chrome DevTools lol. Here’s what Chrome is known for…

Tweet Image 1


Navonil Mukherjee (Neil) Reposted

Offline: I remember scanline (3ds, prman...), hybrid (ligthwave, 3dsmax...), raytracing (imagine, real3d, povray, bmrt...), photon map (mental ray...) - but what was the first real, non-research path tracer? Can't remember! Took me a bit to find this again pointzero.nl/renderers/


Navonil Mukherjee (Neil) Reposted

Programming on Parallel Machines: GPU, Multicore, Clusters and More heather.cs.ucdavis.edu/parprocbook // free ebook


Navonil Mukherjee (Neil) Reposted

tiny_bvh.h v0.8.0 is out now. github.com/jbikker/tinybvh New: - Binned SAH BVH construction for ARM / NEON - BVH traversal using NEON for the AltSOA format. NEON support was added by contributor wuyakuma. The NEON builder halves build time on M2 compared to the reference builder.


Navonil Mukherjee (Neil) Reposted

And here are the slides from my game engine architecture talk at Virginia Tech last week. These are mostly an outline of the topics I discussed while showing a bunch of real-time demos, so they don't contain much of the content of the talk, unfortunately. terathon.com/vtgame_lengyel…

Tweet Image 1

Navonil Mukherjee (Neil) Reposted

tiny_bhv.h version 0.7.5 has been released: github.com/jbikker/tinybvh This version adds support for WASM / @emscripten, brought to you by Christian Oliveros.


Navonil Mukherjee (Neil) Reposted

Intel Open Image Denoise 2.3.1 is out with a couple of fixes and improvements: github.com/RenderKit/oidn… #IamIntel


Navonil Mukherjee (Neil) Reposted

Some details about the great strand hair tech used in Dragon Age: Veilguard. A 6.5ms budget for hair rendering is very impressive! ea.com/technology//ne…


Navonil Mukherjee (Neil) Reposted

tiny_bvh.h version 0.7.0 is out: github.com/jbikker/tinybvh This version adds CWBVH construction (and, for testing only, slow CPU-traversal of the data). CWBVH is the state-of-the-art algorithm for #RTXOff ray tracing on the GPU; it stores data for 8 nodes in just 80 bytes.


Navonil Mukherjee (Neil) Reposted

Performance people, I need your help ! What's the fastest way to open a lot of files on windows ? CreateFileW is slow as hell (slower than reading the file itself), are there ways to make it faster (some flags ? warming the MFT ?), obvisouly without changing the files? Thanks !


Navonil Mukherjee (Neil) Reposted

For those looking forward to spatial splits in tiny_bvh.h: This has now been added, in version 0.5.0. Spatial splits significantly reduce the cost of ray queries in a BVH (by about 25%). An SBVH can still be optimized further with the BVH-post-optimizer. github.com/jbikker/tinybvh


Loading...

Something went wrong.


Something went wrong.