An innovation while pushing boundaries in content delivery
So while I was working on pushing more and more content on older hardware (nVidia GeForce GTX 1050Ti), I ended up with a technique that makes visibility buffer/deferred material rendering faster
for rendering engines that use visibility buffers/material rendering without a DAIS buffer. One notable example is rendering of Nanite geometry in Unreal Engine 5.
Check it out in the graphics programming weekly!UPDATE: turns out I independently rederived Burns and Hunt's original approach with some notable differences:
I do ray-picking on the near plane while they inverse project fragments to get ray directions.
I also have no need for transforming geometry since all my geometry is post-transform from compute while they still have to transform oriented instances.
Cheers,
Baktash
March 7th, 2022
C.L.A.S.H is still kicking...
The world is going to hell in a handbasket. But I've managed to keep things going for now :)
After many months of engine upgrades to support asset streaming and processing via multi-threaded rendering (along with some situational setbacks), I finally have some more updates!
Feast your eyes on a scene of 7.6M triangles, path-traced in post-process using SDF shallow binary LBVHs. The skybox alone is 1M triangles.
Probably the least opportune time for post-apocalyptic scenery, but finally got some free time after a few busy months.
— TooMuchVoltage Software (@TooMuchVoltage) March 7, 2022
The above thread has a link to a rough benchmark and relevant technical details for the curious.
Here's to hoping the world calms down soon.
Cheers,
Baktash
After a literal decade in the making, it is finally here:
Yes it will be running on HighOmega v3.0 (3.021 as of this moment).
Yes it will be path-traced and de-noised. And yes, raytracing will be supported on Radeon VII and the RX 5700 XT.
I cannot put in words the amount of relief I feel from putting this out in the wild as I've had other ventures take up a good chunk of my time recently.
I hope you like this production as much as I've been dreaming of and longing for it for the last ten (10) long years.
Featuring artwork from the wonderful and talented Vittorio Garofoli and Florian Feldbinder.
Follow and hit me up on Twitter if you have any more questions as I will probably be most available there.
Also check it out on Reddit... if that's your thing.
Wishing you times as happy as these are for me,
Baktash
January 9th, 2021
NV to KHR ray tracing migration
Firstly, happy new year if I haven't told you this personally! :).
If you do ray tracing in Vulkan you might find this write up of mine of immense help :).
This will help you move to cross vendor Vulkan raytracing in case your code (still) needs nVidia hardware.
It was also featured in the Graphics Programming Weekly - Issue 166 :).
Cheers,
Baktash
September 9th, 2020
Come check out our i3D 2020 presentation!
You probably don't want to miss this.
I will do live tweaks and source review for this exclusive live demonstration! Don't forget to register here. (it's free this year!)
Cheers,
Baktash
May 21st, 2020
HighOmega v3.02 is here!
UPDATED: June 18th, 2020
I encourage you to try it out by getting it from GitHub and leaving your feedback on reddit. This will be the last release of this demo before an upcoming game. (fingers crossed!)
Your feedback is of utmost importance and will be implemented if it is crucial and within reach.
Cheers,
Baktash
March 2nd, 2020
Mesh booleans seem pretty popular these days!
UPDATED: March 12th, 2020
And we're back in the graphics programming weekly! :)
What a way to start posting for the new year!
Click here so you don't miss all the fun.
It contains the secret ingredient! ;)
Here's how it looks in-engine (... the reddit post has better quality video though):
Cheers,
Baktash
September 24th, 2019
Shadertoy experiments!
So it's been quiet here for a while. And that's because I've been taking a break making Shadertoys :).
Already four of them have been featured in Technically Art which makes me very proud!
Check them out here!
Here's a preview for those of you without a decent card:
Wanna know why everything is so special this time? Check this video out :)
Woohoo!
Baktash
June 12th, 2019
Announcing HighOmega v3.0: King Mackerel. No RTX? No problem :)
UPDATED: June 28th, 2019
Firstly, welcome to our new website! Feel free to look around. Hope you like what you see.
Secondly we just recently released HighOmega's most recent iteration: v3.0 code-named King Mackerel.
This release's rendering pipeline mainly relies on denoising path-traced images produced by either RTX or our very own Voxel-based hybrid path-tracing technique described here.
This means that having an RTX capable card is not necessary for getting denoised path-traced graphics.
Please note that not everything in the demo is path traced: distant geometry, the sky, transmittive surfaces and light shafts use traditional means of rendering.
The aurora borealis is built on a special kind of noise called Triangular Noise developed by Nimitz.
More details can be found here or here.
If you liked the demo don't hesitate to get in touch.
Until next time,
Baktash
About
Est. 2010. Toronto, ON, Canada. Working on cool stuff ever since.
Talks/Write-ups
Toronto ACM SIGGRAPH Tech-Talk
September 29th, 2015
This talk detailed the OpenCL-based fluid simulator in HighOmega v2.0.
It detailed our challenges with OpenGL interop and overcoming crashes faced with Quadro drivers.
i3D 2018 Poster
May 15th – 18th, 2018
This technique is the foundation behind HighOmega v3.0's light-transport based rendering pipeline. It allows for denoised path traced graphics irrespective of whether or not raytracing hardware is present.
The technique as of HighOmega v3.0 release: Technical reddit post here or here
i3D 2020 Poster
September 14th – 18th, 2020
This is a replication of CryTek's Neon Noir raytracing technique with improvisation for dense geometry without proxies. Please note that this was based on an early interview and they have since moved on to a different approach.
The technical details mentioned within can be found in the technical brief.
The source code is available here under a commercial license.
Partners
From february 1st, 2016 to September 30th, 2017 TooMuchVoltage Software Inc. was responsible for extending and maintaining BigData and Machine Learning cloud infrastructure for one of Autodesk Research’s projects. We are proud to have retained our collaborative relationship with Autodesk Research to this day.
SauRayTM inducted us in the nVidia Inception program as most likely the only non-AI program in Inception.
Quite a ride while we were actively promoting SauRayTM :)!
The detailed algorithm is described in our patent application.
For a higher-level picture, you can read our publicly available technical brief.
The source code is available here under a commercial license.
Artwork
Done in collaboration with Somatone's former Art Division.
Cheers,
From february 1st, 2016 to September 30th, 2017 TooMuchVoltage Software Inc. was responsible for extending and maintaining BigData and Machine Learning cloud infrastructure for one of Autodesk Research’s projects. We are proud to have retained our collaborative relationship with Autodesk Research to this day.
SauRayTM inducted us in the 
