What is your opinion on OctaneRender?

Anything and everything that's related to OGRE or the wider graphics field that doesn't fit into the other forums.
Post Reply
Hotshot5000
OGRE Contributor
OGRE Contributor
Posts: 226
Joined: Thu Oct 14, 2010 12:30 pm
x 56

What is your opinion on OctaneRender?

Post by Hotshot5000 »

It's and unbiased renderer. Right now it only works on nVidia since it uses CUDA but they say they will make it work on OpenCL and that will work on everything.

https://home.otoy.com/render/octane-render/overview/

https://home.otoy.com/render/octane-render/faqs/

Is this a potential threat to the more 'classical' Unreal, Unity or OGRE renderers? Right now it seems it's more for offline rendering, but it also works realtime (it's very noisy for now) and its performance will likely improve with future GPUs.
User avatar
dark_sylinc
OGRE Team Member
OGRE Team Member
Posts: 5296
Joined: Sat Jul 21, 2007 4:55 pm
Location: Buenos Aires, Argentina
x 1278
Contact:

Re: What is your opinion on OctaneRender?

Post by dark_sylinc »

This question (Rasterizers vs Raytracers, real time of course) pops up every year; and just like the Year of the Desktop Linux(tm), it never seems to arrive :lol:

Seriously speaking: We're seeing a convergence towards hybrid models; but raytracers are very far in terms of framerate from competing with raytracers.
The main advantage raytracers offer is that everything "just works". Reflections? Refractions? Depth of Field? Global Illumination? You don't need a custom shader or postprocess compositor (aka smoke and mirrors) to implement or fake it. Furthermore the same system for tracing rays in graphics can also be used by the physics.
The main disadvantage is that raytracers have no upper bound. It's not O(1) like a rasterizer. A ray hitting a complex refractive surface could bounce a lot of times before converging to a desirable result. A house of mirrors could take forever to render.

Since Moore's Law is coming to an end; unless we see a major breakthrough in processing power (e.g. quantum computers); I don't see raytracers becoming a threat to rasterizers anytime soon.

But hybrids? That's another deal. That shadow mapping algorithm research sitting in my Trello todo list? Relies on very basic realtime raytracing. Heck, Terra's shadow system (the Terra sample from 2.1) relies on a compute shader raytracing a heightfield. The Instant Radiosity (Global Illumination) we just implemented in 2.1 also relies on raytracing for placing the VPLs. Raytracing is also interesting for AO (not SSAO).
frostbyte
Orc Shaman
Posts: 737
Joined: Fri May 31, 2013 2:28 am
x 65

Re: What is your opinion on OctaneRender?

Post by frostbyte »

2 month ago while investigating about lightField theroy( lots of nice github projects...), i've reached OTOY's website
it seems it's more for offline rendering, but it also works realtime (it's very noisy for now)
i think it runs on 6-teslas for now( or was it 2 teslas )...any way for now its way to heavy...
ok..now i remember....noisy demo-2 teslas( 1080? ) and the guy said that in order to get rid of the noise in real-time you would need 6-teslas
but he also added that they are working on getting rid of the noise- without additional hardware( probably with some fancy interpolation method )
Is this a potential threat to the more 'classical' Unreal, Unity or OGRE renderers?
well, the so called "classical" has come a long way - i've seen some 'classical' rendering demo's almost indistinguishable from 'ground truth' ray-tracers, so don't forget that rasterizers are also making a big progress...

Unity is working with "Imagination technologies" to create a Hybrid-renderer with support for the new hardware based ray-tracer engine
Epic-unreal has announced some kind of collabiration with OTOY, Epic founder- tim swenney have shadered his deep views on the subject....https://twitter.com/timsweeneyepic/stat ... 0905657344
i think its safe to say a new Hybrid-rendering dawn is slowly emerging

the fact is that OTOY is investing in a cloud based solution, so they probably don't see their topTier tech running on commodity hardware anytime soon

regarding hardware ray-tracers, it's nothing new - intel has been working for years on this
regarding imagination- notice they talk about x100 increase in watt/performance... not x100 in raw performance...
i think its purposed at bringing hybrid rendering( avalble today only on high-end gpus ) to mobile and lowend market( targeting mobile vr? )
but the truth is that until we can test it, we can't really know what the fuss is all about....

about 3 years ago iv'e read an interview with some of silicon-graphics top guy, claiming that near-perfect real-time photo-realistic rendering by traditional ray tracing will be achieved within 20 years...
by resurrecting ancient light field theory and using path tracing OTOY has figured out how to make this happen faster...

regardless, rasterizers will always have their role to play, for most games don't need, want or try to achieve photo-realism( angryBirds... )
hybrid-rendering as the name suggests still requires a rasterizer, and full blown blown realtime, true ray-tracing renderer is still beyond reach(in the clouds...)
so overall i don't think it should be considered as a threat but as an opportunity...

Edit: quoting myself :)
he also added that they are working on getting rid of the noise- without additional hardware( probably with some fancy interpolation method )
https://www.innobright.com/
the woods are lovely dark and deep
but i have promises to keep
and miles to code before i sleep
and miles to code before i sleep..

coolest videos link( two minutes paper )...
https://www.youtube.com/user/keeroyz/videos
User avatar
stealth977
Gnoll
Posts: 638
Joined: Mon Dec 15, 2008 6:14 pm
Location: Istanbul, Turkey
x 42

Re: What is your opinion on OctaneRender?

Post by stealth977 »

dark_sylinc wrote: Since Moore's Law is coming to an end; unless we see a major breakthrough in processing power (e.g. quantum computers); I don't see raytracers becoming a threat to rasterizers anytime soon.
FYI, Contrary to the common belief, quantum CPUs are not general purpose CPUs, they are somewhat dedicated hardware for solving some very specific equations. They excel at two equation solving categories:

1 - Solving O(N^2) equations in linear time O(N), for ex: Travelling salesman
2 - Solving complex equations with tendency to converge at a specific result, in linear time, for example Factorization of large integer numbers (aka Security Key Deciphering)

other than the above, and maybe a few other equation types, quantum CPUs perform much worse than usual CPUs when it comes to general purpose computing.

Currently main advantage of quantum CPUs is equation #1 above. Solving exponential difficulty equations in linear time, which can be used to:
a) Perfectly predicting the weather conditions, rain, snow, heat for the next couple of months to couple of years ( =) )
b) Coupled with Deep Learning and Data Vault 2.0 modelling, can be used to predict almost everything, from market share prices of any given stock to crime rate prediction of any city, or possible terrorism targets whatever...
c) Perfect Prediction of anything depending on Data Vault Size & calculation efficiency of quantum CPUs

It is IMHO not a coincidence that one of the biggest players in Quantum Research (Google), also became one of the biggest players of Deep Learning/AI Research by company takeovers and hiring of experienced researchers last year...
Ismail TARIM
Ogitor - Ogre Scene Editor
WWW:http://www.ogitor.org
Repository: https://bitbucket.org/ogitor
Post Reply