Ehy guys !
Here come Rombo.tools for mentalray (soon for Arnold too), a specialized shader pack with latest advancements in CG. We’ve already a lot of scenes ready to be rendered for instant fun and learning. You can download a demo from here. Don’t be scared to register to the Rombo.tools website to get notified of all the new additions coming very soon.
Nice work Max, but i dont get me wrong, mr is not so great lately, in my case with dual xeon cpus mr detects only 1/2 cpus and when you try to render anything with it everything is soooo slow…
So my advice to you, i know u said, is to make shaders for other alternative renders, vray, arnold, renderman… In this moment they have much brighter future than mr 
cheers
Thanks for the feedback.
As for the future, we dont’ belong to anything. We have our own rombo.renderer. Rombo.tools is an high-quality high-performance renderer agnostic framework. We run the same exact code on both mray prman and arnold and mray for sure is not behind the others***. In facts latest mental ray progressive sampler is simply outstanding and GINext (with GPU) too. We tested it together with rombo on quad xeons, latest intel and amd cpus and on render farms , really as expected, - no problems at all. We started from mental ray just because it has less ‘shading features’ than other renderers.
*** we had not a chance to test latest vray but it looks 2017 mental ray is up to vray quality and speed.
https://antoniobosi.com/home/render-comparsion-tests/v-ray-vs-mental-ray-render
ciao, max
Interesting article,
Btw i was using an old mr build, think it was 3.14 or something like that, so its possible the build is way too old to detect multi core systems, so im definitely going to try a new mr and see how it goes 
cheers
Do you mean multi-CPU socketed systems?
mental ray has always utilized multiple cores, since 2004 when I started using it anyway.
yes, multi socket,
so far didnt work for me, until i did an upgrade yesterday to mr 1.01 (3.14.4.9) only then it started to use all my cores
MentalRay is not so bad. Our company worked with Prman.RIS and MentalRay. And for one of the new projects we choose MentalRay again, as it has a number of advantages.
Issue with more than 64 cores is only for Windows version.
Windows introduced this issue with some changes to newer Windows installs forcing developers to change how they find resources.
And it was fixed in mentalray some versions ago.
Okay, my experience with mentalray is a bit dated, but we made a few movies with mentalray and it was hell. We sometimes have large scenes and used large fluid boxes. Empty fluid boxes started to create an alpha, stacked fluid boxes always started to show box borders, fluids were terrible unreliable, I had flickering and popping fluids in a lot of places. Particles rendered sometimes, sometimes they refused to render, sometimes they started to need an incredible amount of time or memory. We had to render every shot at least two or three times only to get all frames done. Creating procedural geometry is terrible and complicated, shader coding is still quite annoying.
“Empty fluid boxes started to create an alpha, stacked fluid boxes always started to show box borders, fluids were terrible unreliable, I had flickering and popping fluids in a lot of places. Particles rendered sometimes, sometimes they refused to render, sometimes they started to need an incredible amount of time or memory.”
Have you had better experiences rendering Maya’s fluids, in other renderers? I’ve only used fluids on a handful of projects so far, and only use Vray or mental, but it seems like most rendering engines won’t really play nice with Maya’s fluids. That said, I’ve been successful with mental ray and fluids in my environment/landscape scenes, but mostly just using them for atmosphere or clouds. I don’t do much animation.
I’m curious if Rombo has a different approach to Maya’s fluids in the works.
Indeed. I tried a few maya fluids with Arnold it worked really fine, exactly as expected. We had a vulcano eruption with heavy fluids and I simply didn’t manage to render the scene with mentalray because of failures. The first attempt with Arnold worked.
Ehy it would be cool if you can simply stop with all these bullshits about mental ray. This thread is about Rombo.tools. Thanks.
For example this is part of the upcoming (late autumn probably) Rombo for Arnold … amongst other things, - it introduces adaptive importance-based sampling in Arnold 5. Wait… what ? … adaptive sampling for the Arnold renderer ?!? Yeah I heard you, however, let’s see a basic scene where we may need adaptive sampling.
It may look an innoncent scene to render… I mean it’s just a simple sphere with an HDR found on internet as domelight… but it’s very difficult to sample because the overall env remains dimmed while there’re a lot of tiny super bright lights… and you can cranck up the sample settings as much as you want that it will never get resolved with plain importance-based brute force sampling. For example before give up I tried with 6400 samples (8x8(camera) x 10x10(specular)). They are already prohibitive settings if one has a real scene and not just a simple sphere. The problem is that they are still not enough.
Not bad but if we zoom in we can see how much still has to be done to get a fully converged image.
That will flicker like hell in anim. Not only. But if you take a look at the first raw image above you may see that the bright tiny lights are more scattered on the metallic surface and should create a kind of metallic halo (that is typical of metals) instead to be kind of clamped. Let’s see where Rombo can help.
Not only we’re 2x faster but we better resemble metallic appearance because the sampling instead to clamp the overall reflection, just fill it with more samples where needed creating a nice metallic halo. And if we zoom in we can see that the image is almost fully converged.
Here we shoot just 4x4 camera rays while the adaptive shading sampler is set to max:2^8, min:2^4 … that means we shoot at max 4096 (4x4 x 256) and at min 256 (4x4 x 16) samples.
You may not even need most of the time to go adaptive but for VFX with HDR plates it really alleviates sampling a lot. And for physically based metals too. And just that we talk about metals before i forget… next time I’ll show you a new generalized BRDF with a gamma parameter to be able to fully model microface distributions instead to just choose between GGX vs Beckmann.
cheers, max
Thanks Haggi !
Here an application for Rombo.Metal rendered with Arnold 5.
We don’t need adaptive sampling on the mecha as long we don’t animate it because reflections have to look broken and rough so probably one can’t really spot what is a sample artifact and what has intentionally to look kinda broken, however I used it on the floor where a lot of fireflies were giving me headache. The cool thing is that sample correlation from Arnold 5 core still works with adaptive sampling, in facts if you look close at the floor you can still see the correlated sampling pattern. Settings have been left to average values however I believe Arnold has one of the fastest core I ever tested and with Rombo addons it gets fully exploited, the image below took exactly 4:30 mins to render at HD720 on a single processor i7Haswell dev machine !
ciao, max
Ehy kids keep an eye on this … upcoming denoiser for arnold.
https://www.facebook.com/rombotools/
https://www.rombo.tools/
ciao, max
thanks baz ! here rombo denoiser for Arnold dealing with strong hdr highlights while keeping full detail of ibl reflections. dof artifacts are also resolved.
ciao, max