what's going on with iray?

[Anonymous]

Hi,

what happens there - NVIDIA released a new iray version and now you have to pay for it? http://www.nvidia.com/object/nvidia-iray.html

This is very disturbing!

Will iray be still a free, integarted rendering engine in 3ds max? 

[DarrenP]

that would be something someone from Autodesk to answer

i know Autodesk is working on a New Rendering Technology for 3ds Max called Autodesk Raytracer which is currently a autodesk labs project and already available in Autocad 2016 & Revit 2016

here is how you can offer feedback for it

The Autodesk Raytracer (ART) is our free plug-in for 3ds Max 2016 that provides easy-to-use physically-based rendering; physically-based as in simulating reality using principles of physics to model the interaction of surfaces and light. The technology preview runs locally on your computer but also has an option for sending rendering jobs to the cloud. You can even send multiple renderings to be processed in parallel, leaving your machine free. And unlike A360 Rendering (introduced in 3ds Max 2016), ART lets you render animations in the cloud as well as single images.

Join the project.

[Anonymous]

uh, sorry, we really don't need to wait and grow another rendering engine in the "childhood". iRay is good and getting better and better and now they really want to take it away from us..? Now, that everything gets much more expensive (my Design Suite Premium is going to be 4 times(!) more expensive with the new subscription model. THANK GOD i still have my perpetual license), they start cutting away features? WoW...This can't be true.

[Anonymous]

Well, Autodesk did add some great features to both IRay and Mental Ray for 3DS Max 2016

http://blog.mentalray.com/2015/04/20/mental-ray-for-3ds-max-2016/

[Anonymous]

yes and that's what we expect from Autodesk -  there would be no sense to upgrade without new improvements. But i don't pay subscription fees to loose features which i got used to...We don't pay for downgrades, do we...

[Anonymous]

it would be really honorable if someone from Autodesk just posts a denial. Because from what i hear even MentaRay might be dropped. Now THAT would be a real shocker.

[Anonymous]

Wow. Nobody using iray/mray? I'm the last? Autodesk not denying? So seems my goose is cooked. Merry Christmas.

[eodeo]

I'm using mental ray. I'm not a fan of iray. I am a fan of quicksilver that is based on mr, but is Autodesk's rendering engine based on nitrous, the Max viewports.

I am curious about ART. If it too is mr based, it might be worthwhile.

[Anonymous]

---

@Anonymous wrote:

 Because from what i hear even MentaRay might be dropped. Now THAT would be a real shocker.

---

Autodesk has already dropped Mental Ray in Autocad 2016 and Revit 2016 has Mental Ray and the Autodesk Ray Tracer engine

my guess is the licensing fees for mental/iray are getting expensive so Autodesk is working on their own rendering engine

[Anonymous]

@Ryan

Uh...Sorry, too expensive for Autodesk? The new, 4x more expensive licensing model of Autodesk should give them money to buy Vray, Cornona and all the others and serve them as subscription presents... And Autodesk couldn't deal with Mray and iRay so they sold them to NVIDIA - why should they be more successful with something new? I'm sure that "new engine" is just something Autodesk uses now to not loose their face to us!

But if Autodesk is really dropping Mental Ray in 3ds max, which would be much worse than dropping iray...Then there should be a special right of cancelation my subscription and i would want my money back! I have no use for 3dsmax without mentaly ray - all my projects for almost a decade depend on it!

[JamesClarke3D]

---

@Anonymous wrote:

And Autodesk couldn't deal with Mray and iRay so they sold them to NVIDIA

---

I realize you're just upset at whatever this overall situation may be, but FWIW Autodesk never owned mental ray. mental images, the company that made mental ray, was acquired by NVIDIA. (source) I won't presume to know or understand the economics and business operations for a company the size of Autodesk, but I assume Autodesk pays NVIDIA to include mentalRay with Max, hence the "*licensed for use by Autodesk" text on this page. 

[Anonymous]

Well i'm not upset until i really know what's going on.

I believe that it is not my business as a customer to know who owns what and so on.

Mental Ray is a part of 3ds Max since a.. Millenium. It has become a significant part of it, all the materials, everything has become more and more optimized for mental ray. I can't believe it will be dropped just like this, without announcing it, without premonition!

I still believe it will be part ot Max 2017, maybe i was crazy doubting that? 

[Anonymous]

I can help here, as I direct the Advanced Rendering Products at NVIDIA.

We just came out with the new Iray for 3ds Max plugin to allow everyone to use the full Iray engine. The Iray that comes with 3ds Max is actually a rendering mode of mental ray, and goes through it for translation. As a result, it's not as quick to start or interactive as the Iray you find in products like Catia, Bunkspeed or even Daz Studio now. It also doesn't have access to the rich material capabilities of MDL, the ability to scale beyond the local machine for distributed rendering (be it interactive or batch), or optional very fast "interactive" ray tracing mode.

We also like that we're now able to bring you our latest functionality and performance across Max versions, and you will see support for Max 2015 and 2014 coming out later this month. Those using earlier Max versions will see their Iray performance double on recent GPUs. We also have a version for Iray for Maya available now, are beta testing one for Revit, and other a couple non-ADSK products too.

We are otherwise hard at work making mental ray better as well, and you will see improvements coming for it too - although Autodesk needs to be the one that talks about details relating to its products.

- Phil

[Anonymous]

Thanks a lot for your help! At least NVIDIA is giving some details to us - although this is not the NVIDIA Forum. Thumbs up for such a spanning support, Autodesk should learn from you. I'm looking forward to mental ray improvements and i hope they will find their way into next Max versions for "free" (well not really free, we all pay autodesk for this kind of improvements).

I completely understand when a company (Autodesk) doesn't want to talk about it's future product versions in detail. But they shouldn't be stubborn when it comes to critical questions like in this thread. They should at least inform their customers if big changes are coming, which would demonstrably affect the workflow of their customers in a negative way. This i'd call fairness between company and long-year-customers. And even if all these rumors are wrong, they easily could just post a short denial rather than stubbornly holding on to their "future-version-silence-policy". 

[eodeo]

Do you know why does mental ray go through translating at all? It was like this since it was first implemented in 3ds Max 6 and after Max 2012 was released I was hoping that it will be fixed, since Autodesk based nitrous viewports and quicksilver on mental ray- committing to mr in a big way.

Now almost 5 years later, and still there appears no sign that they will improve Max' communication to mr. XSI looked to me like it does not have any translating at all, so at least it looks like it can be done.

I'd like to ask a question about iray too. How does it support Maxwell architecture? Are high end Maxwell cards faster than Kepler? I ask since they are in games, but have serious deficit of CUDA cores compared to Kepler.

As an avid mental ray user, I'm very glad that Autodesk based viewports on mental ray. I hope to keep seeing it in my Max.

[Anonymous]

Looks like there's some interest in how things work -

All renderers translate from the native Max scene description to their own definition for rendering - regardless of whether that renderer is ray tracing, DirectX, etc. So translation itself is not the issue, it's that the mental ray framework has different capabilities than the Iray framework.

We initially chose to use mental ray for Iray in Max so that we could leverage all the existing translation support already in place. Back in 2010, that approach easily got Iray into Max a full year earlier that would otherwise have required. But that ease of integration meant that it was saddled with the mental ray infrastructure - which was designed for robust pipeline production and not the interactivity and scaling that the native Iray architecture is. So to get those traits, we needed a new integration.

Quicksilver and Nitrous relate to Autodesk materials based on mental ray shaders (Arch & Design) but have nothing to do with mental ray itself. Both of these renderers were developed entirely by Autodesk and are Direct3D based, with Nitrous doing additional processing for its more advanced effects.

We are working to improve mental ray itself within Max, but can't comment on details before Autodesk does. As always, we work directly with the Max team on all such improvements. I promise to be back here telling you all the good news once we are able to!

Back on Iray and GPU types - we completely reworked Iray to take better advantage of Kepler and Maxwell for last year. If you have these cards, you should see a near doubling in speed between the Iray in Max 2015 and Max 2016. As a result, the latest Iray version will do best on Maxwell which better than Kepler which is then better than Fermi .

You may have the impression that Maxwell has less CUDA cores because we shipped the smaller versions first, but even these have more cores than their Kepler equivalent. The big/full Maxwell's then weight in at 3072 cores vs. Keplers' 2880. But unlike CPUs, the number of cores doesn't always reflect the actual performance, as GPU architectures keep evolving and improving, so the "value" of a single core can easily change between generations, and this will vary between compute applications as they often utilize the GPU differently. That said, core count and core clock speed together give you a linear yardstick for measuring compute performance within a given GPU generation for any GPU ray tracer (e.g., Iray, Octane, VRay-RT, Arion, Mosquito, Furryball, etc.)

I know all of this can be confusing. Wikipedia is a great source for the raw facts. The column there labeled "code name" is your guide to the type/size of GPU being used in the card. The "G" is a constant, the next letter is the architecture (K=Kepler, M=Maxwell, F=Fermi, etc.) and then the number indicates the GPU size, with a 110 or 100 being our largest, the 104 being the standard middle size and the 108 being the standard smaller size, and numbers like 107 & 106 being variations on the middle. A 200# means it's a second run on the same architecture.


- Phil





-----------------------------------------------------------------------------------
This email message is for the sole use of the intended recipient(s) and may contain
confidential information. Any unauthorized review, use, disclosure or distribution
is prohibited. If you are not the intended recipient, please contact the sender by
reply email and destroy all copies of the original message.
-----------------------------------------------------------------------------------

[Anonymous]

Hey Phil...

I'm a beta user and i'm following the discussions on the autodesk forums where the beta testers are aware of what's happening, and going to happen.

quote: "We are working to improve mental ray itself within Max, but can't comment on details before Autodesk does. As always, we work directly with the Max team on all such improvements. I promise to be back here telling you all the good news once we are able to!"

You realise they're saying the same thing? Are you seeing the forums for beta testers? They won't comment or explain anything, because they say NVidia will share this information... So which one is it?... Am I supposed to be following the forums on NVidia or Autodesk to read the news publicly?...

When I'm reading the questions people are posting here I'm very much agreeing with their reason for concerns. Although I cannot say anything since there is an agreement between beta testers and autodesk, I do implore you to contact autodesk and check some facts about who will be sharing what and when....

Phil [NVidia], please don't make the same mistake like autodesk of waiting a week before the new package is released to tell people what changing and what's not.

To everyone having questions like I'm reading in this forum, PLEASE, I seriously urge you to become a beta tester because that's where it's all being placed under the loop and it's there where you can make a difference. You want to be closer to the max-developers who create your package from UI to bug fixes? THEN APPLY FOR BETA-USER. If you have the time to be on these forums, you have the time to be a beta tester. You do, you do, you do!

And most important, you can read the awnsers you're looking for there. + they're LOOKING FOR BETA TESTERS, SO APPLY?!

Regards,

Ethan

[Anonymous]

>>You realise they're saying the same thing?

Not at all. mental ray is a product unto itself that is controlled by NVIDIA. When and how Autodesk exposes its functionality within its products is controlled by Autodesk. So you can read the development blogs on how we're evolving mental ray and Iray, but that isn't necessarily relevant to the readers of this fourm until it's integrated into 3ds Max, etc.

>>please don't make the same mistake like autodesk of waiting a week before the new package is released to tell people what changing and what's not.

Fair enough as it's within the Beta program. Jenni usually does a pretty good job of informing people of rendering improvements up there though.

And your points about participating in ADSK Beta's are quite true. If you can get in you should, and if you do get it, participate like mad because you can make a very real difference. Same goes for our Beta's BTW!

- Phil

[eodeo]

Thank you for your reply.

You may have the impression that Maxwell has less CUDA cores

I didn't want to get technical earlier, but I was comparing 780/ti to 970/980. In most games 970 with 1664 CUDA cores is faster than 780ti with 2880 CUDA cores. I understand that double precision hardware was cut from Maxwell making it much less effective than Kepler in those situations. Not knowing how Iray works and since I have a GTX 970, I was wondering if 970 is faster than 780ti in Iray as well as games.

[Anonymous]

>>In most games 970 with 1664 CUDA cores is faster than 780ti with 2880 CUDA cores.

Yes- that's because Maxwell is faster than Kepler for graphics. Core count is only a reliable metric of performance comparison within the same GPU generation. That 970 also has a core clock of 1050 MHz vs. the 780's speed of 876 MHz. Also note that the780 was a GK110 (large die) while the 970 is a GM204 (medium die) meaning it is out performing the previous generation with half the silicon.

>>I understand that double precision hardware was cut from Maxwell making it much less effective than Kepler in those situations.

Double precision is really only used by analysis, scientific and HPC software. DP is never used in graphics and only some specialized scientific visualization tools use it for ray tracing. Every GPU accelerated ray tracer you'd use in Max is only single precision, so you're not missing anything for use in Max.

>> I was wondering if 970 is faster than 780ti in Iray

It should be. You should also consider the 980 or Titan for twice or four times the memory though.

- Phil