Mental Ray, and Future Proofing it...

[Eamon]

Woohoo! First Post

Ok,

haven't rendered anything thing in a year, but I have been following the developments over on the Maxwell Render thread since the beginning, and now more recently, over at Fryrender.

I think that a Mental Ray Thread has been long overdue, but I filled the gap with Jeff Patton's excellent Mental Ray Shaders thread over at cgtalk.com

 

Now, to the topic of my thread: has it come too late? Whilst fast becoming an industry standard on one hand, does Mental Ray represent an obsolete technology. ie.Photon tracing.

 

A while back, radiosity was the final word in photorealistic rendering, but then photon tracing superceded it. Now it looks like Path Tracing/MTL/unbiased rendering really is the latest, and maybe final stage in the quest for realism-all future developments will merely be in processing power.

 

Enter Maxwell and Fryrender. I'm wondering will Mental Ray be able to compete in this field (as well as on its own strengths)? I know that Loran over at cgtalk.com demonstrated mental ray using a Path Tracing material, but have people utilised this as a workflow?

[AJLynn]

It's not getting much use, but mental ray and Maxwell are for different things.

 

-mental ray is an all-purpose rendered with a lot of users in entertainment. Most movie animators aren't using any kind of GI yet (arch vis is actually the most advanced major field in that area) but with sufficient skill and manpower you can get it to do just about anything - The Matrix used and, and so does South Park.

-Maxwell is for doing still frames of architectural and product scenes with incredible accuracy, without requiring as much expertise or time commitment to use as mental ray. Basically, you can trade off CPU power for manpower under these circumstances.

 

It will take a few more Moore's Law cycles before people start doing serious animation with MLT, so if Mental Images wants to catch up they'll have time to do it. I think the fact that it does have a path tracing system, and the improvements they and Autodesk have been making with FG and the new materials system, show that they're serious about the design vis market.

 

BTW, I've been messing with the new mental ray, and it's really good, and a lot easier than before. I think with some practice I could prefer it to Maxwell.

[pixelperfectg]

Now, to the topic of my thread: has it come too late? Whilst fast becoming an industry standard on one hand, does Mental Ray represent an obsolete technology. ie.Photon tracing.

 

A while back, radiosity was the final word in photorealistic rendering, but then photon tracing superceded it. Now it looks like Path Tracing/MTL/unbiased rendering really is the latest, and maybe final stage in the quest for realism-all future developments will merely be in processing power.

I think you raise a good point. I figure we won't see photons go away per se, because they are still quite good for fast caustics for example. But I think (and hope) we'll just see more options added to our favorite rendering packages. Photons, path tracing, qmc...whatever.

 

There are just too many fields in 3d for one method to cover everyone. I hope we see developers give us more choices in GI calculation methods. That would allow everyone to find the one method that works best for them (or a particular scene)...That's one of the great things about Vray...all the options for GI.

 

But being specific to mental ray, I think we're currently seeing a push to use FG only instead of GI + FG as in the past. Too many parameters to tweak with GI & FG together. A lot of users will find it easier to simply enable FG and use the new presets...Sure, some will still use photons...but I'd bet the majority will use only FG now.

 

As far as path tracing goes, we (mental ray users) have a path tracing shader now (since Max7.5 I think). So that would lead me to believe that at some point a path tracing option might be introduced directly into the rendering rollout. Or better yet, maybe a hybrid of some sort WITH the option of pure path tracing....Of course that's just me thinking outloud...and I could be completely off base.

 

Bottom line, I guess photons may look like obsolete methods to many people these days. Especially with several path-tracing type solutions on the market now.

 

Jeff

[pixelperfectg]

I know that Loran over at cgtalk.com demonstrated mental ray using a Path Tracing material, but have people utilised this as a workflow?

LOL, I'm a dork. I just now noticed that you already mentioned the path tracing shader...sorry about that.

 

I tried it....but IMHO, it's nowhere near useable (far too slow). But I figure there's room for optimization with the path-tracing material. Example, right now to use it we'd have to increase the sampling amounts for the entire image (similar to using a glossy DGS material). If there could be some adaptive sampling applied...then maybe it would be more viable.

[manta]

"As far as path tracing goes, we (mental ray users) have a path tracing shader now (since Max7.5 I think)"

 

Where would we find that...

[pixelperfectg]

http://laurent.renaud.free.fr/loran-tutorial03.html