Experience?

[MG]

I was wondering if there were any people out there having experience with rendering with mental ray 3.0 and Houdini?

Anyways, I haven't really done it yet. And I failed at my first attempt. (Which might be because of my environment variables).

[Jason]

We tried to render a GI pass of the NY pullout scene in mental ray 3.0 during Time Machine, but a bug in mental ray caused the light to flicker when we pulled out of the scene so far. I hear mental images have subsequently fixed this bug but I haven't seen proof of it yet.

I'd love to know if anyone else is using the mental ray output from houdini these days.. it's a lovely renderer but it has a steep learning curve especially with the lack of decent shader front-end and the fact you have to write C++ shaders. I am quite a fan of the renderer, but some serious steps have to be taken by mental images to make it more usable with houdini.

Cheers,

J

[MG]

We tried to render a GI pass of the NY pullout scene in mental ray 3.0 during Time Machine, but a bug in mental ray caused the light to flicker when we pulled out of the scene so far. I hear mental images have subsequently fixed this bug but I haven't seen proof of it yet.

I'd love to know if anyone else is using the mental ray output from houdini these days.. it's a lovely renderer but it has a steep learning curve especially with the lack of decent shader front-end and the fact you have to write C++ shaders. I am quite a fan of the renderer, but some serious steps have to be taken by mental images to make it more usable with houdini.

Cheers,

J

Wow! You have actually worked on the special effects for the "Time Machine"?

About mental ray, well I'd only render still images really. Not because I don't think mental ray is suited for sequential rendering work, but it requires indeed such high cusomizations and careful set ups. Not only in Houdini, but also in XSI (which is the package where mental ray is best integrated).

And I haven't really written custom shaders yet for mental ray. It's quite different, than say, writing RenderMan shaders (and especially with Houdini!).

[betty]

Jason

Interesting to hear you guys where trying to use it in production with Houdini..apart from GI reasons, was there any other benefit for choosing to render in Mental Ray other than Mantra?

Re; "..decent shader front-end..", do you mean a work environment like the render tree in XSI or hypershade in Maya? So therefor i guess all shaders must be programmed from scratch to work in Houdini.

How do you compare Mantra witht Mental Ray? What advantages/disadvantages do you see?

And apart from the front-end interface what else do you think is needed so Mental Ray could be more usable with Houdini?

[Jason]

Well, at the time during Time Machine, we were hammering mantra and we were responsible for a LOT of bugs being fixed. SESI was absolutely incredible, giving us new versions of mantra every 4 days or so..

We took Charlotte Manning on to the show - an incredibly strong mental ray TD who wanted to add to the shot using mental ray's GI - so she made quite a few attempts at getting final gathering to work on the scene. Unfortunately, a bug prevented her from making the images nice so we went for a lightdome style lighting effect on the NY scene.

The cons of mray are: houdini has no shader-tree style thing and the default mental ray shaders are quite a bit lacking as far as ready-to-use-ness is concerned. Writing decent shaders are a fair bit more complex than writing shaders in vex and renderman sl each individual platform requires it's own compile of the shader. We at DD render across NT, linux and IRIX and so it's a pain to compile on all of those platforms all the time.

Mental ray also has a large amount of black magic - BSP depth and so on can really affect your renders. There are a million options which have all kinds of quiality versus performance tradeoffs. It seems a great stills renderer or a great academic renderer. It has nice features like area lights and GI - the main attraction.

So Mantra turns out to be easier to learn and use, although it still does have a slew of settings. The default setting are pretty intelligent and work for most intensity and scale scenes.

And mantra is far quicker..

On average, I'd say that the NY scene were 4hr mantra renders (with lightdome), but 18hr mental ray renders (with low quality GI)

Eventually we were forced into using mantra for the illumination of the shot because mental ray fell over. I would have loved to see mray do it well though.. too bad.

If mental images wrote a houdini front-end like they've written for maya, it'd be an excellent alternate to mantra and prman11.

My 2c.

J.

[betty]

Jason

I think that was a bit more than 2c, very valuable insight....thanks.

And the comments about mantra being faster, well that suprises me! I suppose it shouldn't because as you've seen there's a hell of a lot for me to learn about mantra.

I'll make it a priority to use it for my next project and hopefully i'll have v5.5 then so i don't fall into any silly traps....looking forward to it.

The thing is though all these "things" like caustics and GI are missing which i suppose maybe SESI will work on for future releases!?!?

btw, i was just flicking through 3Dword and i saw some stills from "time machine" including a NY pic,.......nice stuff!

[Jason]

Hi betty,

Very late reponse, but thanks for the compliments on the NY shot - that was a lot of hard work of a team of 7 people, some of whom had never used Houdini before. Very good work for the short period time time that we had to make the shot.

As for GI, etc .. I can only assume that GI is one of the last huge technologies that will be inevitably added to mantra. It's bound to happen, so the only question is "When?"

I am very excited for that day whenever it is, because knowing the mantra development guys, it'll be done right.

Cheers!

J.