I'm quite frustrated about that vray thing.
Just had a fast job - many interior renderings. I don't have a commercial houdini. I modeled everything in xsi, tried to render (had 2 hours to the meeting) but somehow it didn't want to render.
When I exported everything to max vray rendered whatever resolution with GI in a matter of minutes, quite pretty (without shaders).
This really makes me angry because most of the time I'm forced to use max because of this renderer, it gives great quality in no time, is fast and firm.

Now after this foreword, I know a bit how to tweak mantra's GI, but one bounce will never give this beautiful soft look fast. Deadlines are very tight. I'd like to use houdini for work oneday but from what I see it would be impossible now. Are we going to get multibounce, fast global illumination in the future?
I know this is a question without answer, but now the only thing I could do is to use AIR (which is fast and very attractive) or Mental Ray (which is old and clumsy), but I'd need to spend 17 k and would not have the power and control of shading as in mantra.

Oh if only Vray was rman compliant …

Mark? Andrew? Care to comment?

if only Vray was rman compliant …

oh that would be sooooo nice

Not that I wouldn't also like to see Mantra support multiple GI bounces, physically based rendering and be faster and all that, but what is it that you render all the time that require such features? Are you mostly doing virtual sets/interiors/visualizations?

What I am getting at is: would this be the only thing that'd drive you all the way to AIR or mental ray or VRay as your primary renderer? Or would it only cause to you want to render elsewhere for those projects that really benefit from it?

I'm just trying to find out the line where the “advanced” features of mantra (procedurals,point-clouds,archives,etc) don't satisfy an artists needs versus the “prettification” features (multiple bounce gi,spectral sky models,etc).

I have architectural backgroung and therefore my main day to day work is arch renderings.
Not that I'm not bored
I began hating max so I moved to XSI but still after smashing my keyboard I moved back to max just for vray. After some years of XSI i found it really weak compared to houdini so I started learning it heavily with some success.
But still when there is a serious deadline it all ends up in vray.

I know the effect can be achieved in any renderer etc etc, but this damn thing is really fast and really doesn't need testing to get good results, far better than anywhere else after hours spent.

Often the deadline is just too short for testing and you use what you are sure will help you out.
Whether it is an animation or a still, vray operated by quarter brainer will render it.
And sometimes, late at night I really eagerly turn into a quarter brainer and want a fast effect without having to remember everything and set up and think.
By the way, there have to be some reason why DD uses it


I usually write such posts when I'm pissed off and this is just the case.
XSI 5 ceased to cooperate at 3 at night so I grabbed all scene, put to olgood max, vray'd in half an hour and went to bed.
I really hate it. I'd like to use houdini, plant procedural trees and make otl's for everything but for now it looks … impractical for the kind of work I do.

I do believe that VRay gets you good results quickly, but I really doubt that it has much to do with the architecture of VRay, besides multi-bounce and so on. Instead I believe it's got a LOT to do with available shaders. Mantra is sorely lacking in a foundation of good base shaders and corresponding mid-level VOPs- and it shows when you need good results fast. A lot of feature work, however, does require a bit of thought and if you can spend the time, it pays off tremendously. For commercials and personal projects I can really believe that the results in VRay come a lot faster.

I assume SESI are aware of this lack of shader options. It would only help to build a few excellent shaders instead of keeping everything so general, as is their wont.

PS. I see 5+hr frames with VRay (only Commercial division, never Features) at video resolution (12+hrs at HD) rendering here today. If Features did that - and try and imagine the 2k render times! - we'd have our heads cut off by Production.

I absolutely agree, and don't want to criticize mantra or anything else.
But there always come a difference between a big studio like DD and a small few person beginning company with tight budget like me.
Your “production values” for GI are 256 rays, mine hardly ever reaches 64 .
Two different worlds with different needs.
In my free time I'm trying my best to learn rendering technology from several books and I'm beginning to see the power of renderman compliants.
There should be more open shaders prepared in VOPs but probably this is just a matter of time.
I hope for multibounce in mantra because it's a fast way to make images so much lighter and pleasant for the eye.
Not ranting anymore…
Thanks for your replies

P

I wonder what do you render if rendertimes of vray can be beaten by mantra?
I got to the point in which I ceased to understand using another renderer than vray. It's just so blazingly fast. The GI is superb, when I tried to render the same scene (architectural interior with lot of glass and brushed alu) in mantra, rendertime was plainly unacceptable.
I don't think it's only a matter of shaders.
Are rman renderers good only for feature production, and why? I got once it but now I'm thinking again.
I had the same feeling when I load the bolex tutorial ad hit render…slow as hell, what the heck, this is just a very standard setup?
Perhaps I should stop thinking about houdini as a rendering machine but only as geo generator, then load the geo sequence into max and save lightyears of setup? I looks impossible to use mantra in a small <10 person studio and survive, when you need a bit of photorealism.
I have some rman knowledge now and the thing looks so clean and easy and logical, flexible, open etc, but probably it's just for the very big guys and my interest in it is totally impractical.
What is “flexibility”? In some years of work in the business I found speed much more important then the possibility to sink in some rib,mi or anything.
Perhaps the 5 hours you are talking about are with some mblur,dof and everything, but how does mantra compare to this? From my experience what takes an hour in vray, takes 5 hours in MR and ten times more in mantra (not talking about dof and mblur).
I was very hot about houdini for last year, but I can't see how would I use it in no feature film 500 processor and 300 employees studio.
If someone could pour a bucket of cold water on my head and enlighten me I'd be very grateful.

yep its slower to texture/render in Houdini.
stick to max/vray for viz, its fast & simple; ideal.
however if you're doing convoluted fx work rendering is
less of an issue than a fast, stable, readily changable setup.

-cpb

Well, imho it all comes to what kind of imagery you create…For the kind of work that you do, it seems that VRay is the best choice.

VRay is, in my opinion, curently more of a specilaized renderer, where you have out-of-the-box dedicated setup to do fast, clean gi for simple geometry types. It has (obviously) very optimized shaders written with gi speed in mind, and nice tools for tone mapping. All this wrapped conveniently in a plugin library…slap shader, click render..done.

Mental Ray and REYES renderers are general purpose renderers that are designed with user control in mind. Such tools are supposed to be as flexible as the user requires and naturally come as black boxes, with nothing pre-set. However, you can render many primitive types (hair, volumes, etc) in an efficient and controllable way, with shaders you design.

As far as the general speed of computation, it has a lot (if not all) to do with shaders…For example, gi illumination in prman is very slow because shaders are not compiled, but interpreted. Interpreted shaders are great for first generation rays, but for raytracing are reaaaaly slow. In mental ray, until recently, with the default set of shaders, glossy reflections required you to oversample the object itself…ans so on, and so on…that list can continue…In other words, you really have to know what's going on under the surface of these renderers to get fast results…

As for mental ray, this is the software with the most modern architecture of the whole bunch out there…It has the best description format, extensive native API, GPU acceleration, sample merging and many, many more features…While I think that the library integration in xsi leaves a lot to be desired (control wise) saying that mr is old (old????) and clumsy, and slow is totally inaccurate…

Cheers

peliosis
I looks impossible to use mantra in a small <10 person studio and survive, when you need a bit of photorealism.
I was very hot about houdini for last year, but I can't see how would I use it in no feature film 500 processor and 300 employees studio.

My suggestion would be it's not so much an issue of 150 emplyees writing shaders for Mantra and the rest 150 setting up the render. It is more an issue of your experience and knowledge of mantra, shader-writing (which involves math), houdini in general and overall knowledge of CG related disciplines. So in comparison to this VRay+Max, which doesn't really require any of those, is often a more sutable choice for most CG artists. (doesn't it sound too highbrow, I hope?)

My guess would be that you don't need 300 employees. Rather a couple of Jason Iversons and Alan Kaplars. :wink: I.e. somebody who can set up the houdini/mantra pipeline, set up the render, create shaders and such.