Manuka Weta Digitals new renderer

[Kadri]

http://www.fxguide.com/featured/manuka-weta-digitals-new-renderer/

[inkydigit]

Looks pretty neat! 

[Oshyan]

It sounds really cool. I just wonder with these "boutique" renderers what the real impact and interest for the rest of us is. I mean, unless they start selling it (and I guess they *have* moved tools out to commercial sales through orgs before, right? whether through staff leaving to run software companies, or whatever, e.g. Massive, others), if it remains an internal thing then... cool but when do I get to play?

In the broader scope of things though it does interest me in terms of their motivation. Arnold has recently become much more available and claims many of the same benefits, but Manaku sounds even more advanced. When Arnold came out people were claiming it was the end of Renderman, or at least of Vray, since for the latter Arnold directly attacks many areas that have always been ruled by Vray and yet were challenging too. Renderman certainly responded with some major developments, I don't know but I imagine ChaosGroup is planning big things for Vray 4, and possibly sooner than the gap between Vray 2 and 3, due to the pressure from Arnold. But in any case there was at least this sense the last couple years that unique in-house tools forming a key part of the production pipeline might be decreasing in popularity. At least that was my sense, hehe. So now Manuku kind of turns that on its head because you can't get much more critical than a renderer, so clearly Weta is committing to this thing in a big way. It's a little surprising.


Again I wonder what it means for "the rest of us", but ultimately it's always nice to see innovation, so wherever it happens it's cool. It just reminds me a little of Flowline's amazing, in-house-only fluid simulations from years back, doing things nobody else seemed able to touch at the time, and you just wanted to play with it right then! But no, you had to hire them for a job to get access, that was that. So, same here. Cool renderer, we don't get to play. Oh well.

- Oshyan

[Kadri]

Yes in these kind of articles it is the same thing that bothers me despite that we can not buy-use all the software that is already all around.
Still nice to read how this or that studio makes his special-3D effects.
For example "Flowline" or whatever it is called,the fluid simulation program is still an in house tool i think ( http://www.scanlinevfx.com/index.html ).
But those kind of software is good in the end that it makes the others do better-more things.
Bullet kind of destruction was so far away for ordinary users in the past. Now nearly all major programs have it.

Until i need hundreds of apes rendered in Ultra HD and in stereoscopic format i don't mind it much
Not that we couldn't use that power like from a hardware point as every update of our computers ends after 1-2 year in the "I need a faster computer" way.

[goldfarb]

rendering is moving very much in the direction of physically based approaches - so ultimately everyone will be using the same math
all the rest of it is down to your pipeline, Weta clearly believes that they have needs that are not being met by Prman, v-ray etc - at least not met at the cost of having 1000's of render nodes ($$$)
Animal Logic wrote a renderer that they used on the Lego Movie - for reasons that don't make much sense...but they did it anyway...
I'd be wary of thinking that just because Weta (or whoever) writes their own tools that those tools are by definition better than those commercially available - until very recently Pixar used an animation system that people I know who have used it described as 'shit'...and few would ever criticize their animation.

[Oshyan]

Thanks Michael, very good to have an industry perspective weighing in.

- Oshyan

[TheBadger]

Agree with Oshyan's last post. This stuff makes my head spin. Every time a new renderer or 3d soft comes out I start out thinking that it must be better, or do something no one else has (why else would they put out a new product). But no. Usually I get excited for nothing. So its good to hear from someone who is in the trenches.

Still, even knowing that, I keep looking.

[Tangled-Universe]

As far as I know Animal Logic first didn't need a new render for The Lego Movie, but that guy (can't remember the name) from their dev department by himself created a very promising "data-crunching" renderer at home which made it much more efficient to render all those Lego bricks and their shading properties which were otherwise too demanding for their existing pipeline. In the end they decided to give it a go and it looked promising. After that that renderer has been improved/optimized greatly to fit the job.
A funny approach indeed.

This is also what Manuka and Pantaray are about.

Essentially Renderman is still the main rendertool in Weta, but they have developed tools which are specifically designed to deal with massive, truly massive amounts of complex data.
All this stuff is done before it is actually being passed onto the renderer in chewable portions.

Arnold, Vray and many other commercial renderers do use their own accelaration methods within their renderers to improve efficiency, but they are pale in comparison with Manuka performance.

Since Manuka and Pantaray especially have been discussed before, it shows that it is meant to stay an in-house tool.
Probably for good reasons, mainly being competition I can imagine.
Pantaray was a GPU based spherical harmonics geometry sorting monster, allowing Renderman to raytrace the scenes much much faster, because Pantaray kind of pre-filtered the scene's geometry to calculate what would make sense to raytrace and what not.

That's basically all we know about it, but there's no technical information or white-paper out there describing the tech in depth.
Neither will happen to Manuka.

I actually have the feeling the competition is becoming even tougher, thus less and less tech info will be shared as a consequence.
Vray improving, Arnold into play, Weta and ILM's tools for Renderman, but also Pixar/Disney themselves on Renderman are very very obscure about their latest tech called "RIS".

So like Oshyan said, it's nice stuff to read and drool over, but otherwise...

[jo]

That was an interesting article. As a side note, manuka is a NZ native tree. It's not a terribly impressive one, relatively short and scrubby, but it looks nice when it's flowering en masse. Honey from it has good medicinal properties. I have a lot of manuka round my house, in fact I am using it to make an interesting looking balustrade for the stairs. So, some random facts for you!

Regards,

Jo

[Oshyan]

Oh wow, I forgot about the name of this renderer, and then just bought some Manuka Honey recently! Random connection.

- Oshyan

[PabloMack]

One thing was a bit misleading. "Hardware Rendering" to me means they built an embedded computer with PowerPCs or some such embedded processors expressly as a dedicated renderer. It seems to be all they were doing is GPU programming for "Gazebo" with CUDA or OpenCL (they never said). So even in that sense, all rendering is a combination of hardware and software but the article gave no indication that they designed their own dedicated hardware for Gazebo.