PDA

View Full Version : How is modo's network renderer?


Navstar
04-15-2009, 03:55 PM
I'd like to hear from modo users about their experiences with the network renderer. From their website, it sounds really easy to set up.

Is GI and AO smooth? Do simulations need to be baked? Are still frames split up and rendered across the network -- or just whole frames?

Any gotchas or things to look out for when doing a network render?

MikeS369
04-18-2009, 10:43 PM
It's easy to use. Host and clients are linked through a common folder. It's a bucket renderer so buckets are assigned to the different machines. You don't have to bake animations but in certain situations it is advantageous. I don't do much animation so someone else would have more info than me. I'm not sure what you mean by "is GI and AO smooth?" Could you be a little more specific.

Navstar
04-18-2009, 11:46 PM
Before r11, Cinema4D had really noisy Global Illumination and ambient occlusion, making it unsuitable for animation (unless you really cranked up the samples which slowed renders to a crawl)

So can any machine be a render host? Does the host need to render, or can all the slaves render and the host just collects the buckets?

wilgory
04-19-2009, 07:49 AM
Navstar,
Modo's network rendering is so simple, you basically just have to tell it where to look for a job (select a shared folder). Since it is a bucket renderer it will work on parts of an image, not just whole frames, so theoretically if you had 16 machines in slave mode and your image required 16 buckets to render, each machine could tackle a bucket, and pick up where other machines had left off.

Any machine can be a render host or you could switch any machine to enter render slave mode, but I think the current 302 release does not allow the machine that is not in slave mode to send render requests over the network without rendering some buckets itself. That would be a great feature request to allow the host machine render with just the slave computers. I'm sure a lot of people would like to continue to work when something has been added to the render que.

MikeS369
04-22-2009, 10:04 PM
Navstar,
Modo's network rendering is so simple, you basically just have to tell it where to look for a job (select a shared folder). Since it is a bucket renderer it will work on parts of an image, not just whole frames, so theoretically if you had 16 machines in slave mode and your image required 16 buckets to render, each machine could tackle a bucket, and pick up where other machines had left off.

Any machine can be a render host or you could switch any machine to enter render slave mode, but I think the current 302 release does not allow the machine that is not in slave mode to send render requests over the network without rendering some buckets itself. That would be a great feature request to allow the host machine render with just the slave computers. I'm sure a lot of people would like to continue to work when something has been added to the render que.

The only way to do it now is to disconnect one of the comps from the network. I think Lux wants folks to buy another license instead. I understand Lux's point but when you are a freelancer every penny and every second counts.

Navstar
04-22-2009, 11:04 PM
Interesting. With Cinema4D, I can install the render server on any Mac -- even a low end Mac Mini or even a G4 iMac. Since it does not render, it only sends data to slaves and collects frames. So the render server it doesn't need much CPU power.

MikeS369
04-23-2009, 02:23 AM
Interesting. With Cinema4D, I can install the render server on any Mac -- even a low end Mac Mini or even a G4 iMac. Since it does not render, it only sends data to slaves and collects frames. So the render server it doesn't need much CPU power.

What is nice about C4D is that you can render and work at the same time. No need for an additional license.

As for the GI in modo the renderer is like all other renderers. In certain situations noise problems arise. An example is blurry reflections. At times you have to crank up the samples which slows things down. It is not a renderer where the default settings will yield good results every time.

As for speed, it might be a little faster than AR v3. You will get people here claiming it is a lot faster than AR but I haven't found that to be true. You even have people claiming modo's renderer is comprable to Vray. Which is not true. Maybe some day but not right now.

modo's render engine is a good renderer and it is going to get better. Lux seems to focus a lot more development on it than Maxon does with AR. But then again C4D has a lot more functionality and features than modo.

pelos
04-23-2009, 02:37 PM
we use modo for modeling and we send everything to XSI for animation and render with Mental ray, since modo doesn't support character animation yet.

but sometime we use it when we render objects moving for some commercials.

i try to set it up as the document say but i am missing something,
any particular hint?

rufferto
04-23-2009, 02:44 PM
I use both modo and C4d ( AR and final Render as render engines). Here are a couple points to keep in mind with MODO.


C4d gives you

Modo has buckets, which are a big advantage compared to AR, when rendering large images.
ONe disadvantage from Modos buckets however is that the Irradiance is calculated only one computer. So all the other machines are waiting until the fastest one finishes the prepass.
(in comparaison, Final Render also has buckets, but the prepass there is also shared between the slaves, which is much more efficient).

Modo's prepass system is mainly a disadvantage when calculating smaller images, such as movie frames, when the prepass takes a big percentage of the overall render time.
It is not so significant for larger images, where the slaves kick in.

Modo does not have a net render services for single images. You can however get FRAME WARDEN for modo, which does a good job for this (just google it). Frame Warden has also from my experience top notch support from Daniel, the dev.


On a related note, the animation GI in modo is very good in many situatiions.
I prefer the way it is setup in comparaison with C4d: in modo, the animated GI solution is updated on the fly, after each frame, so you quickly react if you see something wrong.
In c4d, you need to wait for the full prepass before seeing how the first frame actually looks like.

I find the Multi pass system in C4d more robust, espcially when outputing PSD files.
Everything is named and placed as it should be. Also C4d allows for AE project output, which is very usefull if you use that route for compositing.
(there is a plugin written for AE support from modo, I have never used it so I cannot comment on it).

Modo's multi pass system however is much more fluid to setup within modo, since you get live feedback of any changes with the render preview (AO, passes etc...).

Object level buffers are easier to setup in C4d in my opinion.
However modo give you material level buffers, which is something lacking in C4d.

hope this is helpful.
cheers

ebbmeister
04-23-2009, 06:36 PM
Currently Modo has no method of interpolating irradiance samples between frames for smmoth, non-flickering animations in the same way C4D or Vray can. Right now, the only way is increase sampling in IR maps or the QMC method at the expense of render time.

Cheers
Phil

ngrava
05-10-2009, 05:10 PM
Currently Modo has no method of interpolating irradiance samples between frames for smmoth, non-flickering animations in the same way C4D or Vray can. Right now, the only way is increase sampling in IR maps or the QMC method at the expense of render time.

Cheers
Phil

Does that really work though? I got weird results with frame interpolation in Vray with Max. A kind of stuttering effect that was different from the GI flickering. Maybe I'm doing it wrong though.

CGTalk Moderation
05-10-2009, 05:10 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.