butterfly net render

[geo_n]

Options/Configure/Configure Platforms, that should bring you to the "LightWave Windows Setup..." tab and then right column at the top.
I asked Paul an option to exclude kray from this switch.
This switch is mainly made to strip the LWS from masterplugins like Spreadsheet Editor which can crash LWSN in certain cases.
Uncheck it, I will add this little quirk to the wiki as well.

That definitely did it . During the wizard I remember it slightly. Thanks a bunch . I think I did 5 installs of lw

[larry_g1s]

Big thanks Erwin. I'll be setting up my newly BNR with your info. Thanks again.

[geo_n]

Has anyone used the pro version to work with AE or fusion? There's very little documentation.

[erwin zwart]

http://www.kraytracing.com/wiki/Network ... _NetRender

Will add more BNR specific stuff later.

[geo_n]

Btw bnr does not check for rendered frames with kray renderer. It will overwrite rendered files.
And there's a nasty vista ui problem with bnr. If I click the bnr tab in taskbar it won't open. I have to minimize all tabs and then click bnr.

[erwin zwart]

BNR does check for rendered frames, when you use the LW image savers.

[geo_n]

BNR does check for rendered frames, when you use the LW image savers.

I see. When I use lw image saver the kray buffers don't get saved out. I put outputtolw tail cmd nothing happens. So I used kray instead and my renders keep getting overwritten but its no biggie.

How about the vista taskbar bug. Its so annoying

[geo_n]

Calling bnr experts.
I'm having some trouble with bnr. When a render is on going and a rendernode is disable and then you enable it, it says node was not present in scene creation so that particular node won't render the ongoing process and cause an error. Is that normal?

How do you guys do your workflow in lightwave with bnr?
When making a project are all objects and images pointing to a network drive from the start?
For example when working on a project
\\PC\Shared\BNR\content\Projectx\images
\\PC\Shared\BNR\content\Projectx\object

Problem is sometimes the path doesn't point there anymore after creating new objects so I keep repackaging the whole scene.

I'm thinking of just doing the project the usual way and then when I need to render to bnr I package the scene. Is this what people are doing? The files are stacking up.

[pok]

Hi,

I'm happy rendering with BNR-kray for a year. My workflow is:

1. working local (mac-pc) animator-modellers etc
2. moving finished scenes to mastercomputer of the farm (where BNR controller is)
3. opening lightwave (from the master or from another pc, never from mac) , setting content directory to the location on the server (always go by my network places/etc/etc... ALWAYS) check for missing stuff (images, objects)
4. do testrender, make GI bakes and set outputs of all outputs to the network location (again by my network places/ etc/etc...) I mean the layout save output AND the kray output)
5. adding scenes to BNR, check scenes in BNR and then happy rendering
6.compositors do copys of the renders to local disks, NEVER rendering, animating, compositing on the same disks...

My BNR renderfarm has on the master PC 2 mapped network drives where my folders of BNR are AND where my content folders are. Dont wanna have a stuppid compositor moving the BNR files!!! Never use spaces in disk names or locations, linux and mac have issues with that, just use underscores. No quicktimes or AVi's always frames (image sequences) no renderfarm can work with that.

1_renderprojects (network drive X) for content folders + renderoutput
2_renderdata (network drive Y) for BNR

(names are for example X and Z) I have on my cluster pc and my pc a little program called "network drive manager." It automount the mapped network drives to the system and rechecks that every 5 minuts. Like if you reboot the system it will remount the disks.

Cause BNR uses more digits in output names, check it and set it on the same like kray-layout, then BNR can check the renders. Or dont let BNR check it, but wait till Kray writes into your render output. Just to be sure.

Why not mac for BNR? Well the drive on a mac are mounted while a pc is mapped. You can let BNR translate that but it is very complicated!!! While a pc will use the correct locations. Dont forget that every OS that is lower than server edition has a limitation of 9 users. So max 10 computers in BNR. Nodes are limited from BNR license. I only use 1 node on a computer.

If needed you can send a scene from local to the BNR farm with the BNR node window, but havent tested it. I use my BNR farm on a pipeline basis, everything works in steps, rendering is just a part from it. You can use the BNR farm during other steps but you cant do the GI on a cluster, testrenders I try to do on 25% size or just raytrace only.

I hope this helps you.

fabian-POK

[geo_n]

Thanks for the detailed reply. I see I am thinking and doing the right workflow for bnr.
1. Do the work locally.
2. Package the scene to master pc for bnr to render.
3. Retarget the content directory for the scene. (somehow that automatic find content directory or force content directory, never works and only works if you're rendering in local pc. Missing maps and objects sometimes from other nodes if you dont set manually)
4. Do a test render to make sure all scene assets work.
5. Finally render it
A bit of juggling from local pc to master pc and traffic if everyone will be queing their work for rendering. But bnr is easy to setup.

I'm trying out smedge and netpal. Netpal seems to be faster and more responsive. Smedge is clearly superior of all. Kat from battlestargallactica only recommends Smedge for netrendering in lightwave. I saw his workflow and it works like backburner for max.
The steps for smedge
1. Do the work locally.
2. Add the scene for rendering and hit render. It will package itself and read from local hardisk direct to ram to other nodes.

There's no master controller so everyone can add or cancel scenes from their workstations so no traffic because there's no master pc and no juggling to add scene in master pc from local pc.


In backburner workflow
1. Do the work locally.
2. Hit netrender within max and scene is packaged with automatic retargetting and sent off to any pc running backburner.

That's it two steps. I think smedge copied this workflow. And the master controller is accessible from all local pc aswell and all artists can edit the render que priority,order,etc.


P.S. How about when one node is disabled and then you enable it while a render is going on. Does it add itself to the render process or says error node not present in scene creation. It seems strange that all nodes have to be on before a render is started.

[geo_n]

Hi fabio,
I asked at newtek what they do with bnr and they have different workflow.
http://www.newtek.com/forums/showthread ... post982233
Maybe its better only two steps(almost like backburner and smedge) but it requires all project files to be on a file server. That means each artist is working from that drive so in a crash I guess all artist go home and stop work until backup is restored


Anyway I think people should check out smedge too.

[fabian-pok]

Hi Guys,

I know that our pipeline has a lot of steps. But you can indeed work directly on the fileserver and yes crashes are fatal for the deadline. Thats why I dont risk it. The client and my boss will kill me if it happens.

But you can make a script that sends everything from a location to your fileserver. Its a bit like a backup script. It opens the lightwave scene file and replaces all paths with the paths for the network location and copies the files to that drive. I'm not the best script wizard, I always ask it to the IT guy. Just do copies not replace please... better to have some files locally.

Just be clean when it commes to files and paths, keep it simpel. Try to use a pipeline structure. BNR is for studio environment, I know about the backburner system, its like the system for Cinema4D and Fusion. But they only render those type of scenes.

But back to your problem with the nodes! First yes I can quit and restart the nodes. (I only render image sequences) So it seems very weard the problem you have. Its like the BNR rendering interface is locked. Please ask the butterfly guy. I had other complex problems and he helped me. It looks like a BNR script problem with the nodes or maybe an communication error.

Greetzzz,

fabian

[geo_n]

Thanks fabian. The nodes are really a wierd problem. Maybe your farm always have nodes on before starting a render that's why you don't get error.
Here we use artists workstatioon as nodes too. So sometimes I asked one artist to turn on node but I already started render process and scene is already rendering. When his node connects, poof, error comes out only for that node.

this is the error that comes out if you turn on a node while bnr is rendering.
02-06 [00:29:43] [PCHP01] Loading Scene:roomshadowwithmatt (port 5605)
02-06 [00:46:32] Scene name not found during create node

[fabian-pok]

Hi geo_n,

Sometimes I let the wrong scene render and then I just restart the nodes of BNR, kinda hardreboot, so I can stop the renderfarm. But they always can go back to enable modus and render. Even when I disable a node or a workstation to check, I always can bring it back to rendering.

Some questions, might be the problem:

1. do you use fixed IP adresses? Always have to.
2. port 5605 has to do with streams of data in a network (router, switches, etc...) maybe check firewalls, virusscanners...they must leave the BNR processes alone.
3.check the write permissions of the disc (maybe the nodes cant write while process started) I think it will be a problem with the network drive.
4.I had the same problem with my mac's, it was the network drive path that was wrong, but you dont use macs...
5.could be an conflict between LWSN for BNR, screamernet LWSN, WSN for fprime.

I fear that if this isnt the problem, then I cant help you. I would ask the BNR dude, he knows his software better than me. Try to reproduce the error on clean computer, if the node errors, screenshot it and mail it to him.

Greetzzz,

fabian