Hi,

I have a simple doubt what the difference between mantra -H host1, host2…

and mantra -n 2 rendering commands…

If I have 4 machines connected with LAN which command I have to use for network rendering….

Thanks,

mantra -H machinename,machinename2,machinename3 (no spaces)

-n will still work, but is the number of threads per machine. Note that in H12 this doesn't work as expected however (SESI is aware). Basically, I have an 8 core machine, and if I use -H on the local machine plus another 12 core machine, only 8 cores get used on the remote machine. You'd expect “use max cpus” to use 8 cores on my local machine and 12 cores on the remote machine for 20 total.

FYI, -H across multiple machines brings diminishing returns depending on how heavy your scene is and how fast your network is. Never use it for sequences of frames as that will be extremely inefficient. It's useful for interactive lighting sessions however.

Caveat: This is all Linux based info, YMMV on Windows/OSX as I've done little to no testing on those platforms.

Cheers,

Peter B

I've done some work on XP a few years back and -H works well too. The only thing you need to bare in mind is to use UNC paths instead. Otherwise mantra will not resolve textures properly.


kuba

I used mantra -n 8 command and rendered using hrender, I didn't see any textures applied. It seems textures are not shared between them..

The ifd bakes in the paths for all textures. You can use a text editor such as vim (use cream as it is vim with a standard gui) to view the ifd file and inspect the texture paths in amongst all the binary data.

If these paths are not valid on the remote hosts, then your images won't show up. Pretty simple.

The advice to use unc paths only works if those remote render target machines have the root machine mounted with the same name. With unc paths, you need to mount the other windows machines with the same machine name and then use the path syntax:

\\ComputerName\SharedFolder\Resource\…

Thanks Jeff its working..

I am wondering is there any way to generate deep map shadow before render and access these shadows from disk directly while rendering smoke…

Thanks,
kare.

Yes it makes a tremendous amount of sense to create your shadow maps exclusively without generating an image.

With shadow map generation enabled for your light(s), create a dedicated ROP Output Driver for shadow maps (just a normal ROP set to Micropolygon rendering).
In the Properties > Render tab, disable the “Create Image From Viewing Camera” toggle render property.
When you hit render, only the deep shadow map is generated.

how can I apply the rendered shadow map to smoke render. I tried with shadow map vop inside shader but it seems not working…

Am I doing correct procedure or is there any other one?

Thanks in advance..

I believe you can bring in the shadowmap from the light source, hope i'm understanding the situation correctly. :?

and you also want to untick “auto generate shadowmap” (unsure whether it's deactivated when you read from disk) just incase it continues to generate shadowmap upon render.

eitht.

Got it have to toggle off the Auto generate Shadow Map in light parameters