It is possible to keep the textures used by DR ?

[doark]

Hello, it's possible I have distributed rendering mode stop sending textures to other machines and instead I copy my sourceimages for other slaves to machines not waste time transfer files while the DR are sending the job to the computer?

Another thing, I am using vray_adv_22501_maya2013_snow_leopard with Transfer missing assets on but this message keeps popping up: warning: Environment variable VRAY_ASSETS_CACHE_PATH is not specified.

Thanks for attention

[snivlem]

You can cache the automatic transfer of files so it only does it once, why not use that?

[doark]

Hi snivlem! Thanks for replying! I do not know if the automatic transfer cache is working! Why could not I render a sequence of frames.

When I render a single frame using the VFB the function transfer files to other machines on the network works! But I do not know how to test the cache because I do not get it to work DR function using Maya Batch Render and not leaving the button render animation only in batch mode off to render the sequence with maya opened using VFB.

I'm using Maya 2013 SP2 in MAC OS 10.7.5.

[Svetlozar Draganov]

Hello doark

When Transfer Missing Assets and Use Cached Assets are turned on - Vray will transfer all missing assets to your slave machine and it will use them when you are rendering animation.
A simple test could be performed by sending a scene to a distributed rendering while assets are not accessible for the Slave machine - if the slave manage to render all assets then everything works correctly. You can run this test firstly on a single frame - and if it gives positive result try it with animation.

According to VRAY_ASSETS_CACHE_PATH:

Vray needs to know in which directory all transfered assets should be copied. Please create a new environment variable with:
name=VRAY_ASSETS_CACHE_PATH
value=PATH_TO_ASSETS_DIRECTORY

If you still have troubles please download the last Nightly Build and run the same test - if this also doesn't help please contact us.

[doark]

Thank you for explanation svetlozar_draganov!

I did some testing here using DR, I managed to make the option Transfer Missing Assets and Assets Use Cached work.

I set up a simple animated scene and called vrayslave on all other machines. Then I put a texture to send a render. I used GI in IM + BF and everything worked wonderfully

But I have a problem! When I go back to my production scene can not do the other machines work using the DR! It is just my local machine calculating. I was thinking it could be my GI configuration that does not allow the DR is working. My configuration is as follows: In the Primary bounce I use BF with 24 Subdiv and 3 in Depth. Secondary bounce I use LC with the option from file! I calculated before using 4000 Subdivs (my resolution is HD720p), with the option Use camera path, Use light cache for glossy rays and Use retrace threshold. (Another doubt, when I increase the retrace threshold what happens? I better increase the value to 5.0 or decrease to 0.5 ?) This configuration does not work with the inclusion of other machines at render time with DR.

[doark]

An idea!

Ok, I was thinking! If the problem is with the LC using the option: from file. Perhaps the problem is that the other machines are failing to receive the map of LC. But now I do not know what to do with the new environment variable? I create it within the Maya.env or is this a thing of the terminal MAC OS?

Thank you for your help and sorry for not understanding the part of creating the environment variable.

[t.petrov]

Putting the env var in Maya.env won't work, because it has to be available for the vrayslaves.

How do you start the vrayslave processes?

If you start the servers manually from the terminal, then you can put the env variable in the ~/.bashrc file.
You can add something like this to the end of the file:

export VRAY_ASSETS_CACHE_PATH=/full/path/here

If you've registered them as services another solution should be used, I can check it up if this is the case for you.

[doark]

Another problem and solution

First I would like to apologize for the delay in answering the topic. But I was on a project with a very small schedule.

Hello t.petrov,

I start going until the vrayslave ChaosGroup in Applications folder then I give two clicks and the terminal opens with vrayslave. I do this on each computer. Now that I have more time to take tests, I saw both things I was doing wrong. Sorry for any confusion. The theory that the LC could be causing problems with DR is false. It was just a coincidence. The problem of the DR does not work on all other machines just happen when I started and another MAC distinct jobs. When I'm only sending jobs to other slaves everything works fine.

Now I'm using a different method. (I learned here: http://polygonspixelsandpaint.tumblr.com/post/701899285) I export a single vrscene of an animation. After I open the terminal and type:

echo "PATH=\$PATH:/Applications/Autodesk/maya2013/vray/bin" >> ~/.bash_profile

shortly after I insert:

vray -sceneFile="/Users/doark/Desktop/3D/Maya/sh05_DR.vrscene" -distributed=2 -renderhost="192.168.12.6" -transferAssets=1

And everything works fine

But now I have a problem with a production scene. The scene is composed of a character with maya hair animated by 200 frames. The problem with export is because he got 13.47GB Have much time to work on this only Prepass:


This is the print of the slave computer:


In this case it is best to use VRAY_ASSETS_CACHE_PATH ?
Thank you for your help.

[Svetlozar Draganov]

13.5GB sounds too much for a character with a hair and probably this is the reason for the slowdown.
Have you tried to render this scene on a single machine ? Does it produce the same result ?
Have you tried to render it without light-cache , or to use precalculated Light-Cache ?