Distributed Rendering - Dark Buckets

[womble]

I know this topic has come up before but we never seem to get a definative answer (if one exists).

When using DR with our render farm all of the slave machines are returning dark buckets.

Now this could be because I'm using LWF on my machine, but the max installs on the slave machines aren't. But I'm pretty sure that when other guys in the office are using DR (they don't use gamma correction as far as I know) that they are getting the dark buckets as well.

There are no other plugins being used, and I tested it over 4 machines (my dual-quad core and 3 quad core slaves) so there shouldn't be any issue with too many machines trying to access the same bitmap files. I'm pretty sure that all CPU's are intel (there shouldn't be any AMD's in the farm).

I'm using brute force and irradiance map (no precalculation).

Any idea's guys? DR is definatly a lot easier when working on test renders than sending to backburner and it would be good to get it working.

Cheers.

[Sascha Selent]

Test with one slave

Hi.
Did you test to bring it to work with only one slave before you try to assign multiple machines?
Do the slaves find the license-server?
Does any firewall block anything?

[3LP]

and it does work correctly with backburner ?

[Sascha Selent]

I solved my problems with DR Spawner with a really crazy workaround

Maybe this is interesting for all of you?

I solved my problems with DR Spawner with a really crazy workaround

Sascha

[cbecker]

I know this topic has come up before but we never seem to get a definative answer (if one exists).

When using DR with our render farm all of the slave machines are returning dark buckets.

Now this could be because I'm using LWF on my machine, but the max installs on the slave machines aren't. But I'm pretty sure that when other guys in the office are using DR (they don't use gamma correction as far as I know) that they are getting the dark buckets as well.

There are no other plugins being used, and I tested it over 4 machines (my dual-quad core and 3 quad core slaves) so there shouldn't be any issue with too many machines trying to access the same bitmap files. I'm pretty sure that all CPU's are intel (there shouldn't be any AMD's in the farm).

I'm using brute force and irradiance map (no precalculation).

Any idea's guys? DR is definatly a lot easier when working on test renders than sending to backburner and it would be good to get it working.

Cheers.

Has anyone found a reason why DR is giving different values back per server bucket? I've attached a picture of what I am talking about.

[dlparisi]

Looks like it can't find an IES file. Is everything UNC paths?

[cbecker]

Looks like it can't find an IES file. Is everything UNC paths?

It appears you were right, well sorta, I didn't have UNC paths, but I did have Mapped network drive for all my textures, however not for the IES light. I forgot to add it.

So if anyone else is reading this thread ... Make sure all your render nodes can find your IES lights as well as your maps (DOH)

thanks guys,
Chris

[Sascha Selent]

Has anyone found a reason why DR is giving different values back per server bucket? I've attached a picture of what I am talking about.

Maybe the good old gamma-problem?

To me it looks like a gamma-setting-problem. Does your scene use gamma-settings enabled? If so - even if you check "Load Enabled State with MAX Files" in preferences - your slaves will not use gamma-settings! The "Load Enabled State with MAX Files" setting has to be done locally in every servers max-ini-file.

If not gamma then some UNC-problem. Mapped drives do not work with DR-rendering via service.

Sascha

EDIT: I better have to read before I post - okay, was an UNC-problem...