C4D + Redshift - materials breaking at random

I have an issue in my local farm as follows:

  • When rendering a scene over deadline (C4D + Redshift) we are experiencing materials breaking on some nodes but not on others.

  • To be clear, not all the materials break, only one or two materials seem to break at a time (they render as white as if no material was assigned)

  • On top of that, sometimes the materials break and sometimes they don’t!

  • if i resubmit the job sometimes it renders without issues

  • The material that breaks can either be PBR with textures or something simple like glass

  • using C4D batch plugin

  • multiple workers on each machine to make use of each GPU for seperate tasks

  • locally rendered scene is fine on all machines

We double and triple check that all texture paths are globalised and on the NAS
We double and triple check that all simulations are baked

I have attached a deadline log for frame 150 (rendered fine) and frame 149 (rendered with a broken texture)

I’m no expert but it looks like the one GPU seems to be going out of core?
is this a redshift issue or something on the deadline side?

Deadline Logs.zip (7.9 KB)

I’m currently working with Maxon to try to fix this issue, as I’m experiencing it as well, but we’re having trouble reproducing it on Maxon’s end without using Deadline.

A couple questions for you:

  • Which C4D/RS versions are you getting it with? For me C4D 2024 seems to be ok, with the issues starting with C4D 2025.
  • Are you rendering in such a fashion where each GPU on a render node renders its own frame? (I,e - GPUs-Per-Task at 1 and Concurrent Tasks ~4 (something higher than one).

My current suspicion is that its somehow related to the machines rendering multiple frames simultaneously with C4D 2025+ (or something in RS 2025+)

Hey Luke

I left this problem alone quite a long time ago so I don’t even remember which version it was breaking on. I can’t even tell you if it was concurrent tasks or not. All I remember is we tried EVE|ryTHING and contacted maxon, deadline, redshit… etc. and eventually…

moved to the amazing thing that is RoyalRender. It costs about 50euros per node per year but their support is incredible and… almost zero issues that aren’t your fault once it gets going

hope that helps in some way!

Hi Alexis,

Thanks for the info!

  • Could you by chance share the ticket number that you/your team used on your correspondence with Maxon? That would help to provide another set of variables for them to look at & compare.
  • With Royal Render, are your machines rendering one-frame-per-machine, or do you have multiple GPUs in a machine rendering separate frames at the same time?

Thanks!
Luke