AWS Thinkbox Discussion Forums

[Maya] Inconsistence Render Images

Howdy folks,

We recently encountered an issue where frames rendered from the same job showed differences in two ways:

  1. Some frames did not have yeti hair rendered while others frame render normally.

  2. Some frames appeared lighter than the others, as shown in the image below:

render_differences

We are using Maya 2023.3 with Redshift 3.5.15 and Yeti 4.2.8.

The frames that lack the Yeti hair rendered or appear lighter do not follow any clear pattern in terms of when or why they occur. If anyone has experience in debugging or fixing this issue, your advice would be greatly appreciated.

Hello

Thanks for reaching out. Are you rendering on AWS? if yes the follow here: https://awsthinkbox.zendesk.com/hc/en-us/articles/9530119943447-Render-on-AWS-Portal-Render-output-is-missing-or-is-not-what-I-expected

If you are rendering on premise then I think it could be because of below:

  1. Yeti license: some machines ran out of license?
  2. Asset files not available on the render nodes which produce this result

We need to isolate render issue from Deadline.

  • To troubleshoot more submit the same job with MayaBatch disabled, there’s an option in the submitter “Use Batch Plugin” uncheck it. The job will now render via MayaCmd Plugin and will print the actual command line arguments to render the job out side of Deadline.
  • Now Look for Full Command in the job report (right click the job> View Job Reports> See the errored tasks> Ctrl+F to search for the word “Full Command”).
  • Copy this full command and run it on the same render node (by logging inside as the same user under which the render ran) which has failed the task via cmd/terminal.
  • If you get the same error it means you are able to reproduce the issue outside of Deadline and the scene/assets/render node need to be fixed.

More info is here: Troubleshooting — Deadline 10.3.0.9 documentation

Hello Zainali,

Thank you for your reply.

We are rendering on premise. To address your points.

  1. In our workflow, we have incorporated the use of Yeti cache, which means that rendering does not necessitate a license. Our IT team has also verified that there have been no license refusal requests recorded on our licensing server.

  2. We have checked the log file on deadline for the frame that did not successfully rendered and it seem like all the references to the scene was loaded.

However, while reading the logs files. I do noticed a pattern that. When the plugin was fully reloaded for the task, yeti will appears. The issue of missing yeti hair render tent to appear when the task used a scene from the previous task and continue render the next frame in the sequence.

Thank you so much for your guide on how to isolate render issue from Deadline.

I can find the Maya Full Command, to start the scene when the plugin first loaded. However, I’m not sure how can I mimic deadline behaviour to continue render the next queued task of the same job in the same machine.

Hello

Happy to help!!

I think I have seen this behaviour before but with another renderer (Redshift on cryptomatte’s output). I need some more information to look into this issue. There is an option in Maya to log script contents in the job report: Deadline Monitor> Tools> Configure Plugins> MayaBatch> Logging: Log Script Contents to Render Log> set it to True, reproduce the issue and share the archived job with me.

Share the archived job which has both the successful and unsuccessful tasks. if you do no have them all in one job, please share two jobs. If you are concerned about the internal information sent on forums send it to me directly or cut us a ticket. from here: Support | AWS Thinkbox

You cannot mimic (at least not easily) the behavior of MayaBatch, which is why I asked you to use MayaCmd. Just render one task see if you get the same results outside of Deadline or not. Most likely the task will render fine because the plugin is re-loaded on every task in Maya Cmd.

Privacy | Site terms | Cookie preferences