Octane 4 - C4D Crashes when Rendering to PV

[klarwin]

Hey there, just wanted to report a possible bug/problem on Octane 4 for the C4D plugin and to see if anyone else is having the same issue.

My specs:
i7
Windows 10
GTX 1080ti
SSD
32Gb RAM

When rendering a sequence to the Picture Viewer, C4D will randomly close. There's no crash dialog or windows dialog. The program will simple quietly close to the desktop. This doesn't happen at any time when using the Live Viewer window.

Additionally, if C4D does not crash during a render, there are instances when my monitor will lose signal completely and the only way to get signal again is to reboot. But windows still responds. I can p can control Spotify from my keyboard hotkeys, so it's not a full on crash. It seems like the GPU stops responding.

I'll update this post if I get any more insight into the problem. Let me know if anyone else is having this issue or something similar in Octane 4 for C4D!

Thanks all!

[aoktar]

Your PC may be out of RAM. Check the render stats from Console and RAM from task manager. Or stability issues for GPU/HW.

[klarwin]

I transferred some RAM over to my machine so now I have 64 GB DDR4. Will update if problem persists!

[klarwin]

Okay, so now my machine has 64GB of DDR4, and the problem is persisting, albeit in a new form.

Before, when I had 32 GB of RAM, I would monitor C4D's memory usage during a render via the Windows Task Manager. Every time it "quit unexpectedly" or turned my monitors off, it was when the RAM usage was at 99% and in a random place in the render range, so there was no correlation between the scene and the increase in RAM usage. Every time I restarted the render, the RAM would rise in usage and then crash in different spots.

So now I'm testing the render with 64GB of RAM and I just did an overnight render. When I turned my monitor back on, only one of my monitors was working and the resolution had dropped down to like 480p. The render was maybe 1/3 of the way done and Octane was stuck in the "beauty samples" phase, but the data at the bottom of the current frame was gone and the render timer was running anyway. Taking a look at my RAM, my machine was running at 99% again in a place where it hadn't crashed before. I had to force-quit C4D and restart to get my monitors back to normal.

I was thinking I might have done something stupid with my textures, but they don't seem to be all that cumbersome. I have maybe 40 4k Alpha mattes and 3 or 4 4k PNG sequences. 457MB in total. Everything else is just via Octane's node editor.

I'll continue to troubleshoot since this final render needs to start in a few days, but let me know if anyone has any thoughts!

Thanks for the help!

[Mako]

I've got the same problem here...
and almost the same configuration.

win 10
1080 TI
32Gb Ram

I'll try to update my GPU drivers and let you know.

[klarwin]

So, after spending two days observing this issue, I've developed a workaround:

What's happening is, each time a frame is done, my RAM usage goes up by anywhere from .1GB to .5GB. Typically, C4D or Octane (not sure which) will dump that cache every now and then. But in my instance, that "cache" never dumps and just goes on growing until there's a crash.

So how do we force Octane/C4D to dump that rising memory cost?

In my 1150 frame render, I've separated the scene into render stages. Since the crashing happened between 200-500 frames for this scene, I've split the project into 6 separate projects, each set to render a different 200-frame section of the sequence. Then I'll render these with the render queue instead of the PV. The runaway RAM usage issue still occurs, but by rendering it in chunks, it forces C4D/Octane to dump that extra memory every time a section completes, thus keeping the RAM usage in check.

It's janky and also won't let me render them all to the same sequence name, so It'll take some compositing to stitch it back up in Nuke/AE. But it works!

Tonight I'm about to start the final render, so I'll update this thread when it's done and let you all know if this janky workaround works on a huge scale!

Good luck!

[klarwin]

Update: Janky workaround totally works. If you're having a problem with Octane crashing while rendering multiple frames of big scenes, try the above ^^^

[aoktar]

Update: Janky workaround totally works. If you're having a problem with Octane crashing while rendering multiple frames of big scenes, try the above ^^^

Rather than this send me your scene for testing and let's find what's that! Also it maybe a fixed memory leaking.

[horizonmark]

klarwin, is this related to turning the AI denoiser on and off? Because I am having the same problem rendering 5k video. I get a random RAM spike across all our different machines towards the end of the render, which causes Cinema 4D to just unexpectedly close.

Unfortunately I won't be able to use your fix, as it happens even if I'm just rendering out one frame. I have a feeling the AI denoiser is causing the RAM spike and Cinema 4D to close.

[pfoley20]

Hey guys, not sure if this is still relevant, but I was facing the same problem using Octane V4, using 2x 2080ti's.

My fix was pretty simple. Remaking another folder to save the outputted frames to while also un-checking "adapt data rate" in the output render tab.