Hi All,
I am completely new to C4D and Arnold, have been taking this time to try to learn both of them and have been really excited about the prospects of this application for my directing work. I have been working off my mac book pro, which isn't ideal but I don't have money for a new computer at the moment. I have recently created a 160 frame animation that I believed to be fairly simple, but as I am going to export it it is giving me around 2 hours per frame render time. Around 600+ hours in total, which isn't realistically sustainable.
I am new to sampling and the program in general and am not sure of the pitfalls or ticked boxes I need to be aware of that are going to send my render times off the deep end. I have worked my samples up incrementally to try to get something manageable but the result at 20 minutes a frame is low res and still really noisy. Anyone have any advice? Specs Below:
Macbook Pro:
2.9 Ghz 6-Core Intel Core i9
32GB 2400 MHz DDR4
Radeon Pro Vega 20 4GB
Intel UHD Graphics 630 1536 MB
Cinema 4D R20.059
Arnold 3.0.2 (Arnold 6.0.2.0)
I thought about using render farms like ZYNC as a last option, but they don't support Arnold 3 in C4D. Only up to 2.6, is 2.6 available for mac download if that is my only option?
Any advice or resources you could give me would be most appreciated!
Thank you.
6 cores (12 logical) isn't a lot, but as you said, there's nothing you can do about that
Hard to say without seeing an Arnold log (Info level) and maybe a screenshot of a render/wireframe to understand what is being rendered.
Surely the diffuse, specular, sss samples are what is killing you.
What about adaptive sampling?
For older versions, just go to the main download page and scroll down to the bottom, where you'll find a link to the download archive.
If you're okay with sending me your scene, I can see how long it would take to render... I have a 64-core/128-thread 3990X that will be idle for a few days... can send you the output files via Dropbox
It would be cool to see the logs from both machines, to get a comparison.
But I think we all expect 128 threads to kick 12 threads ass 😉
Maybe by as much as 10x faster
I debated going all GPU, but opted to go CPU for one more round as I still haven't had the greatest success with Arnold GPU, so I figure I'll let that cook a little longer and have access to all the Arnold CPU features like C4D noises that I use frequently, etc. It certainly is a beast of a chip, though... helps out in many other areas like particle cache's that are multi threaded... AMD is kickin' butt (:
Hi Stephen!
Sorry, took me a second to figure out log files etc. I created a log for one frame of the export, will that work? Obviously the full animation would take a few days haha.
Also I have not tried adaptive sampling, I heard that would make it slower? But could have been misinformed.
Thank you for your help!
Brandon
Now that is the machine I realistically need to be working on! Wish I could upgrade to even 16 cores. I may take you up on that as a short term solution. That would obviously make quick work of this, thank you for the offer!
As Stephen suggested, adaptive sampling, if properly used, could help. Here's what you do:
Does noice work well with animation? or is/was it intended for stills? I know Optix is pretty good for stills, but really bad for animation
Noice works fantastically for animation. You use it after all your frames have been rendered so that noice can make sure that the denoising is temporally coherent.
Optix is intended for IPR, where you want denoising to happen in ms and it doesn't matter if one frame is slightly different from another frame. Noice can afford to be slower and so can do a much better job than optix for both stills and animation.
That is good to know... I've yet to try noice, but I will certainly do some tests with it after what you've said here