Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - Nejc Kilar

Pages: [1] 2 3 ... 70
You are welcome! I mean I'm just thinking out loud here so don't take everything I say as fact. Truth is, I'm a little new to Zen as the 3970x is my first Zen workstation since... Well, since I was a kid sporting an AMD Athlon. Not that I didn't like Zen or Zen+, on the contrary really but at that point in time I already owned a server grade WS  that was and still is speedy :)

850W for a 3970x or a 3990x with a 3060 should be good enough in my opinion. If you aren't doing a lot of GPGPU stuff then going with a lower tier GPU makes perfect sense imho because it won't help you render faster with Corona at all and the Nvidia AI denoiser will work really fast on the 3060 too. Save some cash and buy an extra SSD or something :)

You can think of PBO as sort of an automatic OC yeah. AFAIK there aren't any prerequisites but lower temps and a thing you can't control anyway - power consumption of the chip. To enable it you basically turn it on in the BIOS and that's it. On my ASUS STRIX XE GAMING motherboard it actually came set to AUTO which I think just means ON - as evident by the CPU boosting up to 90c when it ain't explicitly disabled. Not quite sure how it works honestly because I was under the impression that the CPU's wattage should skyrocket with PBO being on but from what I can tell that didn't really happen. The boost clocks are a nice addition although I can't quite see them making a huge impact either. Honestly, somebody who knows more about PBO should probably speak up :)

As for tiny spaces and multiple nodes... Well, the more rendering power you cram into a single package the less heat output you'll have in that room plus it'll be one less machine to mess around with - so just like you said, less worrying about the different updates and all that stuff. If you can splurge for the 3990x and it makes business sense to do so I think that's the way I'd go. It'll last you longer because it is faster for rendering (not so much for other stuff though) and you'll have less machines to tend to.


Looking at those specs I think generally speaking you're good. The only thing that I would reconsider in your place is that 850W PSU for the 3970x and a 3090. AFAIK you are going to be pushing close to its limits if you'll be running the 3970x AND the 3090 full speed so I'd personally feel quite a bit more comfortable with 1000W+ there. Especially if you'll enable PBO.

Memory wise, I dunno, what you've listed seems great. Running a couple of different rendering benchmarks I can't really notice a difference between 3200mhz and 3600mhz (on a 3970x) and the timings on those modules you've selected seems tight enough. To be 99% sure whether that memory is going to work you can check the motherboards QVL list (on their website) but from what I'm seeing on the forums, most non QVL stuff typically works anyway.

Cooling wise, the 3970x does get toasty with the U14s but it is still within spec. If you can splurge extra I'd consider the Thermal Syphon IceGiant but since you're within spec with the U14s I wouldn't prioritize it. The 3990x appears to be easier to cool so the U14s supposedly does an even better job there.,

As for storage, well, 1TB seems a bit on the low side but I reckon you'll be shuffling projects to the HDD once you've completed them. It really depends on how big your projects get too but I'm thinking you've probably taken all that into the account already. Besides, want more space? You can always plug in another SSD later :)

All of the above is of course just my humble opinion :)

Glad you've found my first post helpful, SSS seems like the way to go about it indeed!

And if you are using Corona 7 then creating that wet "coat" on your character should be as simple as toggling the "clearcoat" layer in the new Physical MTL to ON. Remember, the "clearcoat" property basically simulates a "coat" of clear (refractive) and also reflective matter on your material. So if that material is "water" and it's really not too thick then that's how I'd approach it - with the "clearcoat".


I'm a little bit unsure what exactly you are trying to replicate but anything body related my first approach would be to create a Physical MTL, turn on some SSS / Volume and then put a clearcoat on top of it for that "wet" look. Obviously you'd need to play around with finding the right values and applying the right dirt maps but that's how I'd approach the body. Looks very much doable to me unless I'm missing something :)

As soon as you start "Running out of RAM" what happens is that instead of reading scene data from RAM (which is super fast) Corona starts reading it from your SSD / HDD and both are orders of magnitude slower than your RAM. So any time your are running out of RAM it is expected that the rendering process will get slower - a lot slower if you have a lot of data that doesn't fit into your RAM.

So basically, your computer stores things it needs for fast processing in your RAM. If your RAM is full then all the applications need to search for that data on your SSD / HDD which as we've established is a lot slower. Think of it like feeding the CPU with data, if you can't feed it with enough data fast enough it won't help that it's the UBER MILLION CORES CPU as it'll be bottlenecked by having to fetch that data from hardware that's a lot slower. It literally won't be able to calculate at it's fullest potential because the data to calculate can't be delivered to it fast enough.

Your best remedy? Upgrade to more RAM if possible or try to optimize your scenes so that they fit into your RAM (smaller textures / HDRIs, less subdivided geometry etc...).

[C4D] General Discussion / Re: Cloud/dust explosions
« on: 2021-09-07, 09:36:35 »
Many thanks. So the workflow would be to create the simulation in say X-particles, export that as a VDB and then render in corona?

That's how I'd do it yeah :)

[C4D] General Discussion / Re: Cloud/dust explosions
« on: 2021-09-06, 12:43:10 »
Well creating VDB sims requires some sort of VFXy knowledge - so X-Particles is one option, Houdini f.e would be another. Even something realtime could probably get you there like JangaFX's EmberGen. Generally speaking you'd need to do particle / volume simulations in there which is typically focused around controlling your simulations via different forces and such to make them as you want them.

Depending on the complexity of the result that you'd need you could also try modeling similar shapes using polygonal modeling techniques and then maybe sculpt in some of the detail (using Cinema 4D's sculpting). Afterwards you take that mesh and apply a Volume material and try do to the magic in there... But I suspect if you want to get those nice "nebula" like details in the dust cloud itself that you'll have to do some simulating.

[C4D] General Discussion / Re: Cloud/dust explosions
« on: 2021-09-06, 09:16:29 »
I'd give this a go with OpenVDBs because with Corona 7 you can now render VDB simulations using the Corona Volume Grid.

As for the little rocks on the side, while we don't support Particle rendering right now (we are looking at it for v8 but can't promise anything at the moment) you could potentially get away with simulating the particles and then converting them to Geometry. Not the absolute best solution but a solution none the less :)

[Max] General Discussion / Re: Tonemapping - Plz Halp
« on: 2021-08-30, 09:14:09 »
Hi Martin

Thanks for the LUTs.

I can not find the option to load a LUT as a log...



You should find that option in the Render Settings under the Camera menu :)

[C4D] Resolved Bugs / Re: Light material not working.
« on: 2021-08-26, 16:43:36 »
Hey there!

I'm curious, does the entire light mesh disappear or is it more of a case where you see don't see the bitmap and instead just the bright white light?

Thanks for the feedback Cjwidd, are you referring to any specific examples? I might be able to post node graph here.

Yes, I am referring to the example with the violin.

Hey cjwidd,

First of all let me just say that we are super excited to hear you've enjoyed the webinar and that you've found it useful. That was after all our main purpose :)

I'm attaching the node graph screenshot for the violin scene we've used in the webinar. The full credit goes to GeorgeK for coming up with this one but the reason I'm sharing the graph with you is because we've changed the materials every so slightly for the webinar version.

What changed? Well basically GeorgeK tweaked a couple of other material parameters as well and for the webinar we really wanted to emphasize that even if you leave all the other non-clearcoat related parameters the same you'll still get a completely different look & feel of the wood material as long as you introduce the clearcoat and some absorption.

So the node graph attached is the one from the webinar. Hope you'll find it useful and if you have any questions just let us know.

Oh and again, kudos to GeorgeK for coming up with the scene :)

[Max] Bug Reporting / Re: Render elements
« on: 2021-08-25, 09:49:44 »
Thank you for reporting potential bugs to us but multiple topics across multiple forum sections might not be the best way to do it. We have a whole part of the forum dedicated to bug reporting and you may find it here -> .

That said are you encountering this issue in 3ds Max or Cinema 4D (Win / Mac)?

I'm thinking its somehow related to the colors you see displayed on the top of each NME shader / material. Not sure I know what grey means lol.

Or you know we're just talking to a bot :)

Sorry maybe I was a little unclear with that sentence. I'm basically referring to single, dual or quad channel memory set ups. If you've got 1 or 3 sticks in there it might be affecting the performance of the 2990wx. Again, I'm not exactly sure by just how much although I think I remember the 2990wx played "optimally" in a quad channel configuration.

The size of memory plays a role only when you don't have enough of it. If your scene (and all other programs you're running at the same time) fits in the available memory, there should be absolutely no difference between various RAM sizes.

There are other factors which come in play, because I guess the two builds you mention are not exactly identical. For example the RAM in the 96 GB build might be running on lower frequency or be configured in a way that not all the CPU memory channels are used. And that's talking just about memory, there can be many other factors affecting computer performance, not just hardware, but also any currently running software as well.

it turns out that the memory frequency plays a strong role for corona render ?

it turns out that the memory frequency plays a strong role for corona render ?

With Zen and Zen+ CPUs it afaik does. Zen 2 and Zen 3 not so much.

Imho what you are likely seeing here is that certain configurations have faster memory / proper multi-channel configs and possibly some sort of OC going on while others don't. I didn't own the 2990wx myself but if I recall correctly that particular CPU was very picky with not just the memory speed but also multi-channel setups. Do note that you can't really go higher than a certain RAM speed anyway because of the infinity fabric and how that works.

Pages: [1] 2 3 ... 70