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 ... 86
1
[C4D] General Discussion / Re: IPR- real zoom?
« on: Today at 01:15:06 »
If you go under "Cameras" ->  "Navigation" -> "Camera 2D mode" you pretty much get that behavior with the difference being you need to "zoom in" from within the viewport itself (also, the camera menu is located just above your viewport).
It does the exact thing you're describing so hopefully this helps :)

Nonetheless having it in the VFB directly would make accessing and using it even easier so maybe I'd suggest posting this as a feature request (https://chaoscorona.ideas.aha.io/ideas).

Oh and the cool thing is that when you switch back to the "normal" camera mode the 2D repositioning you've made resets and so you don't ever lose the original position of the camera. It just hops back from the 2D one to the OG one.

2
[C4D] I need help! / Re: Video textures
« on: 2024-10-04, 11:36:29 »
Yes indeed, that would be my approach as well.

Create a Light MTL, have it be driven by the animation, and then plug it into a Corona Light which is set to operate in MTL mode (screenshot attached).

Alternatively you can just apply the Light MTL onto a plane and do it that way.

For the Gobo-ish effect you'll need to go with really high directionality and if you'd like a crisp clean projection I think some of our users suggested using standard C4D lights.

3
[C4D] I need help! / Re: Video textures
« on: 2024-09-30, 16:56:17 »
For the video part, yep, the technique in that OG thread still applies.

- Get that video loaded into a native C4D bitmap loader (just drag it into the node editor, Corona Bitmap doesn't support animations but the native bitmap loader does)
- Plug it into either Light MTL or the Self Illumination slot in your Physical MTL
- Render it all out.

That oughta do it :) If you encounter any roadblocks, we are here.

4
As another tip, if everything checks out fine with your hardware, is to make sure all the opacity maps on the leaves have "clip" toggled to on.

5
Hardware / Re: In-house Render Farm
« on: 2024-09-19, 14:47:59 »
Probably the most often asked question for any studio embarking on the animation journey, I feel ya.

My mini render farm is made up of server gear because two reasons - it is the most powerful stuff you can get (if you go 2 socket style) + if you buy used server CPUs it is also relatively affordable. There are however potentially a ton of gotchas with them and unless you're into hardware I would suggest either hiring someone to do it for you or opt for pre-builts or Threadrippers.

When it comes to server gear there are tiny gotchas all over the places potentially - such as in my case where I've bought 2x Epyc 7B13s that run on a single motherboard but because it is a server motherboard the VRMs were overheating pretty much constantly. That is because server gear is designed for server like airflows which is often directed using shrouds and so putting that kind of gear into a normal case (even an airflow focused one like the Fractal Torrent)... Might not be optimal.

That said I managed to get it working by simply wedging a fan between the two CPU coolers and have it point downwards towards the VRMs. That solved the issue in its entirety.

Point is, its a gotcha.

The other gotcha is running regular Windows on these types of configurations - one of my LAN ports doesn't seem to work properly and I'm assuming its the drivers. All the motherboards I've seen for dual Epycs are "rated" for Windows Server and the like.

On top of that you get no fan curves or anything so its often a choice between "tornado " mode and "kinda ish silent but do keep the AC on" mode.

Oh and as a further gotcha, I'm running Noctua fans on the thing and the motherboard constantly thinks they are dying because of the "low" RPM they run at. 1500 RPM is nothing in server world (stuff gets to ~7000 RPM over there) but for us normal folks its already rather loud. So the only way I found to resolve that in my situation is to run them full speed all the time or have a separate fan hub installed (the built in one in Torrent or Meshify or Define works fine, just don't plug the fan readout to the motherboard). Otherwise the fans continuously ramped up to full speed and go back to 0. I think the motherboard was doing CPR to them or something :)

Anyhow, hopefully that illustrates some of the "difficulties" working with server gear.

On the flip side however, a used 96 core 9654 will cost you around 2500€ (around 5000€ if you buy new). You get two of those and you've got some solid power on your hands - and you'll still be below 10k for the whole build whereas a single 96 core 7995wx will cost you 11k.

Building a Threadripper seems easier to me because you've got all the consumer gear available to you and the motherboards tend to have wider compatibility with stuff. As long as you chose compatible components and you don't break them when putting the system together... You should be good. It is a little bit more advanced than consumer stuff in terms of setting it up but imho if you can build and setup a Ryzen / i9 system, you'll be able to do it.

Then there are the prebuilts, probably the most reliable option because yes, you'll pay more (sometimes a lot more) but if something is wrong you typically can call the company that built them and they'll fix it all up for you. Unless you run into one of "those" companies.

I opted for used Epycs and I plan on continue buying them. I have a relatively small office space and with 5 systems I'm already kind of running out of physical space so for me getting the most out of a system (in terms of speed) is paramount.
That might or might not work for you. Do note I'm also super nerdy about hardware so running into gotchas with desktop Epyc setups is ... A bit frustrating but also rewarding once I'm able to solve things.

Hope this wall of text is at least a little bit helpful.

Oh and also, something to consider is that currently some of the higher core count 2S (two CPUs in one motherboard) systems are not quite working as expected in Corona. It is being looked at but just so you know.

6
[C4D] I need help! / Re: OSl flakes in C4D corona??
« on: 2024-09-18, 13:11:02 »
Sounds like a feature request to me, one that I'd also +1. Fortunately we have a new feature request system set up (https://chaoscorona.ideas.aha.io/ideas) and so I'd kindly direct you to open up an idea thread over there so we can keep better track of things (and see what you the users would feel like is a good addition to Corona).

7
[C4D] I need help! / Re: OSl flakes in C4D corona??
« on: 2024-09-12, 17:44:49 »
Howdy! Yeah, from my understanding OSL isn't supported in C4D. May I ask by which method did you approach it? OSL shader something something?

8
Roger that, makes perfect sense. I think our support team is going to take this over. Thank you for reporting this!

9
Hmm, just out of curiosity, is the glitch still there if you lower the bump strength?

Thank you for reporting this!

10
hm hm are you using Interactive Rendering to render your images? If so then you can go under Edit -> Preferences -> Corona -> untoggle "Adjust render resolution to VFB".

That said you probably shouldn't be using Interactive Rendering to render out your final images because IR is a bit more optimized for interactivity and such so the actual fully rendered images might take a bit longer to render out.

It is entirely possible I've misunderstood you here too and so if that is the case please let me know how I got ya wrong :)

11
So this might not be news to some of you following this thread but I was today years old when I realized EXPO / DOCP makes the memory controller & co consume more power. That in itself might seem completely trivial but that extra consumption is added into the max TDP envelope of the CPU as a whole which means more power for the memory controller and IO die, less power for your cores.

Now it isn't a huge difference mind you but on my end I'm noticing about 5% faster rendering when I drop DOCP. The all core boost is at 3.4ghz (up from 3.3ghz - both boosts without SMT turned on mind you) and I'm also seeing about half the cores reaching peak rated frequency on my 5995wx which is 4.5ghz.

Interesting!

12
Howdy!

Those are some really awesome specs either way you flip it but at the same time to me they are two completely different tiers of system.

The 5975WX build is based on the WRX80 workstation platform meaning you get ALL the bells and whistles in terms of expansion capabilities (and other "pro features" like IPMI access that I don't think HEDT TRX50 has) that Zen 3 offers.
That said it is a Zen 3 system so two generations old by now.

On the other hand you have the 7970x which is an HEDT part which means it is a bit more cut down compared to WRX80 (mainly if you need more expansion slots although Zen 4 HEDT has PCI-E gen 5 now compared to 4 on the WRX80) but is noticeably faster than the 5975WX build for when it comes to rendering.

All in all I think you might have to ask yourself what do you need more? Is it rendering speed / compute power or the actual platform itself (aka more expansion slots and IPMI access and the like)?

Personally, if this is mainly to be used for rendering and you don't need a crap ton of IT compliance I'd go with the 7970x. It still has enough expansion slots to fit 2x GPUs (and some of them are PCI-E gen 5) plus if you got slim (think Quadro GPUs) expansion cards you can still have separate addons in there. The ASUS TRX50 Sage for example has 5 usable PCI-E slots which imho is plenty enough. You should double check it but fitting in a single GPU and a couple of NVMEs expanion cards (for up to like 8 drives) should be doable imho. And even that might be overkill depending on your workload.

GPU wise I think a 4080 should be more than fine if you're primarily using the system to do Corona / CPU rendering. Unless you're video editing and doing a crap ton of AI upscaling at the same. I haven't seen many benchmarks but viewport wise 16GB should be more than plenty plus the actual difference in viewport FPS I doubt it'll be noticeable. C4D runs into all kinds of bottlenecks before GPU becomes the biggest problem I think. Do consider VRAM limitations if you're running simulation with the new GPU solvers though.

Memory wise you'll have to go RDIMM (ECC) memory if you go with the 7970x build. Double check this please but I'm pretty certain you can do UDIMM (non ECC) memory on the WRX80. I keep forgetting what I'm running (also on WRX80). It will cost you a little bit more going TRX50 in this case but do remember the 7970x is quite a bit faster for rendering than the 5975wx.

I don't think I'd over complicate it with memory speeds. If you can easily digest the 5600mhz price tag I'd say go with it. If you can't I do think 5200mhz will work just fine as it shouldn't be a noticeable change unless you're high refresh rate gaming or have memory speed intensive workloads (rendering is not).

Storage you've got pretty nicely picked out although I'd maybe suggest looking into Kioxia and WD too. WD has some really sweet deals lately on some of their stuff.

Motherboards I'd say go with whatever floats your boat in terms of the layout and brands you wanna deal with. Personally I'm on the WRX80 ASUS Sage atm and it's a really nice experience. Mileage does vary as everyone has different experiences with different brands and vendors.

Cooling wise, the 7970x won't be "easy" from what I'm reading. It's basically the 3970x on steroids which I do own and can attest it does require solid airflow with a beefy cooler. A lot of people seem to recommend quality AIO / water cooling for this part although between you and me, I'd probably try my fortune with a Noctua U14s with two fans and an airflow focused case (Fractal Design Torrent is great for that imho) - I'm really scared of leaks myself.

So yep, those are my 5 cents. Hope it is helpful :) Oh and don't forget, its all just my own opinion :)

13
Hmm hmm, based on the scores you are getting I'd say those seem like expected results? If I cross reference comparisons between the 13900k and 14900k on the V-Ray benchmark and websites like CPU Monkey then the two should generally score / perform ~about the same.

The Raptor Like chips (14th gen) seemingly did not impress the press either as they got more or less universally panned for not really being an improvement over 13th gen. It is basically a refresh line of CPUs and some of the reviewers called it "the 13th gen but with a higher number" type release.

I'd say some minor differences are to be expected based on BIOS settings, motherboards, drivers, OS funkiness and things like that but overall, looks to me like they should score about the same for the most part.

14
Corona Academy / Re: News
« on: 2024-08-12, 10:31:17 »
Howdy! The 12 release and summer vacays slowed us down a bit more than maybe we initially anticipated but we are back at it and while we can't say exactly when the next video is coming, it is already in the works.

Super excited to hear you've chosen the Academy path of upping your familiarity with Corona! Hope you're enjoying it!

15
Thanks for your input guys! Greatly appreciated.
While the 7970X seems like a powerful option, considering the price difference compared to the 3995WX (at least in Croatia), I'm leaning towards the 7960X for now.
This should still provide good performance for 3ds Max and Corona while staying within my budget.
The only thing that worries me a bit is the limited availability of Corona benchmark results for the 7960X. Are people avoiding it for some reason?

Is there anything else I should consider when making this decision, or any specific recommendations for RAM or other components to pair with the 7960X for my workstation build?

Well well, pozdravljen sosed :)

The 7960x (24 core) is a bit harder to tame than the 7970x (32 core) and especially the 7980x (64 core). Its mainly got to do with the fact that you've got less cores that are consuming more power and so it basically comes down to the fact that you're cooling 350W in a rather smaller area because there's less cores (that are hotter) compared to the higher core count designs.

People seem to recommend an AIO for this chip, I'd probably still stick a Noctua on it and I also wonder what Juraj would do :)

As to why there aren't that many results I do not know, maybe its the above or maybe peeps just opt for the higher core counts. You can also try checking the V-Ray benchmark, it isn't exactly 1:1 with the Corona benchmark but it can still give you a general idea of the performance imho.

And also, since you're in Croatia, if you're comfortable putting the parts together yourself I'd maybe suggest looking into Caseking or Computeruniverse as the place of purchase. They typically have nice deals although do keep in mind you'll have to ship your parts to Germany / Austria in case of an RMA. I fortunately haven't had one yet but besides that they are treating me real nicely so far at those stores (no affiliation, just sharing my personal experience) :)

Pages: [1] 2 3 ... 86