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 - Juraj

Pages: 1 2 [3] 4 5 ... 288
I take what I wrote below back, I forgot it's Substance that introduced the 1/IOR mapping texture. I don't want to tinker with industry standard.


I still think that remapping of texture values from 0-255 to '1-user value' would have been the most reasonable way to go. The current way is just a big question mark, while remapping would have been self-explained if available right in the material's UI. Nothing works without adding extra nodes anymore the way it is right now.

I see a perfect way implement this :- D

- if the value is above 1, it is directly applied as IOR (e.g. value 1.5 = IOR 1.5)
- if the value is below 1, it is converted into 1/IOR (e.g. value 0.5 = 1/0.5 = 2)

Who on earth would use the second method (below 1) to create 1/IOR? Let's just change that one method and we'll have both approaches at same time. Full backwards compatibility, full IOR mode override and full easy flexibility. Everyone could be happy?

Like this:

If the value is above 1, it is applied directly as IOR override (exactly as right now).
If the value is below 1, it will directly map as fraction of existing material IOR. Example value 0.5 will halve the reflectance of material IOR 1.52 (so something like 1.333 I don't know the math from head).

Did I miss something, looks ok to me but I also have brutal migraine right now..

There are paints that bright, even commonly used for high-end wall-painting. Just one of many brands:

You can even buy some "experimental" paints for art that reach above 95 (they claim 99perc. but I call a bit bullshit...). I bought some for fun.

Documentation gives a generic advice and the guidelines keep changing on yearly basis and evolution of material models. The original PBR guidelines were meant to correct the overly bright and saturated colors used in pre-physical game engines. But it looks like we all might have erred too much into dark.

Lastly, rendering engines often suggest lowered albedo to ease up on GI calculations but I would ignore this advice. Image should be primarily plausible not done in way that makes the renderer feel good :- )

Any of you guys running the Silverstone IceGem 360? AIO for 3990x?

Interesting thing. Looks like almost no reviews although it has been out since January?

Hardware / Re: Ryzen 5950x build
« on: 2021-06-09, 21:15:41 »
Unfortunately, it's a lottery with three random factors you cannot influence:

- CPU memory controller binning.
- Motherboard memory support (this can be such picky thing like the distance of traces between CPU and Phase capacitators)
- Memory dies. This can change month from month depending on what the memory manufacturer sourced from dram manufacturer)

What you can do:

- Run the latest bios.
- Manually tweak memory settings. This is alchemy with lot of parameters. You can start by rising voltage little by little (1.36+), but this will increase heat which affects stability so you better have good case air-flow. It's no coincidence high-end kits often go to absurd levels like 1.5V.

Or you can just live with it. You will find it literally doesn't affect almost anything in actual real-world production. We're not talking high-FPS competitive gaming with CPU latency bound cases. But workstation usage. Zero drawbacks.

Hardware / Re: Laptop Suggestions
« on: 2021-06-09, 21:08:19 »

What I liked about the new Legion was the 5800H Ryzen 8 core cpu, 16:10 aspect ratio screen that is QHD and 144hz capable (great if you want to game a bit and for productivity), screen is also 500 nits, rtx 3000 series gpu and if you can go over the gamery look of the 5 pro of course. But you always have the option of the 7 Pro that looks plainer and office alike. Bear in mind that these laptops are a tad bit heavy though :).

100perc. what VUK wrote on all accounts.

The Lenovo Legions are indeed very good balance of features at great price point. Also, Lenovo is the only laptop maker with 16" 16:10 2560x1600px display right now afaik. Looks like they bought exclusivity for it ;- ). The rest only offer 15"/17" 16:9 displays with 2560px or 3840px.

But they are so damn ugly :- (. Peak gamer design unfortunately. Still, great laptops.

I personally don't mind the weight. My Razer Blade 17 pro is 2.6KG and it's just fine. That is actually perfect weight for thin mobile workstation (Dell XPS 17 has same weight with discreet nVidia GPU). Do you guys remember the 5KG + battleships just two generations ago :- ) ?

Hardware / Re: Monitor Recommendations
« on: 2021-06-09, 21:02:53 »
what do you think of LG 32UN650 as a budget option? Looks like it has a similar specs as the Benq PD3200U (or LG 32UD99-W which is not available) but better design and price point.
Or it's better to hold up for a while and wait for those Q2 premieres?

These are very hard to compare since PD3200U is ancient by now :- ) It's 2016 monitor with panel of same year from AU Optronics. LG 32UN650 is last year model only with panel from BOE.
Otherwise there is feature parity indeed.

As for budget option, yup, I would say good choice. Be prepared that with LG Monitors in general, you might count with warranty. One thing you pay with higher-tier monitors, is quality control.
That's not just dead pixels or better factory calibration, it's also to avoid excessive light bleed (not IPS glare, but the artifacts from corners of edge-lit displays).

As for Q2 premiere of displays, well... looks like things are delayed once more. Most brands only introduced their truly high-end offerings at 3K + Euro pricepoints and left the remainder to Q3/Q4..

Hardware / Re: 10 Gigabit Ethernet Performance
« on: 2021-06-09, 19:07:13 »
10G network will not increase any performance anywhere unless there was previously a bottleneck.

Since for reading/writing assets from 3dsMax/Photoshop/etc.. the main bottleneck are those softwares, you won't see much faster loading/saving.
For distributed rendering alone, 10G is massive benefit. Streaming 8K renderings using some medium fraction of megapixels every 60 seconds, even single node can reach the limits, let alone multiple.

Network is like memory, unless you ran against the wall previously, increasing it will not do anything, there is nothing magical about it.

Hardware / Re: Threadripper 3990x vs 3970x
« on: 2021-04-20, 13:34:15 »
These are not my results but my clients, who had forced me to make his WS on Dark Rock 4 Pro

A true client from hell :- )

Using "Dark Rock 4 Pro" as example of reference Air-Cooling skews thing negatively for air-cooling Threadripper, (DR4P) it's the only sTRX socket air-cooler which has horizontally running heatpipes across heatspreader, i.e they only cover the four middle dies. It makes sense, it was designed at 1st generation of chips, still, Noctua made correct design decision at same time..

Dark Rock 4 Pro was never a suitable cooler for 3990X, it does somewhat work for 3960X & 3970X because those have dies in middle of the chip. Which is also why those can be very well cooled by any generic Asetek AIO.

The difference proper TR4 air-cooler like Noctua UH-14s TR4 and Dark Rock 4 Pro on 3990X is simply massive.
Just a reminder, these two coolers had delta of 20 C already with past-generation 2990WX, the first chip with dies that were spread across the full chip.

I've put 3990X under OptimusPC water-block, state-of-art block that is even 2-3C better than the previous winner, Heatkiller IV. After that, it doesn't matter how many radiators it is connected to (in my case 2x420mm) since the issue is heat-transfer from chip, not dissipation from the loop. At room ambient of 25C, the effective boost clocks on long renders are effectively identical between this build and the one with Noctua UH-14s.

That obviously very quickly changes with PBO/Curve Optimizer/Static Voltages/etc.. (any other OC technique), where loop or even the funky tower from Ice-Giant pull quickly ahead, by a large delta. Past 280W (and it's not hard to take 3990X into 700+ W territory), Air is no longer viable.

99perc. of users on this forum are running their PCs stock. That's matter of life for most workstations (discounting XMP/DOCP profile).

Ultimately, my personal suggestion comes down to budget. If you can afford 3990X, get it, even if it scales very little in certain workloads, or particular working conditions.

3970X might indeed be more prudent, more universal option, it does maintain better boost clocks with simple math of diving its 280W power budget only between 4 Dies + 1 MC, a benefit which is partly negatively offset by worse cooling (4 centered dies are harder to cool than 8 spread-out dies).


Buy the one you won't regret spending budget for.
If you have sizeable render-farm, your choice is more partial towards 3970X. If you have only single-workstation that acts as your main rendering machine as well, absolutely go for 3990X.
Air cooling is absolutely viable, zero-issue choice for 3990X. But you have to choose either Noctua NH-U14s TR4, or Ice-Giant. You can ignore the existence of everything else.

Few graphs to show how much Dark Rock Pro 4 sucks, and why it should never be used as argument against air-cooling Threadripper.
Also article from KitGuru that explains the failed design of this cooler.

Hardware / Re: Threadripper 3990x vs 3970x
« on: 2021-04-14, 15:33:18 »
At least some sensible opinions by VUK :- ).

Yeah, 4k+

You don't need more passes with higher resolution, quite oppositely. Noise becomes less perceptible with higher resolution. Just like with real cameras. Noisy high-res image from DSLR still looks good. Noisy low-res shit from cellphone does not.

100 passes for 2K, 4K or 8K should give you the same noise level roughly, but 8K will look good sooner, it may already look good with only 50 passes.

got it.
with passes, what If I change resolution?  do I have to make the tests on the exact output resolution to have consistent results?

With larger resolution you can get away with larger noise, so quite often half the passes are necessary. But it's important to not have too little passes since +/- 100 passes are requires for clean AA.

What happens when you find out you only need 50 passes to have good noise on high-res image, lower the AA/GI ratio to get 100 passes in the same time. If you're rendering frequently very high-res images, you can just make your default ratio something like 8/2 instead of default 16/2 and forget about it. It's better for IR and IR with Optix Denoiser as well since you get faster feedback.

I think I have that as default Corona settings since I don't really want to change it per project like Vray :- ).

I have not used noise level for years. It's just too dependable on detail structure of image. Add detailed carpet into scene and the whole parameter becomes quite useless for evaluation of "render" noise.

hi Juraj,
I`m moving from vray and I`m using noise limit 2% on my setup.   Can you tell me what`s your workflow instead of using noise? sorry if it`s a noob question.

I am using passes, so I have to do some quick local test to see which amount is the right one. Even for animation, passes would equal consistent look without shimmering.

It roughly translates to 2 perc. as well. But it can be 4perc. if the scene has lot of micro-detail (like office type carpet,etc..). The renderer will mistake it for noise.

I have not used noise level for years. It's just too dependable on detail structure of image. Add detailed carpet into scene and the whole parameter becomes quite useless for evaluation of "render" noise.

The primary goal of Sheen is an easy simulation of the effect of subsurface scattering in the fibers, so any further customization options would highly likely deviate from sheen itself (and physical correctness).

Is that to say the sheen layer is meant as a sort of artistic embellishment, or is it offering a physically correct look by default?

Truth be told I don't know myself how to use it properly :- ).

One important aspect I found, is that it's good idea to "map" the intensity of the effect, instead of using pure numerical value. No fabric is smooth enough to create the sheen effect uniformly, the effect follows the structure so the structure needs to be converted into map that drives the effect intensity.

How that conversion should be done, is purely artistic interpretation. It's one of those "fake, but not fake" approaches to PBR that is necessary because real fabric is incredibly complex and mapping shader to a flat plane does pretty bad job most of the time, making "pure" PBR approach impossible. Just like Ambient Occlusion being mapped onto Albedo is necessary for most scanned materials (like rocks, grass, etc...)

I'm reluctant to ask, but how is the Physical Material being handled by external integration partners, e.g. Quixel, Allegorithmic, etc.? I suspect this question is outside the scope of this forum thread, but the direct and obvious implication the Physical Material has for external asset libraries is very extreme in terms of asset migration.

Bump :- ) I already asked this elsewhere and Maru gave elusive answer ("Maybe we are already talking" or something like that).

Would be really good if this was supported by Quixel Megascans exporter early. The current exporter sucks but they should have less excused why their convertion is weird when the new Physical material is industry-standard PBR.

Pages: 1 2 [3] 4 5 ... 288