Author Topic: Juraj's WIPS/Theory/Questions //Update with few PDFs from my recent presentation  (Read 191499 times)

2013-09-22, 13:08:45
Reply #15

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 6517
  • Marcin
    • View Profile
What are the HD Cache settings that might be useful here ? The documentation seems to be for Alpha4 and lot has changed, leaving me bit confused.

I am only changing HD Samples (4096) and K number (9999).

All I can tell you is that raising HD samples makes the precalculations slower but more precise and it pays off when rendering final shots. Setting lower thresholds may speed up rendering but will produce artefacts. Same for raising glossiness threshold. I too would like to see some delicious tech explanations for all features.

2013-09-22, 15:17:19
Reply #16

rafpug

  • Active Users
  • **
  • Posts: 721
    • View Profile
What are the HD Cache settings that might be useful here ? The documentation seems to be for Alpha4 and lot has changed, leaving me bit confused.

I am only changing HD Samples (4096) and K number (9999).

All I can tell you is that raising HD samples makes the precalculations slower but more precise and it pays off when rendering final shots. Setting lower thresholds may speed up rendering but will produce artefacts. Same for raising glossiness threshold. I too would like to see some delicious tech explanations for all features.

Good morning

right maru!

2013-09-22, 15:24:54
Reply #17

rafpug

  • Active Users
  • **
  • Posts: 721
    • View Profile
Thanks Sovcat :- ) Ask away if you need something !

Norwegian house

Here is very rough wip from current FAST project, although one that is pretty ambitious. It features both exteriors, interiors, vegetations. I will do extensive write-up on this project afterwards.





Anyway, this scene has too many glossy materials and I think it is first time the adaptivness might help a lot. Currently having problems updating to Daily version.

Here is some comparison how Maximal Sample Intensity (MIS) affects GI and highlights strength. As you see, around 400 is starts to look "correct", i.e, close to Unbiased. The default 20 is painfully low for realistic Interior scenes imho. I love the unbiased look, but it comes with heavy price of fireflies that never go away. So 400 is good "compromise".



Hello Jurai_Talcik

nice work congratulations.

The effect that this ambient light (ceiling, wall ... and scene) reminds me of the script E-Light


Greats and still good for this work of yours
Raf

2013-09-22, 20:13:40
Reply #18

Polymax

  • Active Users
  • **
  • Posts: 830
  • CG Generalist
    • View Profile
    • maxkagirov.com
I tested and, yes! You a right, 400 best solution!
Corona - the best rendering solution!

2013-09-23, 12:54:10
Reply #19

ondrike

  • Active Users
  • **
  • Posts: 95
    • View Profile
I did some tests and it seems it has no time impact if you use MSI 800, or MSI 20 only +/- 1-2 passes
Corona brought joy to sad world of 3Ds Max :)

2013-09-23, 13:15:54
Reply #20

Juraj Talcik

  • Active Users
  • **
  • Posts: 2897
  • Tinkering away
    • View Profile
    • studio website
It shouldn't have time impact, but specular materials seems to render much longer until they get clear. I actually struggle much with them inside interior to fully clear out, while rest of scene has been completed. I tried the daily build, but I don't know if it was sucess, it all just seemed to go slow, much slower with adaptivness, with no clear visual impact.
talcikdemovicova.com  Website and blog
be.net/jurajtalcik    My Behance portfolio

2013-09-23, 15:24:24
Reply #21

Correntes

  • Active Users
  • **
  • Posts: 74
    • View Profile
    • PRESSRENDER
With MSI at 400 I found that reflections are a pain to clean :S

Corona standard settings have one thing nice. Noise is usually even across the images. And i think its a big plus.

2013-09-23, 15:35:17
Reply #22

Juraj Talcik

  • Active Users
  • **
  • Posts: 2897
  • Tinkering away
    • View Profile
    • studio website
Possible work-around is to use rather low, 50 MSI, and instead, "fake" the materials. Put much more reflectivity, higher albedo (+20/30 RGB value) to materials that get affected by GI a lot (ceiling,etc.). But I don't like this :- ) I love how natural things look when I get close to unbiased look.

Same thing with Vray, unclamped, Subpixel-off settings just look better, but the render times are horrible, so it's either one of it (Sub-if you prefer night highlights, Clamp-if you prefer tonemapping freedom) or both off.

Damn quality/time ratio, just can't have it all. But we can hope adaptivness will solve this eventually.
talcikdemovicova.com  Website and blog
be.net/jurajtalcik    My Behance portfolio

2013-09-23, 15:40:01
Reply #23

Ondra

  • Administrator
  • Active Users
  • *****
  • Posts: 8460
  • Turning coffee to features since 2009
    • View Profile
meh, you cannot get away with MSI=0 unless you use bidir/vcm ;). I.e. adaptivity will not help in all cases
Rendering is magic.
Private scene uploader | How to get minidumps for crashed/frozen 3ds Max | Sorry for short replies, brief responses = more time to develop Corona ;)

2013-09-24, 12:18:10
Reply #24

michaltimko

  • Active Users
  • **
  • Posts: 249
  • Coronaut (c)
    • View Profile
    • Michal Timko | Photorealistic architectural visualizations
I found that noise is more uniform while using bucket mode.

Coronaut!(c)2011

www.michaltimko.net - Coming soon
Supporting Corona in commercial projects since pre-alpha

2013-09-24, 17:26:39
Reply #25

Juraj Talcik

  • Active Users
  • **
  • Posts: 2897
  • Tinkering away
    • View Profile
    • studio website
I found that noise is more uniform while using bucket mode.

The bucket mode is fully functional and ready just as progressive ? Hmm, but I love progressive mode a lot, so I prolly won't it even experiment.

I found similar thing to be with PT+PT, which takes of course, longer to clear out noise, but it doesn't differentiate so much with specular materials.
talcikdemovicova.com  Website and blog
be.net/jurajtalcik    My Behance portfolio

2013-09-24, 18:32:17
Reply #26

DraKeXXI

  • Users
  • *
  • Posts: 4
    • View Profile
    • Alex Ruano
Hi Juraj!

I've always  liked your renders, but....how do you get such nice whites? How do you do the white balance? In camera or in postproduction?
Can you share some tips? There's no way I can get white walls even half as good as yours.

Thanks

2013-09-24, 18:38:00
Reply #27

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 6517
  • Marcin
    • View Profile
The bucket mode is fully functional and ready just as progressive ? Hmm, but I love progressive mode a lot, so I prolly won't it even experiment.

I would say that bucket mode is almost ready (needs some tweaks like visible frames for buckets) but for now, adaptivity is just a "tech preview", it is useful only in some situations like heavy DOF or scenes with lots of flat, uniform areas.

2013-09-24, 21:36:17
Reply #28

Juraj Talcik

  • Active Users
  • **
  • Posts: 2897
  • Tinkering away
    • View Profile
    • studio website
Hi Juraj!

I've always  liked your renders, but....how do you get such nice whites? How do you do the white balance? In camera or in postproduction?
Can you share some tips? There's no way I can get white walls even half as good as yours.

Thanks

Hi Drake,

here is from what I wrote (chaotic) in other thread of mine:

about those white wall question. Btw I had to use sometimes a lot of tricks to get it with Vray. And here is the reason: Maximal sample intensity.
The default 20 is not enough imho for interior scenes. It produces much different "gradient" between highlights and shadows. Naturaly, it looks best at 0=fully unbiased. Then,
you get excellent light diffusion, midtones are huge ! Local contrast is best. But at great cost of sampling and fireflies. Maybe adaptivness will solve this ? Not sure, this was done with
Alpha 5 and I used value 50-80 I think, depending on picture. The big difference can mostly be seen in how bright gets the ceiling. At low value it is dark, while the walls already get burned.
At Unbiased settings, the ceiling is much more closer to walls in contrast. The whole settings can get much brighter before burning. And it can't be compensated with exposure.
Imho it only looks photorealistically when it is set at 0. There might be few reasons (GI caustics ? Some secondary reflections,etc.. didn't study too much) Heh, tough choice tough :- ).
Thankfully I could afford 20+ hours per each picture (but it was nicely clear after 6-8 mostly).

Other stuff about white balance:

I usually use somehow color neutral lightning, my HDRis are desaturated (not black and white !), tweaked, sun is 6500 Kelvin (neutral). Corona doesn't have useful color balance at the moment, as it only allows temperature (but many HDRis have color cast), so I tweak it in Photoshop using LEVELS adjastument layer set on COLOR mode.
talcikdemovicova.com  Website and blog
be.net/jurajtalcik    My Behance portfolio

2013-09-25, 13:05:12
Reply #29

Juraj Talcik

  • Active Users
  • **
  • Posts: 2897
  • Tinkering away
    • View Profile
    • studio website
Building up the exteriors.

talcikdemovicova.com  Website and blog
be.net/jurajtalcik    My Behance portfolio