3dcenter.ru Corona thread

[Ludvik Koutny]

Hello,

when analyzing where visitors of our forum mostly come from, we have noticed that many of them come from a Russian site 3dcenter.ru. Thanks to this information, we have found a thread about Corona on the site. Unfortunately nor me nor Ondra know Russian language so we have read the thread using Google translator, which gave us only very crude and foggy translation.

From that crude translation, i understood that some of the Corona problems have been discussed, and i or Ondra may be able to provide answer for some of those problems. So, here are the answers for some of the problems i understood were discussed.

• You should always use single sided planes in windows for window glass, and apply glass material set to twosided mode. Real glass will slow down rendering a bit, even if you disable shadow casting for it.
• You should use HD cache as secondary GI for most of the interior scenes.
• For curtain material to pass sunlight through, you should not use opacity, but refraction level parameter, with refraction mode set to twosided, and optionally, refraction color to be mapped with opacity map. This is at the moment more efficient solution than mapping opacity. You don't necessarily need to map refraction level if you do not want to, for basic effect of light shining through the curtain, just set refraction mode of curtain material to twosided, and raise refraction level to pass more light through the curtain. Mapping the refraction color will occlude sun based on the map you put in.
  This information might not be true for people with nightly builds, as opacity slowness should be fixed there. Some tests of mapped refraction with twosided mode vs opacity would be appreciated. :)
• For scenes with strong depth of field, it is recommended to lower PT samples value to 2-8 to change ratio of GI vs AA sampling. Less PT samples mean more samples are devoted to AA, which is also responsible for clearing DoF effect. So you will get more samples in your AA and less in GI, which may lead to better balance and better rendertimes.
• There is often a noise in surfaces with sharp, but very weak/faint reflection. That is caused due to the insufficient adaptivity and therefore objects in reflection having insufficient sampling priority. This is a known issue and will be fixed in next alpha release. For those of you who have nightly builds, you can go to Devel/Debug settings and try to use Per-pixel M+C sampler, which should solve this problem at the slight expense of performance.

So i would like to ask people who are both Corona and 3dcenter.ru members to pass these information to guys from 3dcenter.ru, and inform them that they are always welcome on our forum too. :)

[ecximer]

I've already told them. :)

At the expense of opacity in tulle. Here I am personally satisfied with the result and speed. I'll try to do a refraction, but I do not see much sense.

[SHD]

finally)

[Ludvik Koutny]

I've already told them. :)

At the expense of opacity in tulle. Here I am personally satisfied with the result and speed. I'll try to do a refraction, but I do not see much sense.

Problem there was that in public A3 release (not nightly) there was a bug that would sometimes make opacity quite slow ;)

[ecximer]

All right. I understand you. ;)

[Han]

Thanks guys for your attention to us.
I have not worked for two years with 3d Max, now my main software is Cinema 4D, but the Crown has made me re-run the old software)))
In the future, if I can expect integration with Cinema 4D?
You will not be there for such competition - Vray integration worthless, Octane is not out yet, Arion and never will be, there is a mental ray but almost no one uses.
If I will be the integration of the first in line to buy it!

[Ondra]

sure, as soon as I get the money to pay for it, I'll hire someone for it.

[Han]

My first test on a commercial stage. It was originally made ​​by Vray . you can buy it here  zancanaro.com/gspot/?page_id=91
26 min   2500 х1461

1h

1h 30 min

 

[Chakib]

Great render Han of the G spot of the  bed

The displacement for bricks looks nice even if the closeup is a little bit smoother, but it's ok it's nice !

[Ludvik Koutny]

Another info...

When rendering mandelbulb animation:

• If you want it to not deform, then move camera instead of mandelbulb object. Its simply a container for volume in world space, so if you move it, mandelbulb will deform.
• When rendering animation, it's better to render first a single frame, to see how many passes it takes to get clean, and then disable time limit, enable pass limit, and set to the number of passes you got from single frame test. This will ensure noise will be consistent from frame to frame.
• When rendering very fine fractal detail, you might want to lower PT samples amount to capture fine details with more antialiasing.

[Han]

In my test on the wall displacement is very bad, which is a smeared.
How to fix it?

[Ondra]

lower texture filter value in Texture settings, and/or lower max projected triangle size in Corona settings

Also, try tesselating the walls a little

[Triar]

daily build(030313), 3dmax design 2013, windows 8x64

[SHD]

daily build(030313), 3dmax design 2013, windows 8x64

try re-install MSVC

[lacilaci]

Did you copy all the necessary stuff from corona into 3dsmax root directory?? Cause 3ds max is saying it's missing wxwidgets content(new corona custom framebuffer thing...)!

[Triar]

i7 2600k ?

[Ondra]

ask the guy who sent you the daily builds you should not have access to to give you the additional needed libraries (there are 4 of them to be copied into 3dsmax root folder)

[Han]

My second test.
The stage was converted from Cinema 4D - in max. I remembered why I hated the max before :)
It is not yet finished.

 

[alieneye]

3dcenter forum got new section especially for Corona

http://3dcenter.ru/forum/index.php?showforum=178

[Ondra]

sweet, although I would appreciate if there were no Corona builds re-hosted there.

[ecximer]

No, dayliBuilds there doesn't extend, there is only public Alpha3, and as public scripts.
Simply local users not all know English, however as well as I :), and for them all is duplicated that is here at a forum.

[alieneye]

I agreed with Keymaster.
Corona forum must be the origin for installation files, even it is an open Alpha builds.

[ecximer]

Ok. Fixed. :)

[Ondra]

I saw some discussion about portals, here is a comparison how the benchmark scene looks with/without them:
http://corona-renderer.com/stuff/bench-portal/

all differences other than noise are because of HD cache bias, not because of the portals.

[Ondra]

I see somebody asked why on most machines he has very long "transforming vertices" phase. You can tell him to either move the scene+textures to a faster disk, or merging/attaching some of its objects together as a temporary workaround until A5 is out.

[Ondra]

Looks like there is some discussion about comparing results from Corona to Vray. Google translator is failing hard, but from what I understood you are trying to compare pathtracing to pathtracing. You won't find significant differences there, as path tracing never blurs anything, and your scenes are too simple for any significant clamping to occur.

Nevertheless, there seems to be some difference in creases and dents, with Vray output being more "defined" (darker in shadows). This is not because of some interpolation of path tracing or faking in Corona, but probably because Vray by default completely ignores reflect color at secondary bounces (unless you turn reflect caustics on). This means, that Corona is giving correct results, and vray fakes the result by not computing a subset of light paths. If you turn reflect caustics on in Vray, you should get almost the same, correct image.

[ecximer]

Ondra, you can register on our site and carry discussion in language convenient to you, all of us will understand... :)

[Ondra]

No thanks, I'll leave this one to you, i wouldnt even be able to register ;)

[Ondra]

BTW, correctly implemented VCM, bidir, path tracing, light tracing, progressive photon mapping, etc. will ALWAYS give you the same converged result...

[Chakib]

BTW, correctly implemented VCM, bidir, path tracing, light tracing, progressive photon mapping, etc. will ALWAYS give you the same converged result...

Are you serious about Bidir ? it's good now ?  WOW i missed its speed !

[Ondra]

yes, they work. But they are still not fast (and never will be) for the typical scenes you guys do

[Polymax]

War on 3dcenter.ru - Corona VS Vray, Maxwell, Octane )))

[Ondra]

It is so bizzare... Why do you even compare render outputs? You will never be able to match the glossy materials correctly, and even so the differences are too small. If some renderer has a decent gallery of high-quality pictures, then one can assume, that it is able to produce high-quality results, and then it is just a matter of ease and speed with which you can create such renders. And this is mostly subjective...

[Polymax]

It is so bizzare... Why do you even compare render outputs? You will never be able to match the glossy materials correctly, and even so the differences are too small. If some renderer has a decent gallery of high-quality pictures, then one can assume, that it is able to produce high-quality results, and then it is just a matter of ease and speed with which you can create such renders. And this is mostly subjective...

Because it's fun: D and I, Ecximer and another some people defend The Corona...it's just a GAME

[racoonart]

Because it's fun: D and I, Ecximer and another some people defend The Corona...it's just a GAME

Why defend something if you have a bunch of those... :)

[Ondra]

yeah, I guess it is fun to mess with fanboys :D

[Polymax]

There are always limitations and difficulties in the performance of many imitation in all renderers. And like the "war" help to better understand the workings of a product, at least to me :).
Well, it's when I have time of course :)

[Polymax]

aaaand the results of the war :D
Time/Quality - perfect!
PT+PT, no post

[Ondra]

I have invented a new game - drink a shot every time google translate uses "Comrade" in the thread :D

[ecximer]

I poorly understood about what you, but shall be cheerful. :)

[Ondra]

Why doesn't "iDDQD" just send me some of his pictures from Corona and asks for the daily build activation?

[Polymax]

I tell him about it, thank you!

[Stan_But]

Ondra, what are you think about that to make test scene with all existing features for hunting for bugs? This scene may be shared with daily builds and would contained all newly added features too.

something like scene in this theme  http://3dcenter.ru/forum/index.php?showtopic=108803

I think that you can be to make it - as the most expert in  corona. Or someone from us could be make scene and you could be updated it

[Ondra]

in general it is a good idea, but a single scene cannot cover all possibilities - different GI algorithms, settings, lights, ... But is somebody would make it, it would be great. I can edit it, but I am too busy to make it myself

[Stan_But]

I know, but it would help to cut off many bugs in the first few minutes of testing

and many questions like 'how i can do...' could be to closed.

I'm still learning corona, but ok - i'll try to make scene

[DKulakov]

Why doesn't "iDDQD" just send me some of his pictures from Corona and asks for the daily build activation?

Hi, Keymaster!
At last I had time to write here!
Thanks again for a great product! Even in its alpha version, it is capable of much.
However, there is still much work. Can I help in becoming a beta tester?