Problem of cloud noises on surfaces

[flipside]

Problem of cloud noises on surfaces

Here some tests. I disabled secondary GI to see the ir map effect better.


Irmap: 25subdivs, 15samples, clr=0.2



Irmap: 75subdivs, 15samples, clr=0.2
The effect of the subdivs is clearly visible. It even resulted in LOWER rendertimes, probably because with only 25 subdivs, the GI is so rough that the passes have a hard time of cleaning it up.



Irmap: 75subdivs, 15samples, clr=0.3
Effect of clr value also very visible. The quality of each GI sample is good, but there are not enough samples taken to create the detailed shadows. Also look at the rendertime... In a textured scene, the imperfections seen here will not be that apparent. Also, add secondary GI to it and the splotches will not be that apparent either. Then add the fact that the client doesn't know anything about irmaps, so they don't even see if the shadows are detailed anough or not.



QMC GI
Compare this with the second image, it comes pretty close.



Irmap: 75subdivs, 15samples, clr=0.2 + Lightcache

[Micha_cg]

Problem of cloud noises on surfaces

Wouter, thanks for all the infos.

If there is no difference between 66 or 300 subdivs, something is wrong...

I suppose so, but I don't know what. It's impossible to get a smooth GI with min/max -3/0. The test is a typical interior scene, so what could not be good? The GI bounces are not to simple or? And if this example is simple, why I get no smooth result?

I have enabled "show samples" now and have not got visible samples inside the interior, only outside. Is it normal?

Light portal: Sorry, but the IM pass stay buggy. Lightportal and IM cause wrong lighting in every case. Could you test it with 3DSMax? Maybe it's a core bug. I have seen it at a 3DSMax machine some weeks befor. In QMC+LC mode it works perfect.

Multipass: I tested it on several scenes now, but never have seen a enhancement, only little changes. So, I have tested a scene with some ugly little noisy areas - the multipass give me no better image at this critical parts. Do you have compared it by yourself. Sorry, if I'm a pain in your neg, but I want to understand this IM method now. And I want to be sure, that anything works right.

EDIT: Cross post: I see now your examples. I will write later, now I must pick up my daughter from the kindergarden.

[Micha_cg]

Problem of cloud noises on surfaces

Wouter, could you send me your Rhino scene please? Yours looks much smoother than my with the same settings.

e-post at simulacrum point de

EDIT: not necessary anymore - problem founded: the full adaptive mode of the QMC disturb the correct work of the IM. In 1/4 mode it works fine like described.

Now I must think about it. What does it mean? The adaptive sampling is a so nice method, especial for beginners.

[Jean-Paul Paloux_cg]

Problem of cloud noises on surfaces

Irmap: 25subdivs, 15samples, clr=0.2

where do I setup clr ? I don't know that...
thx !
jp

[Micha_cg]

Problem of cloud noises on surfaces

At the IM settings you find three special thresholds, one is the color threshold (clr).

[flipside]

Problem of cloud noises on surfaces

That full adaptive mode is used with qmc, so if it's too slow, you go over to ir map with normal values.

It would be better for beginners to learn the different concepts of vray instead of relying on 1 universal method. At least if you're gonna do lots of work with vray, this is the best thing to do. If you learn to drive a car, you also learn to control the throttle instead of using it like an on/off switch.

[Jean-Paul Paloux_cg]

Problem of cloud noises on surfaces

At the IM settings you find three special thresholds, one is the color threshold (clr).

oh... color threshold ! I see now

[dalomar_cg]

Problem of cloud noises on surfaces

It is an approximation of the GI so don't try to get such clean and perfect results as QMC GI with high subdivs. It is indeed a tool for medium quality, but it depends on what you call medium quality. For many people, what you call medium is already very high. I would call IM high quality, and your quality needs I would call 'exact', like maxwell.

Good point Wouter. I find IM extremely useful and quite effective for scenes of many different levels of quality. It all depends on how you use it, and how you set it up for different scenes. For animations its worth its weight in gold.

BTW what methods of triangulation are you using for these tests. The default Least Squares Fit is fast but adds more blurring to the solution. I've found for scenes with allot of small details the Delone Triangulation method is much better as it is a nonblurring interpolation method. Any thoughts?

[Micha_cg]

Problem of cloud noises on surfaces

I agree. My intention is to find a way for new users too. My experience is, that a fast easy start bring much fun and is a motivation to learn more.

An other question: when is the full adaptive method not recommended? I have the feeling, the speed difference is not so high between manual and automatic. I will test it. Or did somebody test it? Maybe the full adaptive mode is not recommended for heavy caustic stuff only.

Manual workflow: maybe a little tool could help to calculate the right subdiv value for bluryness&quality.

Back to the test scene: I find an interesting effect - it seem to be in full adaptive mode, that the IM need a stronger noise level to work right. My thesis: the noiselevel is not best adapt to the IM process. I will post my tests soon - adaptive IM with very fast calculation in high quality. I hope.

[Micha_cg]

Problem of cloud noises on surfaces

Conclusion:
The goal is to find a way, that beginners and not-daily-user get a option setup that allow to render clean and fast images in less render time.
A first step is done with Vlados unversal method - an full adaptive mode based on QMC/LC. The calculation is slow, but bring stable solutions - slow, mainly for low noise.
My idea is it, to find a full adaptive setup based on IM/LC. The IM calcualtion use approximations and is the base for fast rendering - the user set an image quality level (noise level) and not abstract subdivs everwhere.
My experiments show me:

(1) the full adaptive QMC sampler (1/100 amount 1 noise 0.01) together with IM cause sometimes artefacts (interiors)
(2) artefacts - because the noisethreshold 0.01 is to low for the IM calculation
(3) a noiselevel of 0.005 and lower makes, that the IM calculation works clean and fast
(4) a noiselevel of 0.005 is very strong for the final pass, but can be disarm by an adaptive amount 0.95
(5) no setup of the material subdivs is necessary

Here the example scene compared with the classical manual settings:


I have tested the option setup with a few scenes and it works fine. Manual settings are slower or only a little bit faster. At simple exterior scenes the speed of the full adaptive mode is a little bit slower (50s vs. 70s), but at interior the adaptive mode improve the speed. For interiors works a noiselevel of 0.005 and for exterior a noiselevel of 0.01 fine.

This method could be improved, if the noislevel of final pass and IM pass could be independent or work better together. Than the adaptive amount could be set at 100% at noiselevel above 0.005 (interiors).

Now I hope, after a whole day of test, nobody come and say: it's totaly nonsense. The method is not perfect, but a step in an interesting direction or?

PS: A now I hope I can bring this method together with progressiv rendering ala PQMC.

[joe_b]

Problem of cloud noises on surfaces

re: detail enhancement - I finally added the UI for this yesterday so it should make its first appearance in our next VfS beta (coming within the week or so ) if anyone would like to play around with it prior to our VfR Service Release / VfR4 beta releases.

[Micha_cg]

Problem of cloud noises on surfaces

... here an interior example. I don't know, what kind of dark artefacts are visible at the ground. It dosn't seems to be IM artefacts, because the IM pass was looking good.