A natural situation

[yonosoy]

Yes, is better option PT simple with atmospheric simulation in general.
For this case more. Just a lot of test together in every try ...
It is difficult to work around XML format and the files was oversaved sometime erasing the last tries.
(and a demand for this exercise was not expected...)
Better more over.
Thanks again.

[yonosoy]

Hi again.
Best try this week. Is the same pigs base. 350 spp now.
The image will be substituted with the final run tomorrow late.
Also the IGI solution will be linked.
SS 4, Path tracing simple and some minor changes (more correct scale and expression and detail part erased). The biggest one at the moment.

Cheers ! Thanks in advance.

[OnoSendai]

Amazing!
Looking forward to seeing a more converged image!

[Zom-B]

Hoi yonosoy, that looks quite awesome!!

Anyway I would suggest to modify the box with the SSS medium to not be that uniform

[yonosoy]

Hi.
This is the first good setup of the serie (A_natural_situation_A is bad connected...).
Now a demand for a high resolution is deserved.
Enjoyable.
Thanks for all OnoSendai.
Zom-B agree, for more over.
Thanks in advance.
this small one runs more over and will be substituted.

[yonosoy]

A lately new.
The value of max. coef. of extinction for the volume raised again to 0.015 and a subtle modification on the clouds code...
This last will run to the end because the extreme noise seems to disappear. In earlier test 4K spp was enough.
Faster and bigger volume this time.

Classic edit: now phase function jump from mie_cumulus to uniform when the max. coeff. of extinction is changed (no more when the scene is saved, ...
The pigs was retired and by error the image ... tomorrow a correct pigs and the image.

[OnoSendai]

Hi again.
Best try this week. Is the same pigs base. 350 spp now.
The image will be substituted with the final run tomorrow late.
Also the IGI solution will be linked.
SS 4, Path tracing simple and some minor changes (more correct scale and expression and detail part erased). The biggest one at the moment.

Cheers ! Thanks in advance.

I'm going to leave this one rendering on my computer for a few days.

[yonosoy]

Hi Ono, please correct the value of the displacement on rock B, must be the same tree. Is the fault here.

[OnoSendai]

Sorry, didn't get a chance to correct the displacement.
Here's the results after 90 or so hours.
Unfortunately it's still noisy, I think this is due to the Mie cumulus phase function which is extremely peaky.

[Voytech]

Sorry, didn't get a chance to correct the displacement.
Here's the results after 90 or so hours.
Unfortunately it's still noisy, I think this is due to the Mie cumulus phase function which is extremely peaky.

Just a quick run through noise removal in Adobe.

[Oscar J]

Only thing that stands ot to me is that the mountain looks like it's made out of plastic. Maybe you can fix that and region render. Would fit like a glove in the gallery/frontpage.

[yonosoy]

Hi. After a week of debugging ... nothing conclusive.
The fact that the code don´t fits well with the box that defined it introduces noise. Also is better to remove the residual SSS part in the equation and set the IOR again to 1. But in this position (over the clouds at 6 km of high), Mie is really hard to converge.
Voytech, thanks for the attempt of rescue. Here one that runs across darktable and rawtherapee in linux and finally lightrom in windows (long travel, need a fine measured tune ...).

Oscar, so many thanks. Sadly the shading on the mountain was not performed... still fascinated with clouds.
This one is cooking and has actually 2.1 K. Is the final run for this exercise.
Tonight the 4 K output with the denoised one to contrast with the hope that is clean enough for this stage.
Gallery? ... not this year (thanks for this too).

[Oscar J]

Denoising destroys the realism in my opinion. Sorry.

Aren't there any more optimisations that can be performed?

[OnoSendai]

To speed up the cloud rendering, it's probably best to use the Henyey-Greenstein phase function instead of Mie-cumulus. It will still show the desirable forwards scattering, but will converge much faster.

[yonosoy]

Yes Oscar (thanks, the steps are usually forgoted !).
In Sunday a test with SS 6 and some corrections in the scene gives a good result over 2 K. Sadly this last run is over SS 3 because of some last corrections .... and the hope than a mistake in the setup was the cause of the noise. Better option than an external denoise is the use of indigo´s SS though. For the next situation is a factor to take in account ...

And yes Ono, only one test in this direction but thinking in it again. For this concrete situations of light is spectacular too, to take in account more over. Thanks.

Edit: Voytech, how terrible words one beneath the other! Excuses, thanks because the fact of your interest (this called "attempt" is so far a reasonable and balanced denoise and is not intentiononed nothing more than research for a limit combining some tools, is not an attempt of improvement, believe this... in fact the second one is destroyed ). SORRY, some excess of stress ...
Also the emoticon shown to Oscar must be more a mix than , , and (amazed with the results lately, but the peaks of euphoria and depression distort a lot of things (personal notes)).

So this is the 4 K output (no denoised). A brief try with henyey-greenstein looks very very good, for the next one.
Now is time to move over some other parts of the workflow and the next ISL test will be for other months.