Q about Ray traced Atmosphere

[reck]

Does the samples setting in the cloud layer make any difference when using the ray traced atmosphere option?

I thought I read that the quality (noise) was controlled via the AA when using the ray traced atmosphere option and not the amount of samples.

[Henry Blewer]

I just posted a render using 21 samples in the atmosphere quality. The atmosphere came out very clean! I used 1 for the cloud quality. It may not be necessary to use this in the clouds, but I have not tried this yet.

Edit: I just looked at the cloud layer in my last image. It only uses 9 samples.

[Tangled-Universe]

The quality of your renders when using the raytracer render depends heavily on the AA setting.
The higher the AA the better your raytraced objects look, but also the better the raytraced atmosphere looks.

So the higher AA, the lower amount of atmosphere or cloud samples you'd need to achieve the same visual quality.

For example, if you have a clean render result with 32 atmosphere samples with AA4 then it might be possible that you'd only need 20 atmosphere samples with AA6 to achieve the same clean render (when using the raytraced atmosphere function).

If you have models in your scene which you render with the raytracer it is very likely you will render with an AA setting of 6 or 8.
Consequently this would mean that would not require as many atmosphere samples as you'd normally do when you wanted a perfect clean render.
When rendering with AA6 or 8 it is probably best to keep the samples at default of 16.
One of my latest images (Overcast Weather) uses raytraced atmosphere with only 8 atmosphere samples with fully sampled AA 6 and no noise as result!

Thus, another interesting facet in raytracing the atmosphere is that the adaptive AA sampling can save you quite some rendertime without sacrificing too much of your final quality.This works really well if your scene contains lots of atmosphere and no models.
For instance, try to render with adaptive sampling of 1/16th first samples. It will speed up atmosphere renders pretty much.

Cheers,
Martin

[rcallicotte]

Thanks.  This is good to know and opens up a completely new way to see the renders alongside atmospheric sampling.

[reck]

Thanks for such a detailed response Martin. This answered my question and more.

How do you all this stuff though? Do you have a hotline to planetside central that you can use whenever you have a query? 

[Henry Blewer]

Marin is an alpha tester.

[reck]

Yeah but how do you know all the answers by using alpha software?

[Henry Blewer]

I am sure that Matt and Jo send out explanations of the features they want tested. That's how it was done for OS2.

[reck]

OK so a related question. Why wouldn't you always use ray traced atmosphere, what's the downside?

[Tangled-Universe]

OK so a related question. Why wouldn't you always use ray traced atmosphere, what's the downside?

That's a difficult question. With many things there's a certain "threshold" where raytracing becomes faster than not raytracing.
For example, when using low AA values you might have faster renders with the rasterizer (the non raytraced renderer which renders the terrain and also objects and atmosphere before).

Yeah but how do you know all the answers by using alpha software?

Being an alpha tester sure helps a bit indeed, but don't over-estimate the amount of info we get straight away
Basically we start with release notes similar like you have seen them here before.
All the rest is due to spending many hours with the software.

Cheers
Martin

[Henry Blewer]

There is also the downside of being an alpha tester. Reports, detailed explanations of how you crashed the software. Hours spent testing feature which are not released to the public. Lastly frustration. Such and such works on TesterA's machine but not on yours. This makes testing the failing software on you machine more critical and demanding. Why does it work on A's machine?