Good day

[FrankB]

larger and better quality version here https://img.skitch.com/20120727-bqkirc368bas4y3e433b66rqd3.jpg

[yossam]

I really like this

[Tangled-Universe]

Great stuff, very pretty scene
The lighting direction isn't really to my taste, it gives a very bland shadowless lighting situation, except for the treebark then.

[FrankB]

Yeah it's different in regards to lighting than the usual scenes. This one has a 60 degree sun elevation, so noon-like sunlight.

Actually, I should offer some other stats, too:

+ render time 6.5 hours
+ RTA, RTO
+ highly adaptive AA: AA 20, with 1/16th min samples at threshold 0.15
+ soft shadows (that probably wasn't necessary)
+ detail 0.75
+ GI 3/5

The clouds use a new technique (for me) to arrive at these shapes and lighting.

Cheers,
Frank

[Hetzen]

Very interesting Frank.

I see what you've been playing with. It's a good challenge trying to get things to look right using that angle of lighting. The grass and tree populations look great. Is that some extra shadowing on the clouds?

The sampling on those cherries is a bit harsh though isn't it. I noticed that too in my renders after what Matt had said about theshold not being much use at high levels. I think these settings need further investigation still, like maybe forcing only two levels of sampling with a lower threshold, still trying to get loads of rays into the initial sampling.

Out of interest, how many samples are you using on your main cloud layer? It might be worth trying some crop tests to see if you can get the performance up by reducing the samples until noise starts to creep in. But I guess that would only be of interest if you were animating this.

Cheers

Jon

[FrankB]

Thanks for the comments, Jon.

To answer your questions:
There is no extra shadowing for the clouds, I have just reduced the fake internal scattering to a minimum. It's still a tad too strong in this render. The clouds are primarily "lit" through the light propagation and a high mix value.

You are right in that there is still a lot of white space on the map of custom AA settings :-)
I have simply copied the settings you have posted on the alpha forums. It's my first attempt at using custom settings at all. I still don't really understand what that threshold really means. (what does 0.1 vs 0.05 mean?)

The main cloud layer only uses like 20% of what the slider suggests it should be for optimal. I don't remember the exact value but I guess the samples were between 20 and 30 samples.

Cheers
Frank

[Tangled-Universe]

The threshold level 'decides' how many AA samples are applied.
The lower the threshold the more samples.
Logically, the max samples are determined by the square of the #AA.

If have AA4, then max = 16 samples. Then with 1/4th first samples the minimum number is 4, because (4*4)/4 = 4.
The adaptive sampling algorithm then 'measures' the noise level and depending on that it decides to either apply 4 samples or more, with 16 max.
How it is exactly doing that I don't know, as it could also be that it first uses the minimum samples, measures, and then use more samples if the noise level is still higher than the defined threshold.

How the threshold level and noise level translate into something we can understand and have reference for, is unknown.

I believe the Modo renderer lets you measure the number of samples used for every pixel being rendered at rendertime by hovering your mouse over them.
So to develop something really analytical for this then may be a feature like this would be very handy as you can see if your max number of samples are actually being used.

In your case here, like Matt suggested elsewhere, likely not.

[Hetzen]

As I understand it, Min Samples are the number of first ray hits on a pixel, if the results vary by the threshold level, then more samples are then throw at the pixel until it reaches it's maximum or a threshold level is reached.

So in the settings you used Frank, AA 24 with '1/16th first samples' you will have a minimum of 36 samples finding out what colour that pixel is. With dense geometry having only 4 initial samples, the threshold could be reached without having an accurate representation of that pixel. We've forced the issue in your render, by making sure we have 36 minimum samples to take an average from. With a threshold of 0.15, we are effectively saying, don't go onto the next level of sampling unless there is something really challenging to work out (your 36 results would have to vary by 0.15 between each other to trigger more rays).

In the default settings this threshold has a much lower decimal value, which allows more rays to be sent if that tight threshold level is not achieved.

How this all translates to mixing between adjacent pixels I'm not sure. And could be the reason for the 'blockiness' in your image (and mine looking back).

What is important about this technique is that you get an accurate main colour for your pixel, so that when the camera moves, the same area being sampled now in the next pixel, will give a similar result. Hence less noise between frames.

[FrankB]

Thanks guys, it's a bit clearer now.   

[choronr]

I like the flow and balance of this with the angle of the terrain as it starts from the lower right corner, then upwards to the left. The cloud pattern helps by following the angle thus creating nice overall depth. A pleasing scene.

[Oshyan]

Nice image and good to see you experimenting with these approaches to AA as well. Just a small note, in many of my tests I found that when using highly adaptive AA and RTA, actually *increasing* cloud samples can reduce render time. This is likely because with lower cloud samples RTA has to do more sampling to reach a given noise threshold. It's admittedly a bit unintuitive.

- Oshyan

[TheBadger]

Hey Frank,

I like the clouds where the main part of the clouds are running into the blue sky. There is some nice "wisp, and "fluff" there. There is a link to some unusual storm clouds in open that Zaai posted. Some of the more crazy formations have a bit of what you have in the image here. Probably if someone mixed and matched your cloud work, I bet they could get close to the more complex storm systems.

Sorry to move away from the main focus of the discussion here, but the clouds are what caught my eye.

[FrankB]

Thanks Michael and Bob.

Actually the clouds have been my main focus for this image. The other scene elements have just been added to not have a barren ground 

Frank