Mammatus Layer - WIP

[dandelO]

I thought I'd better at least post a preview image of where I am at the moment with my mammatus files, instead of just gabbing about it with no results to show.
It's now getting a bit easier to manage as I tinker with some different methods than I was before.

Before, I was using an inverted profile cloud to make the tops push downwards instead of up but this was giving me the headache of needing to extend the depth of the entire layer to make each 'finger' extend further. Sample levels were just getting to high for me to manage this way, I've no time for the multitude of test renders I need to do(a lot), with such extreme cloud samples. I needed another way. Imagine, the hiking of samples, when you need to add sometimes hundreds of metres in cloud depth between incremental test rendering, for an extension of the mammatus features, nope, I can't manage that, it's too cumbersome a method for this purpose.

So, I figured, if I could use a fixed depth of cloud, that would keep sample levels manageable for me. The layer is a constant depth of 3000m now, this'll keep the balance of settings/samples reasonably tight, I'll explain the new method of how I'm working it once I'm done, I'm getting closer, I think.

What I need to concentrate on, for the moment, is lighting and contrast. THIS is tricky. I have added a bit of post-render contrast to this image to show the features a bit better(still a bit washed-out), the detail is becoming washed-out with GI and scattering colour very easily because there is very little direct light making its way through the layer. The forms are nearly there, lighting is ongoing.

New WIP...
[attachimg=#]

[dandelO]

I need to somehow invert the lighting effects, the creases should be lighter than the actual protrusions. The light is backwards in most mammatus clouds, that I can see.  Back to the drawing board...
[attachimg=#]

[Henry Blewer]

You are making progress. It would look better if it were darker.

[dandelO]

I have many darker versions, Henry. The above ones use some extra fill lighting to compensate, without it, the above images look roughly like this - brushed metal or haematite, in other words, not cloudy (this has also since been edited to brighten it up, I'm now only testing using GI, no extra filler, I'd like these to be universal to any scene)...
[attachimg=#]

Like I say, WIP. I just picked a couple of decent looking demo images from around a hundred billion. Ha!

[dandelO]

I'm also aware of the bad blending between the protruding and flatter density fractals, too. It'll be a bit smoother than these wip's...

[dandelO]

I also need a more uniform method of fingering( ), power fractals are very erratic and random, hard to tame, if you know what I mean. I'd like a repetitive-noise control in TG, one that I could, for example, warp afterwards. Unfortunately, I'm no function genius but, I'll get there in the end, might be a while...

[Henry Blewer]

I have been reading the function node page http://www.planetside.co.uk/docs/tg2/noderef/ which has helped me get a better grasp of what they function nodes do. Perhaps a perlin noise with a clamp vector(?) function after it would provide more repetitive results.

[choronr]

On very rare occasions have I seen these types of clouds which are near tornado conditions. Great work by you Martin ...keep on.

[domdib]

Very intriguing.

[bla bla 2]

trop bon.

too good

[Oshyan]

Very promising indeed. If your other method was even more promising but render time prohibitive, you could try Raytrace Everything during your tests...

- Oshyan

[dandelO]

The old method was to stretch the length mammatus fingers with the cloud depth setting, I needed very high depths of clouds to make it work correctly. I am using RTE for all of these.

My new method(in these rough images) is to start with a constantly deep enough cloud layer then, I distribute and remove the fingering function from that constant depth layer.
It means I can keep a big 'block' of cloud in the air, like a block of clay or wood, for a decent trade of samples/time, about (3000-4000m thick seems fine) and 'carve' the shapes out of that without needing further stretching to extend the bulges. I'll get back to this one soon...