Hurricane Study v2

Will · January 09, 2008, 12:46:03 PM

nice if its not too late I would say add a little texture to the right side of the hurricane its look to sold and flat. over all great job though.

moodflow · January 09, 2008, 01:45:42 PM

Building something like this procedurally would be a huge project, but I'd love to see it. I think it would require multiple stacks of fractals, done "just right" for the variation needed. Thats why image maps can be very useful, especially if their resolution is beyond the image's native resolution.

moodflow · January 09, 2008, 01:47:06 PM

As I stated on renderosity, this is the best orbital render I've seen yet. There have been others that were good too. And they were yours.

The only improvement I could really see would be to add more detail in the cumulus near the bottom left. But its not a big change.

Mr_Lamppost · January 09, 2008, 08:06:37 PM

WOW!!! The second image is fantastic (Do we expect anything less?).

I can't see my ever getting the time to try but I imagine that a procedural version could be created by defining a central point, manipulating a get position via a localised spiral transformation, feeding the result to a power fractal and controlling the density with a radial distance blend coincident with the centre point. THAT IS A LOT OF NODES

At the moment I am working on converting a specific case to a general solution that would give the radial density function and that is bad enough

old_blaggard · January 09, 2008, 11:00:07 PM

Yeah, I threw together a little thing last night that used a redirect shader and an ungodly organization of function nodes (there aren't that many, but I barely understand what I did) to spin a power fractal into a circular shape - unfortunately there's a seam and the transformation is circular, not spiral. This would a matter of a few minutes if there was a way to get a specific value from, say, a power fractal and then use some kind of "Set" function to assign that value to another point in the plane.

nvseal · January 10, 2008, 08:15:16 AM

I'm afraid I just don't have the math skills yet to try to make something like this one my own. Here is the second angle; I think I liked the first better. As you can see, there are still some issues with mask that need to be worked out.

old_blaggard · January 10, 2008, 11:57:00 AM

Nonetheless, it still looks good

. The detail of the clouds from the high altitude is impressive.

Oshyan · January 11, 2008, 02:33:14 AM

Highly impressive and realistic image, image map shader or no. Great work!

- Oshyan

Seth · January 17, 2008, 01:56:17 PM

congrats for the IOTW !

Tangled-Universe · January 17, 2008, 02:17:24 PM

Quote from: Oshyan on January 11, 2008, 02:33:14 AM
Highly impressive and realistic image, image map shader or no. Great work!

- Oshyan

Indeed, logical choice for awarding it IOTW.
Using image maps this way is just great and serves the purpose very well.

Congratulations again Stephen, well deserved!

Martin

moodflow · January 17, 2008, 05:53:06 PM

Woah! Man, this latest render is amazing....

bigben · January 17, 2008, 07:30:20 PM

Great stuff as usual. One thing I was experimenting with the last time I tried something like this was splitting the mask image up based on the area of the cloud. I used NIH Image to generate a few binary masks based on different particle size ranges, which I then used to carve up the original tonal mask image.

The main reason behind this was to match the cloud height to the relative width. A narrow cloud looks a bit silly when it's 20km thick.

But other than that the only real problem I can see is that the broadest areas of cloud appear to flat across the top. If this is partly due to your mask image you could use the above method to add a "cap" to these flat areas to put some variation back in.