Earth with better Atmosphere
Hi,
I have been playing with planets trying to get a good photo-real one.
My biggest problem has been the atmosphere. I have tried halos, but they disappear halfway out of the fame. You could use a tube primitive, but that would have to be somehow tracked to the camera if you wanted to move. I found another way. It may be no better then the others, but have a look.
I followed Enrico Valenza’s amazing tutorial here:
http://www.enricovalenza.com/realplan.html
However, I made some modifications… Here are my test results.
First, I consolidated the image maps so there are only two: ground (with alpha channel) and cloud (no alpha, needed one). I used the largest possible maps I could find. The only procedural is stucci with its size set to zero just like in the above mentioned tutorial.
The only light source is one sun lamp set slightly yellow (very slightly).
I also have the same sphere setup in the same order. Then only difference here is that I added a fourth sphere for the “fuzzy” atmosphere instead of the tube primitive.
Most of the materials are similar. I did add a color ramp to the surface to accent the fresnel effect of the atmosphere. The alpha on the clouds is also not zero. This is because I told blender to calculate the alpha as I did not have an alpha map for the clouds. The raised alpha effectively “sharpens” the clouds. This in conjunction with the “Mitch” OSA setting and the sharpen node seems to reveal enough cloud detail to make it more believable.
The last fourth sphere is just for the fuzzy atmos. Because of the render layers feature and the nodes, I can screen it on over the normal earth model. The way that the color band is set up the gradient appears to fade off into black. But, because I had to shift the colors right to make room for the black falloff, the fourth sphere has to be set considerably bigger then the other three. The screen makes it so any ships, stars, nebula, etc, can show through a little bit.
All this makes it possible to almost (yes, almost) fly down into the atmosphere. It also is always oriented in the right direction because the material is still light and camera dependant.
If any of you come up with any better ideas I would love to hear them. I think until blender’s volumetrics are fully developed we will have to find shortcuts like this. However, even full volumetrics will have their drawbacks – very long render times. This setup is ok about render times (21.67 sec. for NTSC no vector blur on a decant system).
I will upload the .blend file as soon as I figure out how/where to do so.
have fun,
Kevin
