Jump to content
  • Advertisement
  • entries
  • comments
  • views

Final Earth model

Sign in to follow this  


I think i can safely say that now, everything is close to what i envisionned. In disorder, i fixed the sun color at right angles, the contrasts of the starfield, the quality of the cloud textures and adjusted some parameters. The first screen is an Earth-like planet with an atmosphere thickness of 250 kilometers (gives an "artistic" feeling to it), the second one has a thickness of 100 kilometers, which is more realistic, but looks less nice in my opinion.

Reducing the images to fit in this journal makes them a bit blurry. For fun, i upped the resolution to 1600x1200 with antialiasing x6, planet textures 1024x1024 and took a screenshot (maybe if some of you need a new wallpaper). It still rendered at more than 60 fps on my machine.

Link to wallpaper-ed image (500 Kb)

I can finally move on to other tasks. I need to work on my 4E4 entry, too..
Sign in to follow this  


Recommended Comments

Guest Anonymous Poster


WOW! Awesome work!

Couldn't you get rid of the black atmosphere in the unlit region?
Maybe with a hack like, say deform the unlit portion of the atmosphere sphere to bring that portion inside the planet sphere. Or simply displace the whole atmosphere sphere towards the light source.
All this is you choose not to make the atmosphere truly transparent for the background.

Anyway, it still looks wonderful!

Share this comment

Link to comment
Personally, I think the second one looks muhc better. You might think it doesn't look as nice, but compare it to how it looked without ANY atmo scattering:

Share this comment

Link to comment
Thanks :) Cipher19: i don't know, i still prefer the first one. It might be less realistic, but it has a little something that i like. It's a matter of taste i suppose :) And in space you'll find all sorts of planets and all sorts of atmospheres. Actually this makes me wonder about something: when playing with the scattering/atmosphere parameters, i can easily get an ugly or unrealistic result. Since i'm generating the planets procedurally, i cannot simply choose these values randomly. I wonder if i can evaluate with a kind of function how "aesthetic" the result will look, to reject the combinations that look bad..

Share this comment

Link to comment
That look great! Strangely enough, I like the second one better too, the first atmosphere seems to catch too much light to me.

Share this comment

Link to comment
By the way, i sent the first image as an image of the day on flipcode. I wonder if that site is still alive though.. :)

Share this comment

Link to comment
Great work, indeed!

I really wonder how you solved the issues concerning the lookup tables and the regions that were too dark and so on.

The huge image is really beautiful. There are still some cubemap seams left, but if you get rid of them, it's perfect.

Share this comment

Link to comment
Hey Lutz.

I didn't use lookup tables finally, the quality was just too bad. So i used a vertex shader like you, but i simplified and optimized it in a few parts. It is now around 79 instructions (in ASM).

The two main optimizations are:

- only one ray/sphere intersection test: the one from the sun is avoided by the simplication i already described by mail.

- the main ray/sphere intersection test was optimized in a couple of ASM instructions only. To do that, i cast a ray from the camera to the vertex direction instead of the opposite. This makes it possible to precalculate a few operations on the CPU and upload it to a constant in the vertex shader.

Thanks for the cube map seams, i didn't notice them myself :)

Share this comment

Link to comment
I also did the ray-sphere-intersection for the ray camera->vertex first, but there is a catch: If you are far away from the planet, you will get precision problems. Here is why:

You typically want to know the length s of the ray passing through the atmosphere. So you first compute where the ray hits the sphere around the planet with radius (planetRadius+atmosphereRadius). Then, you subtract the vertex position from this intersection point. Both are big numbers whe you're far away from the planet, so a lot of precision cancels out.

Try zooming into the planet from far away.
Does it flicker?

Share this comment

Link to comment
Lutz: i understand your theory, but i cannot see any flickering. And i've tested both on an NVidia card and an ATI one.. anyway if i get this problem later on, i can revert back to the original algorithm. Until then, it's all good :)

Superpig: i'm afraid not. Actually i'm wondering if i will finish my entry at all. Time is running and i have a deadline for the Minas Tirith project coming beginning september, and since it's a collaborative project with a huge team, i cannot miss it. That means i'll have at most one month to finish everything, and i haven't done a lot yet :(

Share this comment

Link to comment
Damn. I was looking forward to your entry being one of the ones to provide a challenge for my GF7... [sad]

Share this comment

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Advertisement

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

GameDev.net is your game development community. Create an account for your GameDev Portfolio and participate in the largest developer community in the games industry.

Sign me up!