Board index FlightGear Development New features

3D OSG clouds - first step

Discussion and requests for new features. Please note that FlightGear developers are volunteers and may or may not be able to consider these requests.

Re: 3D OSG clouds - first step

Postby Arthur Maiden » Mon May 10, 2010 1:36 am

viewing at an angle doesnt help much :lol:
FGUK - A FlightGear community in the United Kingdom and Republic of Ireland
Callsign: G-PHEN
AVA Callsign: AVA0016 - RETIRED
The development of the S92 has infected my Avatar...
User avatar
Arthur Maiden
 
Posts: 680
Joined: Sat Oct 03, 2009 10:42 pm
Callsign: G-PHEN
Version: GIT
OS: Vista, Ubuntu 10.04

Re: 3D OSG clouds - first step

Postby simbabeat » Tue May 11, 2010 12:08 am

The new 2.0 and on clouds are definitely as close to real life as possible. I look up in the sky sometimes and it looks just like fgfs. Way better than x-plane!
User avatar
simbabeat
 
Posts: 3408
Joined: Sat Sep 12, 2009 12:19 am

Re: 3D OSG clouds - first step

Postby Thorsten » Tue May 11, 2010 6:25 am

The new 2.0 and on clouds are definitely as close to real life as possible. I look up in the sky sometimes and it looks just like fgfs.


*sigh* No, they are not. Stuart has done a really great job with the clouds, they are absolutely marvelous, I am slowly getting there as well with the local weather without performance being down the drain - but whenever I look up into the sky, I realize just how far away from the real thing we are.

Real clouds have self-shadows and also shadow each other. The result, especially with the sun low in the sky, never ceases to surprise me. Clouds in Flightgear don't shadow each other - too expensive in terms of CPU. The real sky has faint fields of haze - we can't really render that. From cruise altitude, visibility of cloud layers and towers is 250 km or more - the standard 3d clouds have 20 km, the local weather clouds offer 30 km - that's nothing like the impression from a real airliner. The real sky has much more variety in shapes and sizes than we render. The real sky has layered clouds fading into patches of small cloudlets - we can't really do that.

Should I go on?

It's one thing to be excited about how good the clouds in Flightgear are. But it's not even remotely true that they could actually measure up against the real thing.
Thorsten
 
Posts: 11765
Joined: Mon Nov 02, 2009 8:33 am

Re: 3D OSG clouds - first step

Postby erik » Tue May 11, 2010 7:08 am

Well, the aim of the project is to create a flight simulator, not a world simulator so tradeoffs are inevitable.
Usually distant objects are rendered using imposters; large rectangles facing the viewer that contain just one (of a fixed number of) textures, giving the impression of depth.
erik
 
Posts: 1706
Joined: Thu Nov 01, 2007 1:41 pm

Re: 3D OSG clouds - first step

Postby simbabeat » Tue May 11, 2010 9:55 pm

Thorsten wrote:
The new 2.0 and on clouds are definitely as close to real life as possible. I look up in the sky sometimes and it looks just like fgfs.


*sigh* No, they are not. Stuart has done a really great job with the clouds, they are absolutely marvelous, I am slowly getting there as well with the local weather without performance being down the drain - but whenever I look up into the sky, I realize just how far away from the real thing we are.

Real clouds have self-shadows and also shadow each other. The result, especially with the sun low in the sky, never ceases to surprise me. Clouds in Flightgear don't shadow each other - too expensive in terms of CPU. The real sky has faint fields of haze - we can't really render that. From cruise altitude, visibility of cloud layers and towers is 250 km or more - the standard 3d clouds have 20 km, the local weather clouds offer 30 km - that's nothing like the impression from a real airliner. The real sky has much more variety in shapes and sizes than we render. The real sky has layered clouds fading into patches of small cloudlets - we can't really do that.

Should I go on?

It's one thing to be excited about how good the clouds in Flightgear are. But it's not even remotely true that they could actually measure up against the real thing.


Hmmm I hadn't thought that in depth about it. You're right. One thing we are missing now is shadows underneath the clouds. The whole cloud right now is the same shade of whatever color. The bottom of the cloud is a lot darker because the sun isn't directly on it. All I meant to say was that the new textures look a lot like the real thing, albeit not perfect.
User avatar
simbabeat
 
Posts: 3408
Joined: Sat Sep 12, 2009 12:19 am

Re: 3D OSG clouds - first step

Postby Arthur Maiden » Wed May 12, 2010 5:37 pm

erik wrote:Well, the aim of the project is to create a flight simulator, not a world simulator so tradeoffs are inevitable.
Usually distant objects are rendered using imposters; large rectangles facing the viewer that contain just one (of a fixed number of) textures, giving the impression of depth.

well tbh, clouds do impact quite a lot on Flying... tell tales for updrafts and weather fronts, also cloud suck is quite an interesting subject
FGUK - A FlightGear community in the United Kingdom and Republic of Ireland
Callsign: G-PHEN
AVA Callsign: AVA0016 - RETIRED
The development of the S92 has infected my Avatar...
User avatar
Arthur Maiden
 
Posts: 680
Joined: Sat Oct 03, 2009 10:42 pm
Callsign: G-PHEN
Version: GIT
OS: Vista, Ubuntu 10.04

Re: 3D OSG clouds - first step

Postby erik » Thu May 13, 2010 8:33 am

True, that's why Thorstens work is much appreciated. But self casting clouds (although that looks wonderful) is of little use for simulation. So if it takes too much computer power it might be left out (for now).
erik
 
Posts: 1706
Joined: Thu Nov 01, 2007 1:41 pm

Re: 3D OSG clouds - first step

Postby Thorsten » Thu May 13, 2010 10:48 am

One thing we are missing now is shadows underneath the clouds. The whole cloud right now is the same shade of whatever color. The bottom of the cloud is a lot darker because the sun isn't directly on it.


That you can have... The shader 3dcloud.vert has a line

Code: Select all
float shade = usrAttr2.g;


which sets the shading to a value obtained somewhere in the core. You can add a line to the shader below overriding that setting stating

Code: Select all
shade = 0.0;


which gives clouds with very dark bottom and away side shading (essentially shade is a number between 0 and 1 where 0 means strong shades, 1 means no shades).

But we can't do clouds shading other clouds. Even if we could spare the CPU power to raytrace the problem, cloud models only look like clouds from the view position. Moreover, their actual position is not the visible position, just their appearance is rotated, but their coordinates (e.g. for a collision finder) are not where the visible layer is. So the cloud models really look like bunches of sandwitch layers hanging in the sky, and so would their shadows on the ground. Thus, the raytracing code would need to deal with yet another model which specified how the cloud casts a shade.

But self casting clouds (although that looks wonderful) is of little use for simulation.


Well, we do lots of things which are eye candy and not useful for simulation, such as texturing the ground or plaing airport terminals. One of the reasons I like flying in real life is the view - so naturally I'm also into great views in simulated flight.
Thorsten
 
Posts: 11765
Joined: Mon Nov 02, 2009 8:33 am

Re: 3D OSG clouds - first step

Postby WooT » Sun May 23, 2010 8:18 pm

But self casting clouds (although that looks wonderful) is of little use for simulation.


In the case of thermal soaring with gliders, the shadows of the clouds on the ground is a very good help for orientation. When you are spiraling over a flat country with little visual references, the shadows allow you to know where the sun is, and thus your orientation. I suppose this is also true for general aviation orientation when flying VFR.

Also the shadows that clouds cast on themselves allow you to know where the thickest part of the cloud is. This is very useful in a glider when you are flying just under the cloud base and cannot see the cloud as a whole. The darkest area is often where you will find the thermal core.
WooT
 
Posts: 92
Joined: Tue Mar 17, 2009 4:09 pm

Previous

Return to New features

Who is online

Users browsing this forum: No registered users and 1 guest