Amazing!

The distance view is terrific
The 3d clouds effect will afford some work but as long AI is able to X-ray these clouds the disadvantage will be on my side (clouds hiding the enemy )

Time to play online then!

I posted this at the Gen, but really thought it might be better
if more heads were involved.%^)
Maybe we have a modder here that is familiar with 3D Studio and could
possibly make some cloud sprites for us. I do not know if we would
be able to use the ones in FS2004,(copyrights?) or the possibility
of acquiring the plug-in add-on that Niniane Wang made for 3D Studio.
In any event, there is a very interesting PDF link I included at the
bottom of this post on how Ms. Wang made 3d clouds for FS2004 and
made them as FPS friendly as possible.
AO
=================================================================
This is from the young lady that did the cloud system for FS2004.
Very interesting stuff......if I could just find the tools.....for free..

CloudsInGames_NinianeWang.mpg


In Niniange Wangs' video, I saw the same 2D Octoganal ring that
we use in EAW for the horizon clouds. But I notice that she was able
to get a few different 2D pictures on the individual octogon faces,
where WE seem to have the same repetitive picture on each face.

If we could make the octogon faces with, for instance, the 2D cloud that
Col. has been showing, with a few massive 3D, it might look right, IF we
can make the massive 3d without a big FPS hit. Maybe the individual
sprites that have to be massed to make a 3d cloud, could be made larger
or smaller than the average 512 x 512 size. (Whichever would have a
smaller FPS hit.) FS2004 uses 512 x 512 3d clouds on average, they also
use 128 x 128 and 64 x 64.





I also would look into the possibility of asking Wang about possibly letting
us have or use a 'light' version of a bmp or something that we could
incorporate into EAW. You can find her with a google search. There has
got to be SOMETHING we can do to make this a possibility, and be FPS
friendly.
AO


edit: From a google search, a PDF from Ms. Wang. A very informative
read!:
Making 3D clouds PDF

A few thoughts on clouds in flightsims.
There are mainly two people that published about them, one is mark harris ( http://www.markmark.net/clouds/ ), who has already been mentioned in this thread.The other is Niniane Wang, Google for
niniane wang clouds

A (nearby) cloud is normally made up of many "sprites" that always face the camera.

Framerate in games is normally determined by three aspects:
- Number of polygons in the scene
- Number of objects to draw (technically "drawcalls", is most importtant in DirectX9 and may not really apply to EAW).
- Number of pixels to render
Many large transparent polys means the number of pixels increases as even completely transparent areas count. An easy way to get an idea on the bottleneck is to increase screen res by say 50% and see whether the FPS also change by 50%, or by 0%, or something in between.

Not having a ZBuffer might be an issue.
OTOH most modern games I know down make a perfect job either and when in a cloud, each sprites basically either is in front of a plane inside the cloud or behind (see Harrisons work for improvements on that). This seems to be ok 99.9% of the time. Of course what should be achieved is that some sprites of the cloud can be in front and some behind the plane.

Guys,
I sent an e-mail to Niniane Wang:


She already gave me a reply:


Can anyone (Ralf?) respond to this? I'm going to e-mail her back and refer her to
this thread. (link.) I'll let her post under my logon if she wishes.
Maybe she will read it and think us all mad.....
Or maybe she might respond.
She already responded to my e-mail, which is already more than I expected.
AO

Sprites don't need any tools to create. They're just simple flat rectangular pictures and can never be 3D. Sprites date back to the Commodore 64 and ZX spectrum, which used them too. They can be rotated and resized but that's about all they can do. In an array they can sit behind eachother and their position in the array will tell the sprite when it's in front or behind another one, however in 3D space they all occupy the same space since they're superimposed onto whatever the programmer wants in front or behind it, unless they use additional coordinate parameters which tells a program when to switch em on. Sprites allways face the camera when visible and can also be in front or behind eachother depending on their drawing order. Closing in on the sprite only zooms in on the picture, that is if a program is told how to use these coordinate parameters.
Flying through sprites is an illusion, only when the sprite occupies a large part of the screen it gives you the idea that you're in it but actually you're in front of it. In the array a program can be told to switch off the first sprite so the 2nd one now seems to be in front of you, the first one can now be told to become visible when the camera pans 180 degrees, giving the ilusion it's behind you. Obviously this will need some programming which keeps track of the camera position so the switch can be made when for instance a camera's X, Y or Z coordinates corresponds with additional coordinate parameters linked to the array and each sprite's location in 3D space, or it's coordinates can be updated so it will seem to be passing the camera, while at the same time it's resized.

Niniane's proposal is exactly what Ralf has done and the only possible improvement is to improve on the picture, use larger sized sprites or create an animation. Another option is to write code which uses current hardware (vertex) features of videocards or newer directX implementations or use the Microprose idea of 3D models.

VonBeerhofen

Hi AO,

Clouds within games is very old technology.
How they are created and the optimizations are very well documented all over the internet.
Here is one of the simpler sources explaining volumetric cloud creation:
http://www.markmark.net/PDFs/RTCloudsForGames_HarrisGDC2002.pdf

The real problem is that EAW uses such an ancient rendering system that practical cloud creation is virtually impossible to implement in that sort of system and still have playable FPS.

One can not effectively backport volumetric cloud creation techniques into a DX6 rendering engine where the Z buffer and Shaders are not able to be used.

Regards sydbod.

Hi Sydbod,
yes....I gathered that from VB's and others post.......
After seeing Col.'s and Ralfs efforts, it was very wishful
thinking on my part......

The past several days I have been trying to utilize some of the tools
that have been made by the community through the years,......and after
reading several readme's and docs, I have been amazed with the amount of
knowledge and experimentation by this community, and of course, embarrassed
by my own lack of knowledge of same......
I can only hope that some of my uninformed questions might have at least
stimulated a thought to a possible solution.

I really enjoyed Niniane Wangs PDF about cloud rendering, as it gave me alot of
information I never knew......but also made me realize that you guys already knew this
stuff.
So,....who has been making horizon clouds and skies? Can any be made that have the
large cumulous clouds that Col. was showing in his screens?

If I learn how to use these tools, I'll eventually make them myself.
(Although it may take me a few......years!
AO

Hi AO.

As Sydbod, and Allen have pointed out several times, our Achilles Heal is our out of date 3D engine, and it's use of DX6.

I think we should start looking at updating our code to DX9, but, and here is the but, how do we do it? We are not up to re writing the entire 3D engine, but is this necessary to implement DX9.

I know Sydbod has been working on a more modern 3D engine, so maybe he could give us some insight into what needs to be done.

Perhaps if we knew what was needed, we could start making some plans.