[rdamurphy]

 Tony, on 12 January 2014 - 07:04 AM, said:

I'm a fan of Kosmos environments, it would be nice if ORTS would be able to do the same. The rain and snow effects in OR look a little to much WIP.

Cheers Tony :sign_thanks:

I'm making progress on that, slow progress, but progress nonetheless...

Robert

[rdamurphy]

The problem with stars is that the smallest thing you can draw on a computer screen is a pixel. Which, in the scale we're using for the sky, is way too large... The solution is really simple. Make the sky a lot further away and the sky dome a lot larger. It really wouldn't be difficult at that point to reproduce the major and minor constellations, and the Milky Way belt, and simply treat them as satellites like the moon and the sun, and put them in the right place by day of the year, extrapolating in either direction.

Realistically, a 2000 meter view distance to a flat plane (the sky) will make huge difference in frame rates than one 20000 meters away. It's a heck of a lot more tiles and vertices. Exponentially more.

Anyone ever done the math problem that you have to tie a rope around the equator, and then raise it one foot, and how much longer would the rope have to be?

Just for grins, try that, the circumference of the earth is 24,901.55 miles or at the equator.

Here's the formula: Circumference = 2 * pi * radius. Just to make it easy: The radius of the earth is approximately 20,920,000 feet. We'll just pretend the earth is a perfect sphere and the rope floats.

Go ahead, give it a try!

In other words, it's a trade off...

Robert

[gpz]

 rdamurphy, on 14 January 2014 - 03:33 AM, said:

Just for grins, try that, the circumference of the earth is 24,901.55 feet at the equator.

Something must be wrong with this number. I'm not in ability to even imagine any distance expressed in feets (and to recognise if dot or comma is used as decimal separator for the first look), but the circumference definitely cannot be smaller than the radius. :sign_thanks:

(Radius is 6357 km, circumference is about 40000 km.)

[rdamurphy]

 gpz, on 14 January 2014 - 04:13 AM, said:

Something must be wrong with this number. I'm not in ability to even imagine any distance expressed in feets (and to recognise if dot or comma is used as decimal separator for the first look), but the circumference definitely cannot be smaller than the radius. :sign_thanks:

(Radius is 6357 km, circumference is about 40000 km.)

You are correct, sir: The circumference should have been in miles, not feet...

Robert

[James Ross]

 rdamurphy, on 14 January 2014 - 03:33 AM, said:

The problem with stars is that the smallest thing you can draw on a computer screen is a pixel. Which, in the scale we're using for the sky, is way too large... The solution is really simple. Make the sky a lot further away and the sky dome a lot larger. It really wouldn't be difficult at that point to reproduce the major and minor constellations, and the Milky Way belt, and simply treat them as satellites like the moon and the sun, and put them in the right place by day of the year, extrapolating in either direction.

Realistically, a 2000 meter view distance to a flat plane (the sky) will make huge difference in frame rates than one 20000 meters away. It's a heck of a lot more tiles and vertices. Exponentially more.

The skybox distance is basically irrelevant, since it is not used to clip or zbuffer anything, and the size of the stars comes from the resolution of the skymap texture (and the size of the stars in that). A higher-resolution skymap would be a good way to improve the relative star sizes.

[Genma Saotome]

 James Ross, on 14 January 2014 - 08:01 AM, said:

... and the size of the stars comes from the resolution of the skymap texture (and the size of the stars in that). A higher-resolution skymap would be a good way to improve the relative star sizes.

Yes.

I went looking for high resolution star maps w/ the idea of finding one for OR that would look better on a high resolution screen and while I found several candidates they all had issues: Some were, essentially, Mercator projections and so the stars at the far "north" and "south" were highly distorted vertical lines. Some put a huge emphasis on the Milky way and others seemed to ignore it.

I realized rather quickly that I was out of my depth as far as being able to recommend any one vs. any of the others and so moved on to something else that might he productive.

James, am I correct that the starmap in use has an ordinary UV mapping such that substituting larger or smaller images will still manage to fully cover the night sky? If so then perhaps by knowing the file name folks can do some experimenting with alternative maps, give us some screenshots, and then we'll have a set of files from which people can choose their preferred image (applying their file after each installed release).

[Lindsayts]

It appears the starmaps used are "StarMap_N.png" and "StarMap_S.png" in the Content directory. The problem in finding a suitable image is that historicaly its been impossible to display the range of magnitudes (brightness) of stars on a printed page so differing magnitudes are displayed as circles of differing sizes. This method has continued into all of the astronomical programs I have so far seen, these are of course not suitable for OR.

It should not be chronically difficult to write a program (it could be almost certainly done be a shell script)to plot star positions and correct colours onto a circular image of sufficient resolution from the SKY2000 master catalog magnitude 6.5 version (this covers all the stars one can see with the naked eye). At its simplest the resulting images (2, North and South) would be circular with the pole being at the centre, the outer rim degrees plus or minus from north facing. Could OR manage such an starmap.
If so I will see what I can come up with.

Lindsay

[eric from trainsim]

Before we make the sky astronomically correct, can we perhaps stop having smoke particles pass thru collision objects?

I'd love to have an enginehouse that doesn't always look like it's on fire...

[markus_GE]

Good idea. Bridges that look like oddly shaped, crashed, smoking saucers with a train going beneath them look somewhat interesting on screenshots :rofl2:

Cheers, Markus

[Genma Saotome]

 Lindsayts, on 14 January 2014 - 12:48 PM, said:

If so I will see what I can come up with.

This might be of use: http://paulbourke.ne...eous/starfield/