Guys, you know I'm not a specialist but I have the feeling most screens of this thread are showing night lights much too strong. I wonder if some specific code should not being implemented to the Celestia engine to dim a night map depending on the distance. I have the feeling a city seen from a few kilometers will show much more light than seen from a few hundreds of kilometers. Of course the visible zone of light will go reducing itself depending on the distance but I feel it's not enough. To me the local "glow" visible at short range should tend to become invisible with distance. What do you think?

Christophe,

I disagree, take a look at the actual city photos shot by astronauts!

Fridger

_________________

Christophe,

I was in a hurry in my previous short response. Qualitatively, I think there is no real problem, though.

The weakening of light according to the inverse square law of physics (~ 1/distance^2) is implemented in OpenGl rendering and thus also in Celestia.

But this law holds first of all in empty space. In addition, atmospheric extinction effects have to be taken care of. This means that a well defined angular dependence of the extinction must also be implemented. This depends on the "optical path length" within the atmosphere the light encounters between its source and the observer. Since many aspects of Celestia's atmospheres are still in their infancy, here we really need improved code, first.

Meanwhile, there are various workarounds one might try. Like using mip-maps for the DXT encoded nightlight textures, which can be individually darkened by hand, the smaller their resolution scale becomes (i.e with increasing mip-map number). There is a suitable NV tool for this operation.

Remember that elsewhere I have also alluded to some work about nightlight textures that still needs to be done

Another simple option is to play with the Mie parameters, which I have done in case of my above images.

Fridger

_________________

Yep, I was thinking in such solution! Now the difference between each level should be not too evident but I think it deserve a test. My problem actually is that i have no tool to extract mipmaps from dds...



Could you post your settings?

Atmosphere
{
Height 60
Lower [ 0.43 0.52 0.65 ]
Upper [ 0.26 0.47 0.84 ]
Sky [ 0.40 0.6 1.0 ]
Sunset [ 1.0 0.6 0.2 ]
CloudHeight 7
CloudSpeed 65
CloudMap "earth-clouds.*"

# Mie 0.001
# MieAsymmetry -0.25
# Rayleigh [ 0.001 0.0025 0.006 ]
# MieScaleHeight 12
# my old values
## Mie 0.0005
## MieAsymmetry -0.15
## Rayleigh [ 0.00025 0.0009 0.0015 ]
## Absorption [ 0.00018 0.00005 0.0 ]
## MieScaleHeight 18
Mie 0.0050
MieAsymmetry -0.35
Rayleigh [ 0.0008121 0.0020775 0.00375 ]
Absorption [ 0.00057 0.0004 0.0 ]
MieScaleHeight 15

}

The sewing and extraction tools are in the old NV texture tools, but only for Windows users.

Fridger

Tx! This settings is pretty nice! I feel the white is a bit over exposed when you look at the day side from the terminator (night in your back) but as know nothing about the mechanism implicated here, I keep your version!

Thanks again Doctor Schrempp...

I applied your revised Mie settings as described in the post above, and I found a noticeable improvement in the terminator and night side lights appearance. These will probably be more dependent upon individual user preferences, but I very much like the effect from your settings here.

Thanks very much for the explanations and code.

I simply love this place.

EDIT:
One other note here... The Virtual Texture night lights are very pleasing to me personally, but I get the feeling that the night lights available from the regular Celestia distribution are too bright. Is this the case? Or, do I understand you to say that the night lights should be rather bright?

At any rate, I very much like the Mie effects. Thanks again, Brain-Dead

Bob,

you can't compare the nightlights from the distribution with the efforts that went into these 32k nightlights, based on the radiation calibrated original data and fancy coloration via real city photos from astronauts.

I agree with you that the nightlights from the distribution are too bright. But I have not used them really since ages.

Fridger

_________________

With the new 2012 map that is out
and the links for some of the old data , now dead .

this needs to be revisited

from the 2006 Rad. Cal. map
http://www.ngdc.noaa.gov/dmsp/download_radcal.html
-- a quick "back of the envelope" ---

the 32bit float image
"F16_20051128_20061224.cloud2.light1.marginal0.fg_15_35_55.full_swath.line_screened.shiftx2y2.op_blended.avg_vis.tif"

has a range from 0 to 6030.7 ( i am using 6031)
-- center of the image, just above the cursor

move this into UCHAR ( divide by 23.65 Approx. )

And downtown NewYork has a max value of 76

--- below a very OVER exaggerated image
then the normal one ( for the 2006 Cal. )


-- the EO 2012 night map ( New York )

and moved so downtown is 76 and not 255



just a one point quick look
other cities might not be the same

So the lights are rather "dark"

The"normalized" and exaggerated map
http://earthobservatory.nasa.gov/Featur ... /page3.php
dose make a nice" wallpaper" ( though the background is WAY to blue )

This is where "t00fri" and i differ
(A good difference of opinion. )

---- edited - new map ------
---- using the 2006 cal. as a "burn" mask -----

-- with NY set to about "160 to 190 average" in the map i am working on
-- screen capture --


screenshots

_________________
"I don't pitch Linux to my friends, I let Microsoft do that for me."
Using OpenSUSE 42.1 & Scientific Linux 6.7

seeing as this page is linked to shatters

A new thread was started
Here:
http://forum.celestialmatters.org/viewtopic.php?t=451

_________________
"I don't pitch Linux to my friends, I let Microsoft do that for me."
Using OpenSUSE 42.1 & Scientific Linux 6.7