There is always a side of the moon on which shadows seem to be cast the wrong way. That makes some of the craters look convex instead of concave.
Since I cannot get the moon map to be right in all situations, I tried blanking it and checked how it would look like only with bump mapping. I got a perfectly flat white moon. I tried to raise the BumpHeight parameter without luck; still a perfectly flat moon. Did anyone get better results ?
My graphic card : ATI Radeon.
Utility of bump mapping for the moon
-
Spaceman Spiff
- Posts: 420
- Joined: 21.02.2002
- With us: 22 years 8 months
- Location: Darmstadt, Germany.
Better results using Albedo Maps with Bumps Maps.
I have much better results, but I can't show them yet, because I have no public web space to park the snapshots I made. I have 7 little jpeg 'thumnails' (320?230) to illustrate my results and this point...
Coming from the 'What Celestia should be' debate (http://www.shatters.net/forum/viewtopic.php?t=163), I'm restricted in input on matters of accuracy ('realism') until I sort out image submission. I left this issue of shadowing out of 'realism' though because it is an easy one to fix given Celestia's great flexibility in data files.
The problem arises because the texture map is an air-brushed 'topographic' map for lunar cartographers. It is designed to show all features illuminated at a constant angle by light from the east - a standard convention - so shadows are fixed. Currently, most other moons in Celestia suffer this problem, esp. Phobos and Saturnian moons.
My solution is to use an albedo map in place of a topographic map, but keeping the bump map. An albedo map is a map that shows suface colour and brightness without shadowing. They are built up as photomosaics from images where the Sun is high up in the sky over a planet or moon. Texture maps for Earth, Mars, all Gas Giants and the Galilean moons delivered with Celestia are already albedo maps. Only planets and moons that have been imaged with enough coverage to gain the data can have complete albedo maps.
I found an almost perfect albedo map for the moon at the USGS Clementine web site:
http://wwwflag.wr.usgs.gov/USGSFlag/Space/clementine/images/albedo.simp750.jpeg
The size is 1080?720, so watch out if you can't deal with non-power-of-two sizes…
The north and south poles still have fixed shadows, but that's because the Moon's axial tilt is only about 1° w.r.t. the ecliptic, and we'll never see the Moon's poles illuminated from on high. The problem cannot be cured, unless nuclear explosions are used in large-scale flash photography.
You can see the fixed shadows give an odd effect over the poles if you speed up time and watch the Moon rotate around it's poles. The shadows always face in to the poles, rather than just away from the Sun… Fortunately, the area concerned is small and often out of the way, so we can live with it.
I also have an excellent result using this technique with Rhea, moon of Saturn, but only from a certain angle. It has bearings on the Voyager tour proposals... More later with images, I hope…
I think albedo maps tend to be presented with a high colour saturation, which I think is not realistic. I have seen the planets through telescopes and they actually have much paler looks to them, but it's a problem easily cured with Paint Shop Pro, e.g. I put a gamma correction of 2.24 on my Moon albedo map for extra brightness.
I think the Bump Map operates on the Texture Map, so if you removed it, the Bump Map has nothing to modify. Likely not a graphics card problem, then. However, I did notice that all a Bump Map does is adjust the colour/brightness of the Texture Map on the object. It does not change the shape of the object from a smooth sphere (as I found when trying to see if the crater Herschel on Mimas became indented). Bump Maps don't cause craters to become pits in a moon's limb. Nor do they make mountains cast long shadows…
Please do not expect full albedo maps for Uranian satellites until at least 2040 (think about it…:wink:).
Spiff.
Coming from the 'What Celestia should be' debate (http://www.shatters.net/forum/viewtopic.php?t=163), I'm restricted in input on matters of accuracy ('realism') until I sort out image submission. I left this issue of shadowing out of 'realism' though because it is an easy one to fix given Celestia's great flexibility in data files.
The problem arises because the texture map is an air-brushed 'topographic' map for lunar cartographers. It is designed to show all features illuminated at a constant angle by light from the east - a standard convention - so shadows are fixed. Currently, most other moons in Celestia suffer this problem, esp. Phobos and Saturnian moons.
My solution is to use an albedo map in place of a topographic map, but keeping the bump map. An albedo map is a map that shows suface colour and brightness without shadowing. They are built up as photomosaics from images where the Sun is high up in the sky over a planet or moon. Texture maps for Earth, Mars, all Gas Giants and the Galilean moons delivered with Celestia are already albedo maps. Only planets and moons that have been imaged with enough coverage to gain the data can have complete albedo maps.
I found an almost perfect albedo map for the moon at the USGS Clementine web site:
http://wwwflag.wr.usgs.gov/USGSFlag/Space/clementine/images/albedo.simp750.jpeg
The size is 1080?720, so watch out if you can't deal with non-power-of-two sizes…
The north and south poles still have fixed shadows, but that's because the Moon's axial tilt is only about 1° w.r.t. the ecliptic, and we'll never see the Moon's poles illuminated from on high. The problem cannot be cured, unless nuclear explosions are used in large-scale flash photography.
You can see the fixed shadows give an odd effect over the poles if you speed up time and watch the Moon rotate around it's poles. The shadows always face in to the poles, rather than just away from the Sun… Fortunately, the area concerned is small and often out of the way, so we can live with it.
I also have an excellent result using this technique with Rhea, moon of Saturn, but only from a certain angle. It has bearings on the Voyager tour proposals... More later with images, I hope…
I think albedo maps tend to be presented with a high colour saturation, which I think is not realistic. I have seen the planets through telescopes and they actually have much paler looks to them, but it's a problem easily cured with Paint Shop Pro, e.g. I put a gamma correction of 2.24 on my Moon albedo map for extra brightness.
I think the Bump Map operates on the Texture Map, so if you removed it, the Bump Map has nothing to modify. Likely not a graphics card problem, then. However, I did notice that all a Bump Map does is adjust the colour/brightness of the Texture Map on the object. It does not change the shape of the object from a smooth sphere (as I found when trying to see if the crater Herschel on Mimas became indented). Bump Maps don't cause craters to become pits in a moon's limb. Nor do they make mountains cast long shadows…
Please do not expect full albedo maps for Uranian satellites until at least 2040 (think about it…:wink:).
Spiff.
picili wrote:In fact I was using a single colour texture map and after close looking, it appears that only a very small part of the moon is bump mapped. That why I guess it's a graphic card problem. I have to investigate...
Can't wait to see your results.
Olivier.
Make sure you have pixel shading capabilities on that card...Best place to look is the users manual or the home page...Also make sure the options in the menu are checked under rendering...
Spiff, if you want image hosing try the site I use http://www.freewebspace.com... You will have to make a small web page in order to get away with it, but that shouldn't be too hard...And its free
I'm trying to teach the cavemen how to play scrabble, its uphill work. The only word they know is Uhh and they dont know how to spell it!
-
Matt McIrvin
- Posts: 312
- Joined: 04.03.2002
- With us: 22 years 8 months
Albedo maps and bump mapping
Spaceman Spiff said:
It can't be cured with literal accuracy, but it might not be quite so hard to get rid of the fixed shadows with a little artistry-- to airbrush them out, so to speak.
Right-- what you want is sometimes called a "displacement map" in computer graphics jargon. All a bump map does is adjust the surface orientation assumed in the lighting model at a given point. This works well for many orientations if the bumps are not too large, but the realism begins to break down when the surface is seen almost edge-on. Displacement maps adjust the displayed position of the point as well; they actually modify the computed geometry of the object. This makes them potentially much more computationally expensive.
The north and south poles still have fixed shadows, but that's because the Moon's axial tilt is only about 1° w.r.t. the ecliptic, and we'll never see the Moon's poles illuminated from on high. The problem cannot be cured, unless nuclear explosions are used in large-scale flash photography.
It can't be cured with literal accuracy, but it might not be quite so hard to get rid of the fixed shadows with a little artistry-- to airbrush them out, so to speak.
However, I did notice that all a Bump Map does is adjust the colour/brightness of the Texture Map on the object. It does not change the shape of the object from a smooth sphere (as I found when trying to see if the crater Herschel on Mimas became indented). Bump Maps don't cause craters to become pits in a moon's limb. Nor do they make mountains cast long shadows…
Right-- what you want is sometimes called a "displacement map" in computer graphics jargon. All a bump map does is adjust the surface orientation assumed in the lighting model at a given point. This works well for many orientations if the bumps are not too large, but the realism begins to break down when the surface is seen almost edge-on. Displacement maps adjust the displayed position of the point as well; they actually modify the computed geometry of the object. This makes them potentially much more computationally expensive.
-
chris
- Site Admin
- Posts: 4211
- Joined: 28.01.2002
- With us: 22 years 9 months
- Location: Seattle, Washington, USA
Utility of bump mapping for the moon
Anonymous wrote:There is always a side of the moon on which shadows seem to be cast the wrong way. That makes some of the craters look convex instead of concave.
Since I cannot get the moon map to be right in all situations, I tried blanking it and checked how it would look like only with bump mapping. I got a perfectly flat white moon. I tried to raise the BumpHeight parameter without luck; still a perfectly flat moon. Did anyone get better results ?
My graphic card : ATI Radeon.
Currently, bump mapping only works with GeForce cards . . . I do plan to implement a version that works with ATI cards using DOT3 texturing, but so far I haven't found the time to do this.
--Chris
-
chris
- Site Admin
- Posts: 4211
- Joined: 28.01.2002
- With us: 22 years 9 months
- Location: Seattle, Washington, USA
Can anyone find a higher resolution version of the Clementine albedo map? At most angles, it does look much better than the current map, especially with bump mapping applied. 1080x720 is an adequate resolution, but I'd like to have one that's at least 4096x2048 for graphics cards that support texture compression.
--Chris
--Chris
Bump mapping...
Currently, bump mapping only works with GeForce cards . . . I do plan to implement a version that works with ATI cards using DOT3 texturing, but so far I haven't found the time to do this.
--Chris
Chris - You mean I won't ever have bump mapping with my
TNT2 Viper 770 card? What a bummer that would be.
Vic
--Chris
Chris - You mean I won't ever have bump mapping with my
TNT2 Viper 770 card? What a bummer that would be.
Vic
-
chris
- Site Admin
- Posts: 4211
- Joined: 28.01.2002
- With us: 22 years 9 months
- Location: Seattle, Washington, USA
Bump mapping...
Vicware wrote:Chris - You mean I won't ever have bump mapping with my
TNT2 Viper 770 card? What a bummer that would be.
The TNT2 only supports embossed bump mapping, which might produce acceptable results but will not look as good as 'true' bumpmapping. Honestly, I'm not sure if I'll getting around to supporting bump mapping for older cards.
--Chris
Bump mapping...
The TNT2 only supports embossed bump mapping, which might produce acceptable results but will not look as good as 'true' bumpmapping. Honestly, I'm not sure if I'll getting around to supporting bump mapping for older cards.
--Chris
That's perfectly OK. If I was in your position, I would feel exactly
the same way. The Viper is already 2-3 years old. I just wanted to
confirm what the facts were.
Is Celestia coded to use embossed bump mapping if it can't do real
bump mapping? Or am I seeing no bump mapping with my card at all?
Maybe when you get time you can put up an updated list of
the current cards you know supports Celestia fully.
Vic
--Chris
That's perfectly OK. If I was in your position, I would feel exactly
the same way. The Viper is already 2-3 years old. I just wanted to
confirm what the facts were.
Is Celestia coded to use embossed bump mapping if it can't do real
bump mapping? Or am I seeing no bump mapping with my card at all?
Maybe when you get time you can put up an updated list of
the current cards you know supports Celestia fully.
Vic
chris wrote:Thanks to a link posted by Hank, I think I've got a 4096x2048 map now . . . An albedo map is just a surface texture where all the detail is due to actual variations in surface brightness rather than local lighting conditions.
--Chris
Are you adding this for future versions? i.e. having the albedoMap = "texture.jpg" or something to this effect usuable for all planets?
Also what would be cool as hell would be adding Spectural Maps...
I'm trying to teach the cavemen how to play scrabble, its uphill work. The only word they know is Uhh and they dont know how to spell it!