SpacedInvader

Hmm.. I've been using the imgur /imgur format all this time thinking that was what was supposed to be native. I guess I'll start using the url version instead. As for the deformity test, I cranked it up to 1000000 as you suggested and it had absolutely no effect on the scaled space mesh when it really had an effect on the terrain. I know this for a fact because I ran into a mountain while on a 500km orbit, but based on the scaled space mesh, I was high above the surface. For the normal map process, I must be doing something wrong while creating it, because I've been following your instructions in the wiki as best as I could. I created the normal map, then selected the red, inverted it, created an alpha channel and copied the red to it, then selected the green, inverted it and copied that to red and also blue and then saved it. My maps still come out purplish... I will have a try with SSBump, but I'm not holding my breath.

I've been using the album embedding for a while, I just had a second image not in the album to share. I've read recently that some people use bump maps for rough terrain shaping and then normal maps for fine tuning it. That being said, the fact that the scaled space mesh is draped over the PQS makes this all the more confusing because I've got the game using the custom height map for the moon, so any PQS output should conform to that and not to the stock munar surface. As for showing the color map in the album, I included that to show that the detail was not coming from the color map, but rather the normal map, though the last picture I shared indicated that to me pretty well. I do like crazybump, but I'm just not in a place to drop $100 on a program I'll probably use for a month and never touch again. I'll have a look at SuperPNG, but again, it looks like GIMP is handling the PNG outputs correctly anyway, unless you think the 90° shift in lighting is still because of the PNG issues even with the change to GIMP?

Here are some pics of the effects I'm talking about with just a flat grey color map to make them easier to see: As you can see, with the custom normal and color in place, you can still make out the surface features of the stock body. That being said, I think we may be missing a map here. The level of detail that the Mun's map gets when you disable the custom normal map, combined with the face that you can still see features with it in place might indicate that they are using both a bump map AND a normal map for added detail. One thing I would point out that also might support this is that your earth map doesn't create any actual bumps that I can see even though it does correctly simulate slope lighting, so maybe there is another map needed to add height variation to the surface. Now, I think I should probably back up and ask exactly which settings I should be using to generate the normal map in the first place because it's entirely possible that I'm going wrong somewhere there and then that's why it's not working. I've been using the following settings: Add height to normal map Wrap 4 sample filter Min Z = 0 Scale = 100 Average RGB height source Alpha field unchanged Anyway, for the craters, yes this was with the planet editor, and as for Earth, I'll get started on looking for some decent data. EDIT: So I think I found something that may be part of the problem. When I go to close out my newly saved normal map with the alpha channel added, I get a message saying not all of the data was saved because the format doesn't support it. Then when I reload it, its got no alpha channel. Does PNG support alpha channels? EDIT2: Switching to GIMP and using it's normal map plugin got rid of the alpha channel truncation issue and significantly enhanced the surface appearance, but I still can't get rid of that stock Mun surface still bleeding through into the new normal map... EDIT3: Also, even with the corrected normal map, the light from the sun is still being rendered 90° east of where the sun is actually hitting the planet. EDIT4: I'm also starting to think that a scale of 100 might be a little much...

So, here's where my lack of photoshop skills is going to show up... When you describe inverting the R etc, does this step happen before or after the normal map has been generated by the Nvidia plugin? Also, I'm having an issue with the normal map for Mars for sure, and potentially the Moon while using these maps corrected for position, though it's hard for me to tell with the latter. Apparently, KSP is refusing to use the new normal maps I've been generating, instead happily wrapping a photoreal image of Mars over Duna's landscape. Is this the result of not yet having converted the normal maps using the process you describe in the Wiki, or could something else be going on here? So far I've been able to flatten them out, make them steeper, affect their ridge shape, and alter their coloration, but I can't seem to get them to shrink. That being said, it might lie in finding the right seed value as changing that definitely had an effect on the output size / shape, just nothing I've been able to predict as of yet. I'm happy to put some effort into Earth, I just didn't want to step on your toes after the work you'd already done. EDIT: So apparently, the normal map problem is a little more complicated than just using the wrong map for the planet. The game actually seems to be using both the stock and the new maps together. Thinking the new map wasn't being implemented at all, I commented out the color map line of the config to see how well the surface variations lined up with the stock surface and the end result was that the surface did match up with stock, but was very much smoothed out. I then went back and commented out the normal map lines and the result was a stock looking scaled space moon with much stronger features. I then went back to using custom normal and color and could easily find features like the canyons of the Mun imprinted in the Moon. I will try to do the normal map conversion tomorrow to see if that fixes this issue. EDIT2: Solution for Moon facing wrong way: initialRotation = 25

After a lot of thought, I think its better to use the Kaguya color map. This is partly because it's the only full coverage color image I've been able to find and so better represents the actual lunar coloration, but also partly because I don't think I have the photoshop-fu to properly get rid of the shadowing in all of the craters without making the thing look fairly disgusting. Does anyone disagree? @AndreyATGB: I'm still waiting for your permission to use these, but I'm also starting to question if your statement that you didn't modify them means that you don't really need to give me permission? Now, I think I'll be done with the improved topography and color / normal maps for the inner planets (excluding Earth) by the end of this week. I already have a good color map for Mars and only need to create a normal map, I have elevation and a color map for Venus, though the latter is going to need some touch up, and I'm in the process of acquiring the Messenger mission data for Mercury, which should have both elevation and color data for me to work with. As NathanKell has requested, I'll be offering both a high resolution option (8192x4096 for height and color with normal probably being fine at 2048x1024) and a lower resolution option (4096x2048 height map, 2048x1024 color, and 1024x512 normal). A disclaimer though, as I am not going to tweak any of the PQS settings beyond what is necessary to get them into the game and establish the correct elevation ranges. This is because I will be waiting to work on the higher level detail until Nathan can get the revamped loader in place for enhanced tweakability. That being said, these files will definitely offer much better terrain than what is currently available. A note on the Moon as well. I do plan on disabling the voronoi craters for the Moon when the new loader is finished, at least if and until I can find a way to drastically shrink the size of the output craters. This is because the output craters are quite large and do not exist in reality. The high resolution height map, at least in the 8192x4096 size has enough detail to generate craters of the size that the mod outputs, so they are both redundant and unrealistic. That being said, if I can find a way to shrink their diameter significantly, maybe we can use them to generate the multitude of very small craters that cover much of the Moon's surface. @NathanKell: Considering you've already put work into generating a pretty good terrain for Earth, I've held off on doing it in favor of the other planets, but it's also probably the easiest to find data on. The question I have for you is whether or not I should just let it be in favor of other planets, or if I should spend some time trying to improve it?

Just checking in to see if there has been any new movement on this? I've been using 2.2 for a while now with almost no crashes, so I think it's still ready for release, but I'm still really hoping for global texture management rather than just parts so I can run all of the high-res planet textures I've been working on for RSS without having to worry as much about the memory cap.

The thing is that KSP flips the maps horizontally and then needs them to be shifted 90° west for everything to line up. If you pay close attention to direction specific features like the horseshoe shaped crater that shown in both versions in my album that is shaped like a 'U' pointing at about 030°, I'm betting yours is flipped so that same feature would be pointing at about 330° instead. The 90° shift is one I had to figure out for Mars, but seems to hold true across all of the planets so far, and is specifically implemented so the correct latitude and longitude coordinates work within the game. I never tried your config, but my guess is that if you try to place yourself at a known landmark like the Apollo 15 landing site, you can't go there just by landing at the correct coordinates. This is exactly what I was dealing with when I had "lost" the Valles Marineris when working on Mars. It was there, but because of the map reversal and the 90° shift, it was not anywhere near where I thought it might be. So, back on the topic of what I did to your map, all I did was flip it horizontally and then move the left 25% and shift it to the right. The result is that, aside from features pointing the other way, it will look almost identical to your original, but have correct coordinates associated with it. As for the spherical thing, I do see the normal map's effects, but I'm not seeing that it looks significantly out of shape. I'm guessing that the overly bright poles are the result of multiple layers of slices of the planet getting lit by the sun at the same time, resulting in a spike in luminosity at those points. That being said, I'm still working with 5.5 as my dev build, so I haven't gotten to the new v6 just yet. Now that most of the available mods have been upgraded to 23.5 and the new RPL is out, I might just update both my dev build and my play build to the newer version. What has me concerned right now is the 90° shift in sun lighting for the scaled space map, though maybe that will be something that's solved with the newer version. EDIT: With your permission, I'll upload the corrected color and normal maps which you can use with the height map I shared a page or two ago, and hopefully that will correct your issues with the wrong side facing the wrong way.

Tracking up the data path from that site takes you to another artist's site with more maps, and you can go even up to sites like the International Jupiter Watch, all of which contain various amounts of imagery. So far I've found 6 different maps for Io and Europa, but they are not only extremely blurry, but very low resolution. My guess is that we may not be able to use much of this as is, but maybe with some processing and a little artistic license, we'll get something that looks good enough to view from orbit. As for licensing, most everything I've see is either free for public, non-commercial use, free for use with some attribution, or up to the original owner of the image to decide. I don't think RSS is in any way commercial, so I think we're fine for the most part, and I don't even think permissions would be hard to come by from image owners, though I'm not really sure just how much effort I want to put into tracking down a lot of people to ask if we can use their image... If it becomes too much, we may just use their work as a guide to create something non-photoreal, but at least real looking. I should also point out that the USGS has a lot of maps of the less commonly imaged bodies like Jupiter's moons which might be good to work with. EDIT: Can anyone with better photoshop-fu than me describe a way to get rid of the shadows on the moon's craters? I've been trying to use an inverted copy of the image selectively blended to "white out" the black shadows and then texturize and darken the resulting white marks to approximate a match with the light side of each crater. The result seems like it works somewhat, but I've noticed a few issues which I need to fix: 1: This one I don't get at all, but the scaled space color map has it's lighting offset by 90° to the East, so that the sun is actually shining on the day / night transition and the opposite day / night transition is squarely in the center of the dark side of the sphere. 2: The actual light recorded on the sun facing side of the craters is quite bright near the poles, so much so that it make the resulting globe look odd. I'm thinking I may not only need to push up the shadows, but tone down the whites a little to get a more even gray across the whole surface. 3: For some reason, when I'm trying to blend the two images together, I often end up with some weird overhangs near the edge of the map, leading to a clear wrap line on the finished globe. I'm really not sure what's going on here because I'm definitely working with two identically sized images, but one or the other keeps drifting off the edge somehow. So the last three images are using AndreyATGB's colormap, but corrected to the KSP requirements of being horizontally flipped and then shifted 90° West. There is some strange lighting at the poles for both color maps, and both have the same 90° East angle of incidence for the surface lighting. I didn't include the redone color map for the last three because I feel like I might need permission to re-release it... EDIT #2: I guess the question now is whether or not I should keep working on my darker version of the color map, or if AndreyATGB's corrected for surface shift is going to be good?

I guess that would make sense... I know some at least exist... I've seen more than a few on various PBS and other science shows in the past. It may even work to find someone else's decent interpretation of them and use that if possible.

Ok, if you give me a list of moons we're going to do, probably both for using stock planets and also PF planets, I'll start looking for good imagery for them. I think if I can find at least hemispherical shots, we'll be able to do a pretty good job of synthesizing some decent terrain.

Ok, we can put together two packages or at least offer high resolution maps for those who want them. One thing I will point out, however, is that we only have such detailed information for the four inner planets and the Moon, everything else is most likely just going to need lower resolution color and normal maps. This also brings up the question of how we want to handle the outer planet moons. Should we try to mimic what we know of their surface with custom height maps, or should we just stretch out the stock height maps to the approximate correct variation in elevation?

So I found a better version of the moon surface image. It still has the shadows, but most of the scan / stitch lines are gone except a couple near the poles and maybe at the copy-paste line where I move 90° to the other side... haven't gotten that far yet because the thing is 5.5GB currently, so I'm going to need to downsample the thing to 8192x4096. That being said, shadows alone might be easier to deal with than shadows and lots of lines... I'll put up the proper sized link tonight when I get home from work and have a chance to put some time into it.

Awe shucks! T'aint nuthin I'm just a little surprised that all of the GIS / Cartography skills I learned in college are getting put to use making a mod for a game

113MB?!?! That's a little crazy... But it prompted me to go and do what I should have done in the first place and have a look at the EarthNRM.png file already in RSS to see how big it is. Turns out, while the height map and color map are both 8192x4096, the normal map is only 2048x1024. Perhaps we should limit that file down to save RAM usage and keep it in line with what's already in place. As far as making the normal map, I'd tried to use that plugin, but always got the message that it needed to be an RGB file with three or four channels selected, so I just assumed that you couldn't do it with a gray scale heightmap. Yeah, I agree the image isn't the best, though I've yet to find a decent shadow free composite that wasn't obviously a composite with hard lines where the different layers exist. As for using the height map as a filter, it's not a bad idea. My thinking is to invert it so the craters are lighter and then maybe a lighter colors filter? The problem is going to be the scan / layer lines, which don't seem to want to go away, even with various attempts to blur or smooth them out. That being said, I'm still looking for a better source image.

I'm guessing it would need to be the case, though this isn't an area where I'm very well versed. Alternatively, you can walk me through the process and I can use my 16GB to take a whack at it. I'm also curious about how we can go about making the colormap using this image: It's not perfect as it has some stitch lines and the poles don't look all that great, but its from the same dataset as the heightmap, so it should line up perfectly once I correct for the 90° shift that KSP imparts. I'm guessing we don't want to just use it as is and that there is some degree of processing that needs to be done? I'm also still going through all of the output data from the LRO to see if there is a better image to work with.

I think it's probably good... the deformity value of 19905 is the elevation difference from low to high for the Moon and the heightmap resolution is as high as we're going to get. The actual output tiff from the mapping program I'm using is something like 46000px x 23000px, so it's good interpolated data rather than extrapolated from something smaller. EDIT: Plus, any noise functions we tweak in the future are not going to alter the overall height of the terrain by more than a couple hundred meters, so I think that won't affect the difference between the normal map and the heightmap once we get the generic loader and are able to really tweak the terrain.

Ok, so I've finished the 8192 x 4096 heightmap of the Moon, and I have the full image from the same satellite as well, though I'm really not impressed with it because it's monochrome and quite dark.I'm drawing a complete blank, however, on where to get a normal map. The surface roughness I mentioned earlier does seem like it might be that, but it's got lots of holes in it that the other two datasets just don't have for some reason. Googling really only produced some links to premade normal maps and also links to creating them in photoshop. Any thoughts about what to do here? EDIT: A shot from a random point on the Moon with the new map: EDIT: Here is the heightmap and here is the code to put into the cfg: PQSMod_VertexHeightMap { heightMapDeformity = 19905.0 //7500 heightMap = GameData/RealSolarSystem/Plugins/PluginData/MoonHeight.png }

I've found the Lunar Reconnaissance Orbiter data which is up to 128px / deg for elevation (way, way more than we can work with) and includes a camera which has global wide angle coverage. I'm currently downsizing the elevation data to 8192 x 4096 and getting the full coverage visual image as well. I'm not sure about normal map, but would that be the surface roughness map? The LRO also provides that...

I'll see what I can do... I'm sure that extreme resolution elevation data of the Moon have to exist somewhere...

BTW, does the Moon still need to be done? I remember reading something about 40 pages ago about the Moon being done in one way or another, but I'm not sure if it needed a heightmap and terrain improvements still?

So producing a practice heightmap is going to take longer than I thought. That being said, does anyone have any ideas / requests as to what simple terrain features should be in there? I've already got some reference hills to help with setting up heightmap deformity properly, as well as a patch of rolling hills and some cliffs, but I'm not sure just what else to put there. Thoughts? Also, I've started working on the Venus heightmap.

You know, the funny / sad thing, is that I actually looked for it at both 180° off location (0° central meridian instead of 180°) AND 90° in the wrong direction (as if it were just flipped horizontally), but didn't even think to look on the fourth compass point... I was really starting to get seriously frustrated at the game's apparent inability to do what I wanted it to do, and all along the problem was firmly sitting between the chair and the keyboard. Anyway, tonight I'm going to produce a practice heightmap with various slopes and simple terrain features so I can start getting a better handle on the PQS functions. I'll share it here of course and maybe we can start putting together some documentation for the PQS system for the community...

Nevermind, I'm doing it tonight... Just copy and paste this code into your RealSolarSystem.cfg file PQS { Duna { maxLevel = 14 maxQuadLenghtsPerFrame = 0.8 //0.03 -- and yes, typo is correct. PQSMod_VertexHeightMap // doubles { heightMapOffset = 0.0 //-2000.0 heightMapDeformity = 29457.0 //15600.0 //7000 // 5000 heightMap = GameData/RealSolarSystem/Plugins/PluginData/MarsHeight.png } } } And then place this file in your GameData/RealSolarSystem/Plugins/PluginData directory. You know, all this tinkering with Mars has made me want to go back and read Kim Stanley Robinson's Mars trilogy again after nearly 20 years... I guess we aren't really done yet, though, considering we still have to figure out some nice noise settings, new ground textures / coloration, and a scaled space mesh. And then we have the rest of the solar system to figure out...

Ok, I have the corrected heightmap finished. I'll be putting it up here sometime tomorrow for everyone to play with.

Found it! So it was completely in the wrong longitude and that's why I couldn't find it... It's supposed to be at 72.5°W 5°S, but I found it at 162°E. It's definitely flipped horizontally, but I'm not sure if there is also an issue with my MOLA export settings that might have shifted its position even more. Will let you know what I find, but for now, suffice it to say that I think this version of the map really only needs some fine tuning of the noise to make it look nicer and more realistic. EDIT: For the record, those pictures are taken with all stock PQS in place.