Hacker News Reader

Comment by JKCalhoun | Hacker News Reader

I recently took an old-school "voxel" algorithm and munged some NASA moon terrain data to try to create a game.

I really have enjoyed learning about the Moon's geography just from flying around the terrain, picking locations for "lunar bases" and such. Fascinating cracks and valleys, grooves, a straight mountain range — in general things I was not expecting on a planetoid without water/erosion.

It really is stark how different the far side of the Moon is from the near side. Insanely pock-marked with so many craters.... I assumed there was an Earth component to that in the same way the Earth has locked the Moon into an orbit/rotation that is always facing us.

I'm become very keen to learn more about the Moon now.

View on HN

Can you talk more about the moon data munging?

I just started some Blender tutorials and the first thing I want to try once I get some tutorials under my belt import dats from the Moon/Mars.

Sure.

For the Moon the (images) data seems to be bucketed with regard to pixels-per-degree (PPD), the more coarse data set being 64ppd data (finer resolution for the Moon goes at least as high as 512ppd that I have found). Sadly, my workflow (image editors, and Blender as you mentioned) are already straining with the 64ppd data-set so I did not try to go higher resolution. It means though that my Moon is wildly small — the lunar bases that seem to be a paltry 7x7 or 9x9 voxels are in fact the size of a small town.

I began with NASA's "CGI Moon Kit" [1] which, as I say, is 64ppd. There is both a "color data" image and an "elevation data" image (where the greyscale value of the image represents elevation). Polar data was garbage (and would look wildly distorted in my tile-based voxel algorithm so I confess that I tossed it — cropped everything outside 60-degrees latitude basically - I just limit fuel in the game to keep pesky kids from trying to get too deep into the polar regions).

Even at 64ppd and poles cropped, I still ended up with two images that were 23040 × 11264 pixels in size. This turned out to be a challenge for most of the image editing apps I have. Blender handled it (but barely).

With both color + elevation images I followed an online tutorial [2] meant for cartographers wanting to use Blender for relief shading. NASA's moon map had no shadows (probably a good thing). I was not going to try to real-time shade/light the voxel terrain so I settled on "burning in" the shadows with Blender. While my MacBook ran some 8 hours or so (over night) to generate the relief-shaded lunar map, the result was quite nice. As a result, it is near sunset everywhere on my Moon, ha ha.

With a relief-shaded image representing the color and a same-sized image representing the elevation, I wrote a custom app to read both files into memory and then generate the correct number of 512 x 512 tiles that contain both the elevation data and RGB data in a single raw file. Now I have 45 x 22 raw tiles covering the entire 64ppd Moon. The voxel algorithm uses simple math to figure out which tile a given voxel is to be found, reads the elevation/color data.

In hindsight, I could have output PNG files or something with alpha channel (where alpha would be interpreted as elevation). As it turned out, the Moon color was so close to greyscale that I ended up treating it as such and used only the green channel when creating my Moon tiles (so the tile data has only a single 8-bit value for elevation and a single 8-bit value for "color" per voxel — cut the raw tile size in half).

Still wrapping up the game [3]. I'll put the sources up on Github end of the month, I think.

[1] https://svs.gsfc.nasa.gov/4720

[2] https://somethingaboutmaps.wordpress.com/blender-relief-tuto...

[2] https://mooncraft2000.com

Fascinating. Are there upcoming plans from any space agency to improve that 64PPD with upcoming or existing satellites or Earth-based capture of moon topography?

There are 512ppd data sets [1] already. As I say, I found them unwieldy with my workflow. A new workflow would be required to process these into "game tiles".

[1] https://astrogeology.usgs.gov/search/map/Moon/LRO/LOLA/Lunar...

Whoops. I missed that you mentioned higher PPD is available. If your workflow was improved to accommodate it, I presume it would produce more, for lack of a better phrase, life-like terrain? Would a much higher PPD be required to reach a threshold to improve the experience. And are there plans coming to get finer detail than 512PPD?

It occurred to me that, beyond 512ppd, you could "synthesize" higher-res detail. I don't know how much of a cop-out that is though.

There may be >512ppd data though. IANAS (I am not a selenologist).

I did some cropping of the LOLA dataset to a .raw that Unity3D could understand as terrain input: https://github.com/amb/SplitTIFF ... a couple of years ago

Happy to hear any better alternatives for the task.

That is very cool. I starred your repo, if I try to up the PPD I'll have a closer look.

I made it to Ptolemaeus depot without crashing! And now Parry. Though it didn't register by landing at Cop, and I then I tried moving over a craft length and I crashed.

Fun game! It would be nice if the altimeter had a high res mode for the last few meters.

I either watch out the window the last bit and watch the base lights to judge the last few meters or I just watch the vertical velocity nixie (numeric display) and feather the throttle to keep Vert V. close to zero — eventually you touch down gently.

Thanks for the feedback. Hopefully I'll get more of that (and likely your request will come up again as well).

Becouse of the orbital lock it’s interesting to think of it like a shield for earth. I imagine a lot of those craters were by asteroids that would have hit earth, esp the perfectly round ones in the center. The oblong ones or ones towards the edges might have come at an angle that wouldn’t have hit earth. Seems like a fascinating way to study earths impact history if we end up exploring the dark side of the moon in more detail.

I also am fascinated by the idea of Nemesis [1] — an infrequent event that causes mass-impact throughout our solar system.

[1] https://en.wikipedia.org/wiki/Nemesis_(hypothetical_star)

Do we have an idea of what percentage of impacts were prevented this way? What likelihood is there that the moon saved humanity (more than once?) since the time that we descended from the trees?

I imagine once you figure out which craters correspond to ones that came in at an angle which would have hit earth and compare that to all the impacts you have had on earth one could figure it out.

The thing is that on the moon the craters remain visible for a much longer period of time while on earth they get covered up by plate shifts, erosion, ocean impacts, etc. So instead it might be a way to calculate how many asteroids likely have hit earth in general throughout history whether we have found the craters on earth or not. Then you could calculate based on the size of the craters on the moon which ones would have been earth catastrophic events (taking into account any diminishing impact energy due to our atmosphere). Which would tell us how many the moon saved the earth from. In terms of how many since human ancestors were evolving we would have to then date earth catastrophic craters on the moon. I'm not sure if this can be done through photographs or if we would have to get surface samples.

Then comparing that number to the fraction of earth covered/protected by the moon to the total area on earth we would have a decent idea of how many of these events happened one earth even if we haven't discovered the evidence for them on earth.

I believe consensus is that the moon will eventually drift away from the Earth. Do such impacts, predominantly on the far side of the moon, apply enough force to change that prediction? Are such projections of whether the moon leaves Earth orbit (or instead collides with Earth) accounting for this?

The moon is drifting away from Earth at a rate of about four centimeters per year [1]. Inbound impacts on the far side won't significantly affect this because of the way orbital mechanics works. To lower an orbital altitude, you need thrust opposing the orbital vector. (This is why spacecraft orient the way they do, with engines pointing in the direction of travel, for retrofire.) Thrusting toward or away from the orbital center of gravity (in this case Earth) can affect the circularity of the orbit, but that's all. In the case of the moon intercepting an impactor that'd otherwise strike Earth, the thrust vector is necessarily toward Earth, so the effect would be to render the moon's orbit infinitesimally more elliptical, with the increase along the major axis balanced by an equally minimal decrease along the minor axis. (To the extent the force vector of the impact is displaced from the axis through the moon's own center of gravity, that will affect the spin of the moon, but not its orbit.)

According to the linked article, it will take about fifteen billion years for the moon to escape Earth orbit entirely. I haven't checked, but I believe that's considerably longer than it is expected to take for the Sun to expand as a red giant and render Earth uninhabitable. Certainly more wondrous than either event would be for anything then living to remember having once had distant ancestors who called themselves "human"...

[1] https://www.google.com/amp/s/public.nrao.edu/ask/what-happen...

It's compelling to me that the Earth and its moon are so tightly linked that their relationship is projected to outlast a life-providing sun. I believe this is moreso than any other moon-planet pairing in the solar system? Given what we know of incoming asteroids and comets, the probability of their impacts hitting with enough force at a trajectory you describe with enough significance to change the moon's orbit is expected to be low . And we're relatively confident of that?

I don't have numbers or references here, but my intuition from a lifetime of space nerding is that, if the moon were struck with enough retrograde force to materially lower its orbit, in the best case that impact would liberate debris of mass and vector such that our first concern would be planetwide disaster response. (In the worst case, it might shatter the moon entirely and distribute a significant fraction of its total mass across Earth's surface; in such a scenario, the survival of eukaryotic life anywhere on Earth qualifies as a million-to-one-shot win.)

But the odds of such a collision I suspect are significantly lower even than of a similarly sized impactor striking Earth. Granted that the moon helps a great deal in sweeping space near Earth of objects that might otherwise enter our atmosphere, but something with the kind of kinetic energy it would take to produce these outcomes isn't going to spend long enough within the gravitational influence of the Earth-Moon system to be significantly affected before impact. It'd be like trying to swerve an ICBM warhead by blowing really hard on it, and I choose that simile with care - if this ever does happen, I think it might constitute a clear indication that someone somewhere in the universe really hates us!

Reply on news.ycombinator.com