I just received the following from Shin-ichi Sobue at JAXA:I wish I could post some images but I'm having a bit of trouble with the Data Archive website -- got registered OK, but my searches seem to return database errors. And I couldn't seem to find my way to Terrain Camera data. Perhaps it's because my brain is somewhat mushy from a day of football and beer But I give up for the night and give you guys the link in the hopes you can figure out how to get pretty pictures out of this website!

I've attached the whole doc that Shin-ichi sent me, which contains a little more info about the data sets in the release.

--Emily

First thing I wanted was TOPO data - and managed to get through registration, data search - all worked. Got my email with an FTP link for the 400 meg data file.

Download 4kb/sec.

I'll try again later - I wonder if it'll be mirrored by the PDS themselves.

(100 - 150kb/sec now)

Are they going to release the actual HD camera data?

That's copyrighted by the broadcast company that built the camera I believe - so no. The release specifically states



I managed to buy an import of the blu-ray disk they did and it's worth it for the 15 or so minutes that's included, but they've taken hours and hours of footage since then.

The question for me is - what to do with a .sl2 file.

I'm registered but I haven't figured it out yet - trying to get TC images or orthomaps in a given area, I search successfully but get 'no data'.

Terrain Camera data as far as I understand it comes in the LISM dataset.

Phil

OK, getting somewhere... if you run the 3D Moon Java app you can view datasets on a globe, and rotate it to view any region. Works great for the altimetry, GRS, magnetometer etc. Then I display TC on that - just a few spots. Tycho is one, Mare Moscoviense another. So back to the database search. Select LISM datasets, search on Tycho coordinates, then choose TC orthomap, and bingo, lots of datasets.

Presumably the database will be populated with other TC scenes later.

Phil

EDIT: I've ordered a TC map scene (2 degrees by 2 degrees, on N rim of TYcho, Surveyor 7 area... I'll be able to FTP it soon. We'll see what we get.

Phil

I'm downloading the elevation data (at least I hope it's the right file - LALT_GGT_NUM.sl2). Should be interesting...

--- still struggling with the format. My image appears to be a 8192 x 8192 pixel 32 bit image. So far nothing is opening it. Any suggestions (I mean from somebody who has successfully opened one!)

Edit: Oh poop! I didn't see this before. Should have done my homework first:

About ".sl2"file format
The form of the file compresses tar. Please uncompresses and use.
(example) tar xvf dddd.sl2
dddd.sl2 : Acquisition file name


Phil

I'm glad you're slogging through this and not me!

Via Twitter it seems Mike Howard's been having some success with the topographic data.

Well - I managed to look at the resulting file - after a header it's lat, long, alt

I don't have anything I know of that can turn that into something usefull for me, such as a high bit png/tiff/tga etc etc.

I'm on it. If my cats will leave me alone for awhile.

It's 5760x2880 floating point values. What format would you ideally like?

P.S. For the images - I haven't looked at those, but the elevation data has PDS headers, I wonder if the images do too. Maybe IMG2PNG could be adapted? I'd volunteer, but I don't do Windows, or scripting languages. (Actually I think IMG2PNG is an .exe, IIRC.)

finally i can get ride of my "artists concept " topo map . that will be wroth while .

I'm going to have to go out to dinner pretty soon, but I should be able to produce a 5760x2880 bump map PNG, normal map PNG, and binary stream of floats within the next 24 hours, assuming the data is good. Unless somebody else can do it faster.

If these files have PDS headers followed by floating point data IMG2PNG might already be able to read them. If not, modifying it to do so is trivial for me - the most complex step is probably getting the data so I can test it ;-). Can't do it until about 20 hours from now though.

I have been waiting more than 10 years (!) for a nice, global DEM of the Moon so this is something that really interests me.

IMG2PNG can't do it as it is. The problem might be the header, but I'm not sure. A solution along these lines would be great. Any unzip-type app will extract the IMG file from the downloaded file, but I can't get anything else to open the IMG.

Phil

The best analogy I can come up with is the values in HRSC Dems - which a guy on HRSC kindly wrote an IDL routine to churn into something usefull



This data set is in km if i remember, so it would need to be multiplied by 1000 first - than then have the negative value of the lowest negative altitude added (i.e. if the lowest is -1580 metres, add 1580 to make them all positive values starting at 0)

To be honest, just a raw array of values normally can be convinced to work with ImageJ or Photoshop.

i just looked at LALT_GGT_NUM.TAB in ghex edit
see screen shot
Click to view attachment

on the right -- the 3 columns
top one


is looking like excel is going to get puled out
i will need to pull the middle value
first is lat.
second is the gray value
and the third is the long.

As text loaded straight into ImageJ - it's lat long alt - there's a header that describes it once unzipped (I used 7zip) - you may have it muddled somewhere.

Thanks to mhoward's processing and my running it through ImageJ - we have a 16bit Tif for those that like such things. It looks like the attached 1/4 size JPG and is found HERE - About 26 meg

Also, something we've been waiting for for a long time: a proper surface normal map for the Moon.

LALT surface normal map (22MB PNG)

enhanced LALT surface normal map (26MB PNG)

The first version is technically accurate AFAIK; the second is "enhanced" a bit (the normal vector x and y are magnified somewhat). Use whichever one works best. Attached is a quarter-sized version of the enhanced model.

These Celestia screenshots use mhoward's regular normal map:

Click to view attachmentClick to view attachment

Click to view attachmentClick to view attachment

My cellphone probably can't compete with stuff rendered on a full-blown computer, but an experimental version of my Moon Globe app looks better using this data, too.

for celestia you might want to run the 16bit tiff through nms to make a spherical corrected normal map

If somebody does that, I'd be curious to know if the result is different from mine. I'm glad to know about those tools, but I don't have time to look into them further now.

P.S. If anybody prefers raw 16-bit binary integer data over the 16-bit TIFF file, I've got that - just PM me let me know download it from here, while my bandwidth lasts. 5760x2880 16-bit signed integer values, values converted from km to m to retain precision.

nms (Normal Map Spherical) just applies a correction for the spherical geometry of a planet. The slope in x is increased by a factor of (1/cos(latitude)) to account for the fact that the map is horizontally compressed near the poles. If you wrote a tool especially to make normal maps for planets, it may already be doing this. If not, it's an easy correction to include.

--Chris

Thanks for producing this! What heights to the minimum and maximum pixel values represent in this tiff?

EDIT: Probably a dumb question. After opening the zip, I see that the pixel units are apparently meters.

--Chris

Thanks, Chris. I've got the sine of the latitude in my math, so I think I'm doing the same thing, maybe in a slightly different way. Still would be curious to know if the result comes out exactly the same.

Here are a few more screenshots:

Click to view attachmentClick to view attachment
Click to view attachment

And rotation movie: http://www.youtube.com/watch?v=arMmSCQiuic

Awesome movie.

Now that I'm working with the normal map, I notice it's pretty rough around the poles. I'm working on smoothing it out.

Having this DEM is awesome, for the first time I'm able to get lunar renders I'm happy with. Here is a test render. The data is rendered as a DEM (not a normal or bump map) meaning it is possible to spot some slight irregularities (mountains etc.) on the limb, especially in the south. It is rendered using data from the 16 bit integer file created by mhoward and shows the Moon as it might look as seen from the Earth:
Click to view attachment

And here's another one showing the Moon from a distance of 4000 km from the Moon's center. The field of view is 35 degrees:
Click to view attachment

At this close range it would be nice to have a slightly higher resolution DEM although this is not a major issue.

Another one, this time 6100 km from the Moon's center, FOV 35 degrees. The "subspacecraft" point is lat=-10, lon=270 degrees west. This is probably my best one so far:
Click to view attachment

Yet another one, this time from an altitude of 300 km and with a FOV of 45 degrees. This one is intended to show the limitations of the DEM, i.e. insufficient resolution for closeups like this one. It will be fun to do again in the future once the higher resolution LRO data arrives.
Click to view attachment

This is a lovely dataset to play with.

Outstanding! The DEM is indeed a wonderful product. I'm a bit concerned right now with what seems to be a very limited release of TC data. As far as I can see it is indeed limited to the small areas visible in the 3DMoon global viewer, but I assume more will be added soon.

One wee problem in 3DMoon - check out the RSAT gravity data - you can plot the Clementine or topography data first, then overlay an RSAT dataset, and flick it on and off to see how it lines up with topography... but it doesn't! They have accidentally flipped it 180 degrees in longitude. I expect it will be fixed soon.


Phil

I processed with the 16-bit height data with the nms tool, and I get a normal map that's rougher around the poles. You must be capping the spherical correction factor to some maximum value? My normal map is almost identical to yours except within a couple degrees of the poles. It sure does seems like the polar regions need some filtering--artifacts are apparent near the poles even when slopes aren't amplified by the spherical correction factor.

--Chris

i just made a normal with my hacked copy of nms ( mine uses unsigned 16 bit)
the zip is a png image
using my default settings and the default 16 bit tiff from above
5670x2880.png
http://www.zshare.net/download/6793328103b3d90b/

a 1024x512.png image
Click to view attachment

two screen shots
the first one is the NEW one at default settings
and the second is my OLD hand drawn height map
Click to view attachment
Click to view attachment

"probably my best so far" - yes, I'd say that deserves to be a Lunar Picture of the Day...

Phil

Bjorn - absolutely stunning pics, thank you!

Do you take requests? I could really do with a couple of nice pics of the crater "Eddington", for obvious reasons...

Bjor (and JohnVV!) amazing work! Seeing oblique Mare Crisium on the west as seen from Earth on the east! Amazing! Langrenus and Humboldt as round craters! Something you can't see from down here

Outstanding work, Bjorn!

This may be the first time a computer render I've seen be so realistic - in fact with a bit of fiddling around in Photoshop (I assume you didn't do the 2.2 gamma correction judging by contrast?) you could pass it along as a photograph to unsuspecting folks.

Click to view attachmentClick to view attachment

It does look quite photographic ! Congratulations for all the nice renders on this thread.

Yes - you are right. I think I was doing that for some reason when processing the MOLA data. (In my case it's a minimum value - apparently I've done mine differently somehow, but it's equivalent.)



Indeed. I've got a somewhat smoothed out version now that I'll post for people to use if they want to: LALT_normalmapV2.png (21.5MB PNG)

Others may be able to do a better job filtering it; this is sufficient for my purposes.

After adding the LALT normal map, I've been tweaking the lighting parameters in Moon Globe to try to make it look more realistic (something I've already spent a lot of time doing - it's not easy to get this to work on a first-generation iPhone). Here's a quick comparison:

Yes, these are probably the most realistic renders I have ever done. And as you suspected there's no 2.2 gamma correction - your versions are even more realistic.


Normally not but here's a quick and dirty render with Eddington visible. The "subspacecraft" point is 22°N, 95°W, the distance from the Moon's center is 6100 km and the FOV is 35°. The subsolar longitude is 15°W.

Click to view attachment

I should probably make a rotation movie of this thing...

Quick, oh yes, but dirty....? Stunning.

That's extraordinary, Bjorn.

Earlier today I updated the second version of my normal map (link above), to smooth over a few spots where there are glitches in the source data.

Finally figured out self shadowing terrain that I've been struggling with for a while... a couple of what I THINK are realistic renders, plus one with UBER-EXAGER-TERRAIN and a final one that's exagerated, but only to look a bit more realistic. It seems I have to turn up above and beyond what I'd expect the numbers to require for realistic terrain - perhaps because of the comparatively low level of detail in the terrain models.

And because the resolution isn't that good - a bit of foreshortening to get a look at Eddington for Stu

These Kaguya-derived images are absolutely stunning, great work everyone. This actually feels like a planetary encounter, with new images of a new world for us all to drool over...

And thanks for the Eddington pic Doug, much appreciated. I've some big talks coming up before Christmas and really want to be able to show off that crater, for obvious reasons :-)

I like the exaggerated one. Somehow on a globe it doesn't seem to be terribly out of place. I think it's useful for understanding the terrain and relative elevations and crater depths.

I wonder when we'll see the first program that accurately renders this topographical data so we can really plan our lunar observations...

Excellent renders, everyone!