very recently, the new LOLA
ltimeter) data of the LRO
became available at the NASA PDS archive
http://nssdc.gsfc.nasa.gov/nmc/experime ... 09-031A-04
http://pds-geosciences.wustl.edu/missio ... efault.htm
or perhaps simpler from here
In addition, a dedicated WEB site was opened, where Celestia-specific data are offered for download by the LOLA team, notably a 32k normalmap VT (virtual texture) set!
The VT's were obtained by the LOLA team with my Nmtools
, as available from this CM site:
At last, we can enjoy a beautiful hires normalmap for the Moon besides the recent one from Kaguya-LALT, although the sampling density is still somewhat low. Therefore at the higher VT levels, one can clearly see the orbital ground tracks of LRO (stripe pattern). Of course we expect this to improve quite soon.
Unfortunately, the situation about a good quality basemap of the Moon is still unsatisfactory. The oldish Celementine data
are full of nasty artefacts and also show a considerable non-linear pixel mismatch
with the recent Kaguya and LOLA normalmaps.
Fortunately John Van Vliet
went through a tedious remapping procedure recently and has now completed a nice matching base map!
Being uploaded to the Motherlode right now
I thought, it might be useful to initiate a surely ongoing LRO-LOLA discussion in this forum with a few neat shots based on these new data. For now, I just show the Moon's appearance with a 8k non-VT texture set
: the basemap is John VV's remapped 8k moon and the 8k normalmap
, I redid myself using the official 64 pixel/degree data <=> 22.5k
Since the 16 bit PDS binary .IMG format refers to (little-endian) non-signed (0..65535) rather than signed 16 bit integers (-32768..32767)
, some code adaptation of my 'nm2pow2' tool was necessary: The single command-line for generating the 8k normalmap then reads
nm2pow2 23040 < LDEM_64.IMG | nm2half 16384 | nms 1753.784 16384 2.5 > nmout8k.png
with the 16bit read method in 'nm2pow2' adapted to unsigned 16bit integer
input. Essentially, this amounts to re-declaring the row of 2 byte data from short
to unsigned short
and subtracting 0x8000 = 32768 to convert back to signed 16 bit for the remaining calculations.
But note :
By this conversion, the zero level for the altitude shifted by a certain amount as follows:
According to the accompanying LBL file,
radius = DN * scale-factor + offset
offset = nominal Moon radius of 1737400 m = 1737.4 km
scale factor = 0.5
and DN (= Digital Number) is in meters, ranging within 0..65565
To convert to signed 2 byte integers within (-32768.. 32767)
, I had to redefine DN -> DN - 32768 (i.e short C++ type
. Hence in order to retain the same radial distances, I have to redefine also
offset -> offset + scale factor * 32768 = 1753.784 km, i.e. the value I used above.
radius = DN * scale factor + offset = (DN - 32768) * scale factor + (offset + 32768 * scale factor)
where DN are the unsigned
2 byte input integers, that were internally converted. That hopefully explains the somewhat bigger Moon radius that is required in my 'nms' tool. I am planning to soon release a new version of the Nmtools accepting BOTH signed and unsigned 16 bit integer input.
Now, after these longish, yet hopefully somewhat instructive notes, let me display some nice shots obtained with only 8k non-tiled textures
A full view to start...
Throughout, click for bigger images!
The Schroedinger crater at a zoom level that is almost too much for 8k textures
Rather few craters with a familiar mountain range
Finally, some Celestia.Sci
images with colored, new galaxies for "dreaming"
A heavily cratered Moon with our neighbor LMC galaxy in the background:
The Moon with the beautiful orange elliptical M60
and overlapping blue NGC 4674
galaxy in the background
that look pretty similar in the SDSS photographic survey:
And last not least, the Moon with the (elusive) Sagittarius dwarf galaxy that is about to collide with our Milky Way
. See the peripheral cloud of globulars?
Of course, I have turned up the galaxy brightness, to make that dwarf more conspicuous...