It is currently Wed, 22-11-17, 20:10 GMT

All times are UTC




Post new topic Reply to topic  [ 13 posts ] 
Author Message
PostPosted: Mon, 27-02-12, 15:38 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4510
Location: Hamburg, Germany
Hi all,

after lots of work (and frustration) with the two recently available hires surface maps of the Moon, I felt it may be useful to write up a résumé of my detailed comparisons of these two scientific imaging data sets.

The title of this threat anticipates my personal conclusions...

The maps to be compared are:


For specific comparisons, I use here just 16k versions. These exhibit already plenty of detail, while this image size can still be handled without big delays in GIMP, which has considerable advantages.

For Celestia(.Sci) applications, a crucial usability criterion is a good alignment of the surface map in question with the official

LRO-LOLA normalmap VT set (32k)
http://www.celestiamotherlode.net/catal ... on_id=1549


Chang'e 2 map: PROs
================
  • The Sun is at a high angle in the Chinese imaging, so shadows are essentially missing. This makes the map perfectly suited for a dynamical shadow rendering via a normalmap!
  • The vertical stripes throughout the dark central mare areas are considerably weaker than in case of the LRO-WAC map (see image below). De-striping (ISIS3) can be successful without too much deterioration of map quality.

Chang'e 2 map: CONs
================
  • The biggest drawback are the disturbing alignment problems with the
    LRO-LOLA normal map! There are several issues that I have carefully analyzed:

    The aspect ratio of the original Chinese map is NOT 2:1. After close examination I was able to locate the underlying reason. It demonstrates that the assembly quality of the Chang'e 2 images into a simple cylindrical map leaves much to be desired:

    After scaling the original 218527 x 109166 (.jp2) map down to a standard 16k texture height of 8192 pixel with the original aspect ratio of 218527 / 109166, the corresponding width becomes 16399 pixel, i.e 15 pixel larger than the canonical 16384 pixel, corresponding to a proper width : height ratio of 2:1.

    I then was able to spot the place where these extra 15 pixel went:

    Image

    In this image I zoomed-in on part of the vertical line, along which the east and west ends of the simple cylindrical map are joined in 3d rendering. While the joining is much better hidden than in case of the LRO-WAC map, you see clearly a redoubling of features along a vertical line in the center of the original 16399 x 8192 map on the left. This is exemplified by the two orange arrows pointing to TWO tiny craters instead of only one.

    Their distance is precisely 15 pixels!

    AHA...Bingo.

    The zoomed image on the right then shows how the image looks after cropping this excess strip of 15 pixels from the east end of the simple cylindrical map. The corresponding texture now has precisely 16384 x 8192 pixels. After adding some optimal global offset to readjust the global alignment optimally, there remained still various local areas with an intolerable misalignment w.r.to the LRO-LOLA normal map.

    Sigh....

LRO-WAC map: PROs
=================
  • The LRO-WAC surface map matches perfectly with the LRO-LOLA normalmap
    (no surprise ;-))

LRO-WAC map: CONs
=================
  • The relatively low position of the sun in the LRO-WAC map leads to dark fixed shadows in many craters, and notably to unpleasant artefacts, when combined with dynamical shadow rendering through the LRO-LOLA normalmap!

    Here is an explicit illustration:

    First, with the normalmap deactivated, let us look at the (disturbing) line where the two ends of the simple cylindrical LRO-WAC surface map were joined in Celestia's 3D rendering:

    Image

    You clearly note that the fixed illumination of the crater rims is in opposite directions left and right of that joining line! To the right of the line, the crater illumination is towards the left, while on the left of the line the crater illumination is towards the right. Without a normalmap this frequently leads to visual effects like "inverted craters" looking like mountains...

    Now, with the LRO-LOLA normalmap activated, things get even worse, since we note a really disturbing artefact: A sudden drop in brightness of a whole surface strip to the right of the joining line! The normalmap now has placed all shadows in the same direction, which implies that it placed shadows over the previously illuminated areas on the right. Hence the dark overall impression.

    Image
  • The vertical stripes throughout the dark mare areas are considerably more disturbing and more resistant to de-striping by a combination of lowpass and highpass filters (ISIS3), than those of the Chinese'e 2 map. The reason is that the stripes have irregular widths and sharp boundaries. They can certainly be eliminated but hardly without affecting the crisp oveall imaging quality of the original.

    [click on image by all means!]
    Image


Personal Verdict:
=============

After much frustrating previous work with artefacts and misalignment of the early Celementine surface maps, things have certainly improved with these two recent maps of our Moon. BUT, altogether, BOTH maps still leave a lot to be desired, at least as concerns their application in Celestia(.Sci).

No doubt, a number of the emphasized map deficiencies may be eliminated with an excessive amount of image manipulation work. John van Vliet's 64k VT maps represent such an example.

But unfortunately the crisp appearance of the original maps is lost by the required large number of correcting actions... In my view, Moon maps exceeding 16k - 32k are barely worthwhile at this time for rendering in Celestia(.Sci).

Fridger


Top
 Profile  
 
 Post subject: re
PostPosted: Wed, 18-04-12, 9:12 GMT 
Offline
User avatar

Joined: Tue, 04-09-07, 21:55 GMT
Posts: 773
Location: N 42.38846 W 83.45456
just looking over the site , it has been a while

nice post

the shadows ARE A bleep-e-d bleep

hopefully that can be solved

the stripes ??? yes / no / maybe ???

the destripe tool in isis 3.3 is working very well on the old Venus data and that is VERY stripped , even more than the Moon .


Top
 Profile  
 
 Post subject: Re: re
PostPosted: Wed, 18-04-12, 17:24 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4510
Location: Hamburg, Germany
John Van Vliet wrote:
just looking over the site , it has been a while

nice post

the shadows ARE A bleep-e-d bleep

hopefully that can be solved

the stripes ??? yes / no / maybe ???

the destripe tool in isis 3.3 is working very well on the old Venus data and that is VERY stripped , even more than the Moon .


I also use the Isis3.3 destriper and have pretty much experience with it. There are many different types of stripes, some vanish with just a single lowpass & highpass run. Others need a long list of such combinations:

Here is a nastier ISIS 2 script from the isis site as an illustration:

Code:
#!/bin/sh
dstripe from=$1 to=d1 mode=vertical vlnl=301 vlns=1 vhnl=1 vhns=3
dstripe from=d1 to=d2 mode=vertical vlnl=701 vlns=1  vhnl=1   vhns=15
/bin/rm d1.cub
dstripe from=d2 to=d3 mode=vertical vlnl=701 vlns=1  vhnl=1   vhns=25
/bin/rm d2.cub
dstripe from=d3 to=d4 mode=vertical vlnl=7011 vlns=1  vhnl=1  vhns=51
/bin/rm d3.cub
dstripe from=d4 to=d5 mode=vertical vlnl=4515 vlns=1 vhnl=1  vhns=151
/bin/rm d4.cub
dstripe from=d5 to=d6 mode=vertical vlnl=8191 vlns=1 vhnl=1 vhns=501
/bin/rm d5.cub
dstripe from=d6 to=d7 mode=vertical vlnl=8191 vlns=1 vhnl=1 vhns=1403
/bin/rm d6.cub


There is usually a high price to pay, if many destripe iterations are required: image contrast decreases, which I HATE!


Unfortunately the LRO-WAC stripes are really resistant.


Fridger


Top
 Profile  
 
 Post subject:
PostPosted: Tue, 01-05-12, 21:14 GMT 
Offline
User avatar

Joined: Thu, 25-10-07, 15:20 GMT
Posts: 992
Location: NE PA, USA
Hi Fridger,

I've been looking into this the last couple of days and was amazed at how many things changed. The Planetary Data System (PDS) for one. I got to educate myself all over again on how to use it. I think that this is your answer though for getting a really good moon texture. The LRO WAC map was created from images at low sun angles as you stated above. But there are probably millions of images taken that were not used in the mosaic.

The PDS should let you filter your search for images with similar, more pleasing, sun angles but it is tedious. There are also codes/map coordinates in the file name that should give you the info needed to "fix" certain bad areas with different images. Again tedious. The one thing I didn't see was a filter to limit lat/lon. Only time. The PDS used to have that feature. At least for Mars it did. EDIT Here is the tool I was referring to and trying to find. It may help. http://pds-imaging.jpl.nasa.gov/search/ ... uickSearch Just got to click on the LRO link to the left.

I never did look closely at these maps to see how much work they really need. Still running a laptop with no frills. :? Puts a limit on what I can do.

But I've been thinking of rebuilding a new computer, re educate myself on how all this software works and try to create my own mosaic from the "raw" or calibrated data library. Mars and the Moon. I think it would be better to start from scratch. :wink: Remember how that spec map that was published for the Blue Marble lined up with the water ways? :shock:

cartrite


Top
 Profile  
 
 Post subject:
PostPosted: Tue, 01-05-12, 23:02 GMT 
Offline
User avatar

Joined: Thu, 25-10-07, 15:20 GMT
Posts: 992
Location: NE PA, USA
Hi Fridger,
I did a quick search on LROC from the link above with an incidence angle set at min = 0 and max = 2. I got a count of 7646 images. I didn't look at them yet but.......... That's a lot of images. :wink: At the very least there should be enough to repair the original mosaic to suit Celestia's texture needs. Hopefully they will be a good enough quality to do a whole mosaic at those incidence angles. That may be enough to get me motivated to start this :D
A little more time will tell.
cartrite


Top
 Profile  
 
 Post subject: Re: re
PostPosted: Tue, 01-05-12, 23:31 GMT 
Offline
User avatar

Joined: Thu, 25-10-07, 15:20 GMT
Posts: 992
Location: NE PA, USA
John Van Vliet wrote:
the destripe tool in isis 3.3 is working very well on the old Venus data and that is VERY stripped , even more than the Moon .


I think I may be missing something here :? From what I remember, the destripe tool was meant for images. I thought all the Venus data was topographic. I think the destripe tool would degrade the data since it's not really an image but height data/dem.

cartrite


Top
 Profile  
 
 Post subject:
PostPosted: Wed, 02-05-12, 13:45 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4510
Location: Hamburg, Germany
cartrite wrote:
Hi Fridger,
I did a quick search on LROC from the link above with an incidence angle set at min = 0 and max = 2. I got a count of 7646 images. I didn't look at them yet but.......... That's a lot of images. :wink: At the very least there should be enough to repair the original mosaic to suit Celestia's texture needs. Hopefully they will be a good enough quality to do a whole mosaic at those incidence angles. That may be enough to get me motivated to start this :D
A little more time will tell.
cartrite


Hi Steve,

did you overlook my last respective thread? See what I did there and how nice the 16k result looks!

http://forum.celestialmatters.org/viewtopic.php?t=428
http://forum.celestialmatters.org/viewt ... 28&start=2

Fridger


Top
 Profile  
 
 Post subject:
PostPosted: Wed, 02-05-12, 21:47 GMT 
Offline
User avatar

Joined: Thu, 25-10-07, 15:20 GMT
Posts: 992
Location: NE PA, USA
t00fri wrote:
cartrite wrote:
Hi Fridger,
I did a quick search on LROC from the link above with an incidence angle set at min = 0 and max = 2. I got a count of 7646 images. I didn't look at them yet but.......... That's a lot of images. :wink: At the very least there should be enough to repair the original mosaic to suit Celestia's texture needs. Hopefully they will be a good enough quality to do a whole mosaic at those incidence angles. That may be enough to get me motivated to start this :D
A little more time will tell.
cartrite


Hi Steve,

did you overlook my last respective thread? See what I did there and how nice the 16k result looks!

http://forum.celestialmatters.org/viewtopic.php?t=428
http://forum.celestialmatters.org/viewt ... 28&start=2

Fridger


I just had a look. Really nice. And I guess you know tedious!!!!! :wink:
I never looked at the dates from your threads. :? I thought that there was still an alignment problem. I see that the bigger size has it's problems. I had that problem a few years ago when I was trying to create a cloud map with the gimp.

I still want to try to process my own map with raw image data. Just not ready yet. Need to get back to Linux!!!!!!!!!!!!!!!!! Stuck with Win 7 for now. Not to complain but I Can't run ISIS with Windows. :wink:

cartrite


Top
 Profile  
 
 Post subject:
PostPosted: Wed, 02-05-12, 22:25 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4510
Location: Hamburg, Germany
Steve,

the bigger size (say 32k) would require 4*77 = 308 screens (1600 x1200) to align by hand! This kind of alignment work is better done manually. Moreover, coloration options as I described require GIMP operation and anything >= 32k is getting very slow...

Good luck with your project!

Fridger


Top
 Profile  
 
 Post subject:
PostPosted: Wed, 02-05-12, 23:19 GMT 
Offline
User avatar

Joined: Thu, 25-10-07, 15:20 GMT
Posts: 992
Location: NE PA, USA
Hi Fridger,
I'm not planning on realigning anything. I am planning to construct a new mosaic from raw images using the LOLA file as a shape model with ISIS. This should avoid any alignment problems. I'm also planning to use only images that have incidence angles approaching 0. This should turn out like the Change map. My first step will be looking at the data. Thanks for the Good Luck. I'll need it. :wink:
cartrite


Top
 Profile  
 
 Post subject:
PostPosted: Thu, 03-05-12, 9:27 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4510
Location: Hamburg, Germany
cartrite wrote:
Hi Fridger,
I'm not planning on realigning anything. I am planning to construct a new mosaic from raw images using the LOLA file as a shape model with ISIS. This should avoid any alignment problems. I'm also planning to use only images that have incidence angles approaching 0. This should turn out like the Change map. My first step will be looking at the data. Thanks for the Good Luck. I'll need it. :wink:
cartrite


Steve,

I did understand your plans, but I referred to this sentence of yours:

cartrite wrote:
... I see that the bigger size has it's problems.


albeit not being sure what you really meant ;-)

Fridger


Top
 Profile  
 
 Post subject:
PostPosted: Thu, 03-05-12, 10:04 GMT 
Offline
User avatar

Joined: Thu, 25-10-07, 15:20 GMT
Posts: 992
Location: NE PA, USA
Oh,
I was referring to using the Gimp with very large files. Difficult to say the least.

cartrite


Top
 Profile  
 
 Post subject: re
PostPosted: Sat, 16-06-12, 7:57 GMT 
Offline
User avatar

Joined: Tue, 04-09-07, 21:55 GMT
Posts: 773
Location: N 42.38846 W 83.45456
Quote:
Oh,
I was referring to using the Gimp with very large files. Difficult to say the least.

you can but you need to cut it up
for a 64 k or a 128 k map ,tile it to a 8x4 grid or 16 x 8


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 13 posts ] 

All times are UTC


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group