Building a shape model

The normal way (Bob's way) is described in a text file called "densify.txt". I have created a script to automate things in a standard way.

• Copy the directory "shape3/support3" and put it in the working directory and call it support

   cd <workingDir>
   cp -r ../shape3/support3 support

• Edit the buildShape.sh file
  • Change the suffix to H2-Test1-
  • Comment out (#) the line that runs dummer

    • dumber < tmpRun.txt | tee -a shape.log # normally used

  • Remove the comment (#) from the line
    • #cp SHAPEFILES/SHAPE.TXT SHAPEFILES/dumb$suffix # used only with initial base model Q=32
• Run the program

   buildShape.sh

• When it is done, it will have already run view_shape (via another script viewShape.sh)
• Look at the files view_01.jpg (01 to 10) and shape.gif

This will create several different resolution shape models, each with a factor of 2 higher resolution. The 512 is the highest resolution model, with 1.6 million vectors.

One you create the high resolution model, you can install it in the directory SHAPEFILES. This is done by removing the symbolic link to SHAPE.TXT and replacing it with the new file that you've created.

•   ln -s shapeH2-Test1-512 SHAPE.TXT

You can run register and load the shape model to see how the new shape model looks.

Note, this shape model does not have albedo in it. If you want to have that, then you need to change the program from densify to densifyA.

Bigmap

• Create a bigmap using the TAG site definition.

   bigmap < MTAG15.in

• Display the map

   showmap
   MTAG15
   convert MTAG15.pgm mtag15.jpg
   open mtag15.jpg

You can also view it using other programs

•  view_map_rgb
   MTAG15
   1 1 2 45
   convert view.ppm output1.jpg
  
   view_map_stereo
   MTAG15
   convert view.pgm output2.jpg

coverage

You can use coverage_p to show how well a region has been tiled.

•  map_coverage
   MTAG15
   0 .0012
   convert coverage_m.pgm coverage.jpg

Building maplets using bigmap

Previously, we created maplets with the initial topography from the shape model. Once we go beyond 75cm (for Bennu), then the bigmap will have the higher resolution, which will be needed. To do this, we "fill" the maplet with the data from the bigmap rather than the shape model.

Running lithos

• c
• TEST07
• m
• MTAG15
• 200 200
• .00075 49
• i
• TEST07
• y
• .5
• n
• e
• n
• 0 50 .2 .2 0 4
• b - fill with map
• y - list possibilities
• n - Auto pick
• MTAG15

Then just continue processing like normal. It is unlikely that you will need to align to a picture. You now have decent topography with albedo, so you should be able to just 'auto align' to get the images aligned -- that is option "1" and submenu "0" auto align

Resetting the albedo in lithos

If you have some topography that isn't good in your template, you may want to replace it. If you can load the data from another maplet, that works well. However, if you cannot, you can remove the template, setting it to 0. Then as you build a template, you are not fighting bad data.

Within lithos and the landmark is already loaded

• a - reset albedo menu
• y - reset slopes
• y - reset albedo

Then you can do several "build templates" (0 0 44)

The RECENT command in LITHOS

There are times when you need to re-enter the same landmark. There is a shortcut in lithos that lets you use the landmark that you were just using, RECENT. You can either type "RECENT", or "R" or "r".

This assumes that you've already worked with a landmark and typed in its full name. You can enter this in at the following menus.

• i - input menu
• o - overlap/limb menu

Overlaps Menu in LITHOS

Overlaps are used by SPC to help constrain a solution. Each maplet will have its edges overlapped by other maplets. SPC will identify which maplets should have overlaps, and then it will share information between them to reduce error.

Overlaps must be set in lithos. This is done in the 'o' menu. This is done as part of the normal scripts that you have been given.

Also in those scripts, they run the overlap menu that has an option to set the limbs. A maplet is set at a limb when any part of its surface is in an images' limb. What lithos does is reviews every image and identifies where the limb is in that image. Once found, it will see if the the maplet should be present. If so, it will record that image as a limb image for the maplet. More info: Attach map to maps or limbs

Steps in SPC (taken from the standard scripts)

 o - overlap menu
 r - use recent landmark
 y - clear out old data
 1 - automatically detect overlaps

 o - overlap menu
 r - use recent landmark
 n - do not remove old data (we want to keep what we just did)
 3 - search for limb images
 n - clear.  Do no remove old limb images
 1 3 5 - Search parameters.  

When you are done, lithos will not have a landmark loaded. Thus, it is useful to reload the landmark

 i 
 r
 n
 n

BIGMAP error

When you run bigmap it will generate two files

• SIGMAS.TXT -- a text file that lists the residual error. Each time you run it, it will add an entry. The first number is the maximum deviation. The second number is the average deviation.

• SIGMAS.pgm -- an image showing a pixel-by-pixel representation of the deviation. DN of 255 correlates to the maximum deviation number in SIGMAS.TXT

The residual deviation is computed for every pixel by taking the height value from every maplet that was used to determine the height. In bigmap, it will average those values. The residual (or sigma) is the standard deviation of those heights.