Converter for 3d mesh "VST" (ViSTa) files into modern 3d mesh files.
Created to reconstruct the 3d model of "Troy sandtrap", the final location of NASA Spirit/MER2/MER-a rover
(Image credits: https://www.planetary.org/space-images/spirit-a-monument-to-exploration)
3 versions of the converter are under development:
- VST-converter.html: raw javascript; developed stopped at "patch" level and continued on VST-converter-kaitai.html
- VST-converter-kaitai.html: based on VST parsing library created with kaitai-struct; requires VST.js in same folder; this file was automatically created by kaitai-struct using my VST.ksy file. Such file can also be used to automatically create a VST library for other languages (java, c, c++. Perl, PHP, Python...)
- x3d-viewer:based on vst2x3d cpp source, mixed with above versions, enhanced with massive help from AI (Openai ChatGPT, Claude Sonnet and others)
- others: other tools found around on forum or repositories
- VST2OBJ: converts to Wavefront .obj format; DOS executable; no source code
- VST2X3D: converts to X3D format; DOS execxutable; source code available
NASA used for Spirit and Opportunity rovers (MER - Mars Exploration Rovers) some peculiar file formats to store 3d reconstructions of terrain:
- .ht : Heightmap: The file contains 1 to 3 bands, each band representing a depth map; first band contains raw values; second band contains also interpolated values; 3rd band, if present, contains estimated errors. Example folder.
- .pfb: "Performer Fast Binary", used by "IRIS Performer" or "OpenGL Performer" from Silicon Graphics (www.sgi.com), now dead; format details unknown. Example folder . See below for further details.
- .iv: Silicon Graphics "Iris Inventor" format; not officially documented, but here there is C source of ivconv, a program able to convert it to other formats, and here there is the python script iv2dae.py (requires python collada package); modern software OpenInventor should be able to open them
- .vst: Used by ViSTa commercial software; widely documented in this PDF, also stored in this repository for reference
- Example folder (also contains greyscale textures in uncompressed SGI RGB format )
- C source code for SGI RGB format conversion: readrgb.c, readrgb.h
- kaitai-struct .ksy file for images in SGI RGB format (only greyscale 1024x1024)
- kaitai-generated javascript library to read SGI RGB format
- A .vst file contains multiple "Levels Of Depth" (LODs), i.e. multiple "3d files", each one with different resolution.
- The higher the index of the LOD in the file, the higher is its resolution (= number of triangles)
- Each LOD can contain 1 ore more "patches".
- Each patch can contain a number of "arrays"
- An array can be interpreted as a "group of faces" (triangles strip) or a "point cloud", depending on a patch flag at offset 0x04 in patch header (0 = face, 1 = point cloud)
In case the array represents a triangle strip, its values must be read stright or in reverse depending on position of triplet in the array:
All of this means that each .VST file can be converted to a number of 3d files (one per each LOD) with different resolutions.
Archive of .vst files for MER Spirit: https://pds-imaging.jpl.nasa.gov/data/mer/spirit/mer2mw_0xxx/data/
There were 4 cameras onboard:
- 1 Pancam
- 1 Navcam
- 2 Hazcams (front and rear)
Products/images are grouped by "site", i.e. reference points fixed along rover course; at each site 1 ore more "observations" may be associated.
ply magic string
format ascii 1.0 ascii/binary, format version number
comment made by anonymous a comment starts by "comment" keyword
element vertex VVVV number of vertex in the file
property float32 x ok
property float32 y ok
property float32 z ok
element face FFFF number of faces in the file
property list uint8 int32 vertex_index each face is a list of int32 pointers to vertex; how many pointers is
given by an uint8 value before the values
end_header delimits the end of the header
0 0 0 start of vertex list
0 0 1
0 1 1
0 1 0
1 0 0
1 0 1
1 1 1
1 1 0
3 0 1 2 start of face list; in VST format all faces are triangles.
3 7 6 5
3 0 4 5
3 1 5 6
3 2 6 7
3 3 7 4
- Navcam - mer2mw_0xxx
- Hazcam - mer2mw_0xxx (as above)
- Pancam - mer2mw_0xxx (as above)
Query to list all available XYZ products for a specified site:
Split:
- https://pds-imaging.jpl.nasa.gov/solr/pds_archives/search?
- pds.INSTRUMENT_HOST_ID=MERA/MER2/Spirit
- pds.atlas_instrument_name=pancam
- identifier=*xyl*\ - product type for XYZ is XYL
- pds.ROVER_MOTION_COUNTER_SITE=137
- rows=0 - to get just results count
Once you get the count of results, use this query to get all of them in CSV format:
query (please use needed number of rows/result)
https://pds-imaging.jpl.nasa.gov/solr/pds_archives/search? identifier=*xyl* pds.FILTER_NAME=*l* pds.INSTRUMENT_HOST_ID=MERA/MER2/Spirit pds.atlas_instrument_name=pancam pds.ROVER_MOTION_COUNTER_SITE=137 wt=csv rows=1000 fl=FILE_NAME fl=FILTER_NAME fl=SEQUENCE_ID fl=INSTRUMENT_HOST_ID
- Navcam - mer2om_0xxx
- Hazcam - mer2om_0xxx (as above)
- Pancam - mer2om_0xxx (as above)
- Basic instructions: https://pds-imaging.jpl.nasa.gov/tools/atlas/api/
- Detailed instructions: https://pds.nasa.gov/services/pds4_pds_search_protocol.pdf (PDF)
- Third party detailed instructions (by apache.org): https://solr.apache.org/guide/8_11/the-standard-query-parser.html
To look for available images for texturing,you can use this query, which can be split in:
- https://pds-imaging.jpl.nasa.gov/solr/pds_archives/search? * - base address*
- mission=mars*exploration*rover - mission
- product_type="mer" - processed images
- pds.ROVER_INSTRUMENT_ELEVATION=[-90 to 5] - elevation of pancam (interval)
- pds.ROVER_MOTION_COUNTER_SITE=137 - Site 137 (final position)
- pds.FILTER_NAME=*L7* - Filters are L2, L3, L4, L5, l6, L7; combine L2 + L5 + L7 to create colore texture
- pds.eye=left - left camera
A more advanced query allows listing only the filenames corresponding to specified parameter, and allows combining multiple values for filter by "OR" operator:
Split:
- https://pds-imaging.jpl.nasa.gov/solr/pds_archives/search?
- pds.FILTER_NAME=*l2*
- pds.FILTER_NAME=*l7* - Repeating same parameter with different values results in "OR"
- spacecraft_name=spirit
- atlas_instrument_name=pancam
- mission=mars*exploration*rover
- pds.ROVER_MOTION_COUNTER_SITE=137
- wt=json - output format: default=json; available: xml, csv, ..?
- fl=FILE_NAME - fields in output
- rows=1000 - output size
All links refer to navcam images for sol 1869 and site 137, the last site before Spirit got trapped in the "sand trap" of Troy, where its mission ended.
- raw image products:
- https://pds-imaging.jpl.nasa.gov/data/mer/spirit/mer2no_0xxx/data/sol1869/edr/ (eff = Full Frame, eth = Thumbnails)
- JPG versions of IMG products:
- 3d products (VIL, in VST format; .HT (height map), .pfb (Silicon Graphics OpenGL Performer, or Iris Performer):
Explanation of generic 2N292280938...B100P0703L0M1 name:
- 2 = MER2
- N = NAVCAM
- 292280938 = Sol 1869
- ... = Product type (XYZ, VIL, FFL,...)
- B1 = Site 137
- 00 = Position 0
- P = Navcam/Pancam observation
- 0703 = Observation n.0703
- L = Left camera
- 0 = no filter
- M = Creator of the file
- 1 = version of the file
- Point cloud - XYL (XYZ linearized), .img extension, multispectral BSQ image
- Multi-resolution textured mesh - VIL (ViSTa format, .vst extension)
- Height Map(.ht) (label)
- 3d multimesh file - Silicon Graphics OpenGL Perfomer Fast Binary format (.pfb) (label) (Only SGI OpenGL Performer and SGI Perfly can open these files, but they are no more downloadable since Silicon Graphics disappeared from internet years ago. Ancient project OpenPFB attempts opening some of these files, but how to display them? I am working to my own PFB-converter, which currently can only extract from PFB files the list of depth-images and textures to pass to Blender NAVCAM importer or to my enhanced version of Blender importer, now able to import from navcam, pancam and hazcam: https://github.com/jumpjack/Blender-Navcam-Importer.)
- Silicon Graphics IRIS Open Inventor (.iv) (For internal use, not released to public, destroyed after processing) (Inventors tool for Windows, Coin3d, python iv2dae.py converter, ivconv C source )
Visual representation of XYZ product; not technically useful because it mixes X, Y and Z data, but useful for a visual overview:
Visible image:
