1. Shape: A 3D Modeling Tool for Astrophysics. Wolfgang Steffen, Nicholas Koning, Stephan Wenger, Christophe Morisset, Marcus Magnor. IEEE Transactions on Visualization and Computer Graphics, Vol. 6, No. 1. January 2007. ​arXiv Link. Submitted to arXiv 10-03-2010.

Outline:

1. Introduction

• Claims that the development of "effective methods for the reconstruction of the 3D structure of astrophysical objects is […] an issue of growing importance in astronomy."
  • Astrophysical photographs only provide a 2D integration (some sort of column density map, if you will) of the emission and absorption along the line of sight.
    • No information with regard to the depth of the object photographed.
  • Some objects have symmetry properties that allow for a more "automatic reconstruction," i.e. Planetary nebulae (PPN, PN).
  • No symmetries? Can determine the depth from information contained in the velocity field.
    • Some astrophysical objects cannot allow for such a mapping, i.e. Interstellar clouds.
• What about the current methods used in computational astrophysics? Astrophysical modeling utilizes parallelized codes on supercomputers (like AstoBEAR).
  • Authors claim: "While such simulations produce insight into generic astrophysical processes, they are rarely suitable for elucidating the properties and structure of particular objects." What properties and what objects?
    • Planetary nebulae in particular. (Why? Symmetry reasons? Hard to apply dynamically?)
    • "For the correct physical interpretation of observational data, information about the object's 3D shape has to be available. Obtaining new structural information and insight on particular objects is the main purpose of the application of [Shape]."
• Morpho-kinematical: As applied to modeling involves only structural (morphological) and velocity (kinematic) information.
  • In contrast to dynamical simulations: Include effects of forces and temporal evolution from a set of simpler initial and boundary conditions.
    • "…, the outcome of dynamical simulations is not predictable in detail and very hard to tune to a specific object."

Ultimately authors claim Shape will remedy the shortcomings of the current state of astrophysical reconstructions by applying structure modeling techniques of commercial animation softwares, while adding systems that are necessary for astrophysical research. They implement more physically accurate modeling of radiation transfer from the sources to the observer.

2. Related Work

2.1 Automatic-Reconstruction Methods

• Utilization of Doppler shift methods make use of the correlation between velocity and position of emitting gas relative to a local point of reference. Allows derivation of "depth information" from doppler shift data.
• "Many objects contain several different kinematic subsystems which may have different relations between velocity and position."
• As mentioned before PNs have an inherent spherical or axial symmetry due to their evolution (from symmetrical sources). Can use a symmetry assumption to reconstruct the missing spatial dimension. However a caveat is that these symmetries can be disturbed by statistical effects or external influences (turbulence!).

2.2 User-Driven Systems

• These shortcomings mentioned above can be avoided by having an interactive modeling technique. So Shape is "user-driven."

3. Observational Data in Astronomy

• Authors claim that observational data in astronomy comes from two essential sources: photographs, and spectrographs.
  • Advent of HST —> High resolution imagery available for a large number of objects.
  • Doppler-shift measurements —> 3D reconstruction. High resolution spectrographs measure the shift of any observed spectral line w.r.t its known reference wavelength. Therefore the relative velocity of emitter can be measured.
    • Complex astrophysical objects may have distortions of their spectral lines.
• With Shape, a narrow slit is used to select only a narrow, but long region of the object to capture spatially varying wavelength information. Yields a PV-diagram.

4. The Shape System

As author's summarize:

• Tools to define a spatial emissivity and velocity field.
• Establish a mapping between Doppler-shift and position.
• Produce output that can be compared with observed images and spectra.
• Shape: The tools to define the spatial structure and velocity field should be interactive.

Ultimately the workflow involves interactive modeling, and automatic optimization.

4.1 Java Implementation

A few compatibility issues with Mac OS X. "The necessary Java software has become available only recently [2007] and only for 64-bit systems." This section doesn't really apply and note the time the paper was submitted. Slightly out of date. When downloading from their website, they will instruct you on any particularities with Mac OS X.

4.2 Interactive Modeling

• Two main components: 3D modeling view where we define geometry and behavior of the object, and 2D screen where simulated appearance of model can be compared to observation.
• Implementation of "meshes."
  • Sphere, torus, cone, cube for mesh creation process that can be deformed via modifiers. These meshes can be used as volumetric objects, thin shells with a specified thickness.
  • Scaling, translation and rotation. Also have squeeze, squish, shear and twist modifiers.
• Density, color and velocity can be defined as a function of position.
• Modeling/visualization can be done in arbitrary or in various actual physical units.
• Particle velocities can be specified in a model —> time evolution can be predicted, assuming ballistic expansion (i.e. constant velocity/momentum).
• Can import data from external simulations in order to visualize and analyze the results (see first example). Models created Shape can also be exported and used as input for other software.

4.3 Image Rendering

• Three renderers: particle, grid, and mesh. There is a physical renderer as well. This renderer takes into account opacity and other radiation transport effects. During the publication of this article, the physical renderer was still being developed.
  • Particle: Random particle distribution to sample model emissivity and velocity space.
  • Grid: Particle positions used to sample physical properties of the object. Density of particles distributed in the 3D grid.
  • Mesh: Sampling of the physical properties by searching the voxels of the grid that overlap with the model's mesh.

4.4 Synthetic Observation

• In order to make the reconstruction of observational data reliable, Shape takes into account the model, and also the properties of the measurement devices.
  • Color of the substructure can be used as assigned to the 3D model, the velocity along the line of sight as given by the Doppler-shift can be color coded (red vs. blue).
  • Implementation of channel maps: where only a given velocity (or wavelength) range is projected onto the xy-plane.
  • Light echoes (paraboloids)

4.5 Automatic Optimization

4.6 Automatic Reconstruction

4.7 Plot, Animation, and Movie Modules

• Plotting data, animating model parameters, and the display of animation results. Making 1D spectral line profiles.

5. Results and Example Applications

May want to access the article in order to read about these if they are of interest. I primarily care about section 5.1.

5.1 Validation of Shape with Hydrodynamical Simulations

• Using Shape to characterize the velocity field of numerical simulations of some basic types of planetary nebulae.
  • The simulated hydro data was filtered according to density.
  • Imported as a particle system, including the velocity information.
  • A 3D mesh and density distribution was then fitted to a large scale object.

5.2 Reconstruction f the Saturn Nebula

5.3 Nova RS Ophiuchi

5.4 Content production for digital media

6. Future Developments

• "Long-term plans are guided by potential applications that the system has. Such plans include the incorportation of interactive hydrodynamical simulations making use of multi-processor graphics processing units."

7. Conclusions

Download Shape here: ​Shape - Downloads