Difference between revisions of "ESA Summer of Code in Space/Project Ideas"

From BRL-CAD
(initial stab at ideas for SOCIS, starting with the GSoC list)
 
(eliminate some ideas only loosely relevant to space work)
Line 62: Line 62:
 
|-
 
|-
 
|'''[[GED Transactions]]:''' Migrating functionality from our MGED geometry editor into our LIBGED library provided excellent code reuse, but now we need transactions so that actions can be undone.
 
|'''[[GED Transactions]]:''' Migrating functionality from our MGED geometry editor into our LIBGED library provided excellent code reuse, but now we need transactions so that actions can be undone.
|C
 
|HUGE
 
|EASY
 
|-
 
|'''[[Add exec option to search]]:'''  Our LIBGED library provides a ''search'' command very similar to the UNIX ''find'' command for scanning through geometry.  Implement the -exec option.
 
|C
 
|HUGE
 
|EASY
 
|-
 
|'''[[Geometry Selection Functionality]]:'''  Implement a new selection command that lets you temporarily group objects based on spatial, attribute, or geometric properties.
 
 
|C
 
|C
 
|HUGE
 
|HUGE
Line 80: Line 70:
 
|HUGE
 
|HUGE
 
|HARD
 
|HARD
|-
 
|'''[[Space Partitioning for Tessellation]]:'''  Technically an optimization task, make our geometry converters run an order of magnitude faster by using spatial partitioning during tessellation.
 
|C
 
|BIG
 
|HARD
 
|-
 
|'''[[libsvn within Geometry Database Format]]:'''  Our ''Geometry Service'' project provides comprehensive edit history by using a revision control system.  Hook our .g database on the backend ftw.
 
|C
 
|HUGE
 
|HARD
 
|-
 
|'''[[Overlap tool]]:''' Resolving geometric interferences (aka overlaps) is a common geometry editing activity.  Design an awesome interface for resolving conflicts.
 
|Tcl
 
|HUGE
 
|EASY
 
 
|}
 
|}
  
Line 143: Line 118:
 
|C
 
|C
 
|BIG
 
|BIG
|EASY
 
|-
 
|'''[[MGED to Archer Command Migration]]:'''  Our next-generation editor (Archer) needs to have the same (700+) commands that are in MGED.  Help make our next CAD interface awesome.
 
|C and Tcl
 
|HUGE
 
|EASY
 
|-
 
|'''[[General Tree Walker]]:'''  We have a half dozen or more functions that will traverse a geometry hierarchy in different ways.  There only needs to be one.
 
|C
 
|HUGE
 
|EASY
 
|-
 
|'''[[Rework of libbu/libbn to not require Tcl]]:'''  Tcl's '''''C API''''' is used throughout BRL-CAD providing great abstractions, but we want our basic utility and numeric libraries (LIBBU & LIBBN) to stand alone.
 
|C
 
|HUGE
 
 
|EASY
 
|EASY
 
|-
 
|-
Line 191: Line 151:
 
|HUGE
 
|HUGE
 
|EASY
 
|EASY
|-
 
|'''[[2D Sketch Editor]]:'''  MGED has a sucky 2D sketch editor written in pure Tcl/Tk.  Archer has nothing.  Fix the old one or write a new one.
 
|Tcl (presently) and/or C
 
|HUGE
 
|HARD
 
|-
 
|'''[[Ayam Editor Feature Integration]]:'''  The Ayam project has support for NURBS editing, we do not.  Collaboration and reuse is a good thing.
 
|C/Tcl
 
|BIG
 
|MEDIUM
 
 
|-
 
|-
 
|'''[[GUI Integration of Analysis Tools]]:'''  There are a ''lot'' of visualization tools in BRL-CAD, but most aren't integrated with the GUI.  Visualizing directly within the GUI would improve usability.
 
|'''[[GUI Integration of Analysis Tools]]:'''  There are a ''lot'' of visualization tools in BRL-CAD, but most aren't integrated with the GUI.  Visualizing directly within the GUI would improve usability.
Line 209: Line 159:
 
|'''[[Level of Detail Wireframes]]:''' Geometry is drawn with fixed wireframe detail which can screw performance and make shape recognition difficult.  Detail needs to adjust based on view size.
 
|'''[[Level of Detail Wireframes]]:''' Geometry is drawn with fixed wireframe detail which can screw performance and make shape recognition difficult.  Detail needs to adjust based on view size.
 
|C
 
|C
|HUGE
 
|EASY
 
|-
 
|'''[[BoT Editing]]:'''  Our unstructured triangle mesh geometry (Bag of Triangles = BoT) is a common import format, but our BoT edit support sucks.  We need something better.
 
|C
 
|HUGE
 
|HARD
 
|-
 
|'''[[NMG Editing]]:'''  Our structured polygonal mesh geometry (N-Manifold Geometry = NMG) is a common export format, but our NMG edit support is non-existent.  We need something.
 
|C
 
|HUGE
 
|HARD
 
|-
 
|'''[[Visualizing Constructive Solid Geometry (CSG)]]:'''  CSG forms a directed acyclic graph.  Tree view widgets show a simple hierarchical representation, but we need something better.
 
|C/C++
 
 
|HUGE
 
|HUGE
 
|EASY
 
|EASY
Line 238: Line 173:
 
!Impact
 
!Impact
 
!Difficulty
 
!Difficulty
|-
 
|'''[[Shader Enhancements]]:'''  We have a functioning custom shader system in BRL-CAD, but there are now folks that specialize and there's lot of room for improvements.
 
|C
 
|BIG
 
|EASY
 
 
|-
 
|-
 
|'''[[Material and Shader Objects]]:'''  This one is a biggie but easy.  Implement new object entities for describing shaders and material properties, use them during ray tracing.
 
|'''[[Material and Shader Objects]]:'''  This one is a biggie but easy.  Implement new object entities for describing shaders and material properties, use them during ray tracing.
Line 248: Line 178:
 
|HUGE
 
|HUGE
 
|EASY
 
|EASY
|-
 
|'''[[NMG Raytracing Performance Improvement]]:'''  BRL-CAD's structured mesh geometry (N-Manifold Geometry = NMG) needs some time in a profiler.  Malloc during raytracing is bad.
 
|C
 
|BIG
 
|HARD
 
|-
 
|'''[[Generalized abstracted spacial partitioning capability]]:''' Need more be said?  If you're familiar with BSPs, KD trees, and grid structures, then this one is for you.
 
|C
 
|HUGE
 
|HARD
 
 
|-
 
|-
 
|'''[[High Dynamic Range Support]]:'''  We don't have displays supporting this yet, but that's never stopped us before.  Implement support for images with more than 8-bits per channel.
 
|'''[[High Dynamic Range Support]]:'''  We don't have displays supporting this yet, but that's never stopped us before.  Implement support for images with more than 8-bits per channel.
Line 286: Line 206:
 
!Difficulty
 
!Difficulty
 
|-
 
|-
|'''[[Synchronize Wiki with Docbook]]:'''  We use Docbook for most of our user documentation but find editing a wiki page much easier to use.  Set up a system so the two are always in sync.
+
|'''[[Materials Database]]:'''  Create a Materials Database web site for collecting, managing, and providing programmatic interfaces to material properties.
 
|Depends
 
|Depends
|HUGE
+
|BIG
 
|MEDIUM
 
|MEDIUM
 
|-
 
|-
 
|'''[[Benchmark Performance Database]]:'''  BRL-CAD's Benchmark has been used for two decades to compare performance  across configurations.  Build a database website.  
 
|'''[[Benchmark Performance Database]]:'''  BRL-CAD's Benchmark has been used for two decades to compare performance  across configurations.  Build a database website.  
|Depends
 
|BIG
 
|MEDIUM
 
|-
 
|'''[[Materials Database]]:'''  Create a Materials Database web site for collecting, managing, and providing programmatic interfaces to material properties.
 
 
|Depends
 
|Depends
 
|BIG
 
|BIG
 
|MEDIUM
 
|MEDIUM
 
|}
 
|}
 
== Other Tool Projects ==
 
 
Some of these project ideas aren't as detailed or elaborate as the ones above, but would be interesting and useful nonetheless.  Please discuss with us before proposing one of these ideas.
 
 
* [[Automated exploded view tool]]  (C/C++/Tcl)
 
* [[Automated cutaway view tool]]  (C/C++/Tcl)
 
* [[Converter completion so all current formats have both import and export]]  (C)
 
* [[GUI Animation editor/creator]]  (C/Tcl)
 
* [[Blender file format converter]]  (C)
 
  
  

Revision as of 18:57, 1 July 2011

The list of possible projects below should serve as a good starting point for new developers that would like to get involved in working on BRL-CAD. The ideas below range from the very hard and math intense to the very easy, see the link to each for more details. You are also welcome to scale the scope of the project up or down as needed as the ideas suggested below are merely starting points.

BRL-CAD Priorities

Be sure to read up on our application process for getting started with your proposal submission if you have not done so already including our checklist. See the diagram at the bottom of this page for an overview of our major project focus areas.

Project Categories

Geometry Conversion Projects

When it comes to geometry conversion, STEP and IGES are the priority. STEP is an absurdly complicated format, but we're one of the first open source projects to implement a working importer. IGES is a much older standard, but more time-tested too. Both have a variety of potential high-priority projects associated with the.

Idea Languages Impact Difficulty
STEP Libraries: We took over the NIST STEP Class Library (SCL) project during our development of our g-step exporter. It needs a lot of cleanup. C/C++ HUGE EASY
STEP exporter: We have an importer, we need a comprehensive exporter with support for implicit CSG, NURBS, or polygonal mesh geometry. C/C++ HUGE EASY
STEP importer improvements: We have an importer, but it's preliminary. Add support for importing hierarchy information, polygonal geometry, and implicit geometry. C/C++ HUGE MEDIUM
IGES import improvements: We have extensive support for the International Geometry Exchange Standard (IGES) with our g-iges and iges-g converters, but they need updating. C HUGE MEDIUM
Geometry Conversion Library: Probably our biggest open source asset is our extensive collection of importers and exporters. Turn them all into a universal conversion library. C/C++ HUGE EASY
Voxelize command: Convert geometry into voxel data sets by shooting a grid of rays. The finite element analysis and volumetric rendering folks will love you. C BIG EASY


Geometry Processing Projects

As part of our new GUI development efforts, a lot of code had to be moved out of application space into a library. That library is our geometry editing library, LIBGED. A lot of time, attention, and emphasis is going on there to make sure that all of our core functionality is reusable by applications. More than 100k lines of code have already moved, but there's still a lot more work needed.

Idea Languages Impact Difficulty
GED Transactions: Migrating functionality from our MGED geometry editor into our LIBGED library provided excellent code reuse, but now we need transactions so that actions can be undone. C HUGE EASY
Geometric Constraint Solver: Our LIBPC parametric constraint library is the work-in-progress foundation of being able to validate and describe geometry relationships. C/C++ HUGE HARD


NURBS (Non-Uniform Rational B-Splines) Projects

BRL-CAD has recently implemented raytracing support for NURBS-based boundary representation geometry. It's one of our hot-topic priority focus areas with high impact and high visibility. They are also pretty hard so be careful to have sufficient familiarity and lots of milestones.

Idea Languages Impact Difficulty
NURBS Intersections: Evaluate the intersection of two NURBS surfaces. You'll get a 3D curve or a point, line segment, or even a plane depending how the surfaces overlap. C/C++ HUGE HARD
NURBS Tessellation: Given a NURBS object (i.e., a collection of NURBS surfaces that enclose space), evaluate a polygonal mesh. It has to be fast. C/C++ HUGE MEDIUM
Implicit to NURBS conversion: We implement lots of primitives like sphere and torus defined with implicit parameters. Given one of our implicit primitives, describe it using NURBS surfaces. C/C++ HUGE MEDIUM
Plate Mode NURBS raytracing: Imported NURBS geometry often does NOT enclose space (i.e., not solid), so add support for specifying an implicit thickness. C/C++ HUGE HARD


Code Refactoring Projects

BRL-CAD has more than a million lines of code so we have to continually refactor and clean up code for maintainability. Cruft happens. These projects help reduce the complexity and make things better. They're not very hard at all, but exceedingly valuable to our project vitality.

Idea Languages Impact Difficulty
Code Reduction: BRL-CAD is more than 1M lines of code. Approximately 10% of that is duplication. Help reduce that percentage. C BIG EASY
Consolidate image processing: We have 100+ image processing tools that independently read and write file data. Needs much reuse love. C HUGE EASY
Fix Bugs: We take pride in our breadth of functionality and time-tested stability. We still have bugs. Identify some and fix 'em! C DEPENDS EASY


Graphical User Interface (GUI) Projects

The primary environment in which users interact with BRL-CAD geometry is the MGED geometry editor. The "next generation" interface for MGED, dubbed Archer, has been under development for some time but there is a lot of functionality that still needs to be migrated and new features needed by MGED, Archer, or both.

Idea Languages Impact Difficulty
New Cross-Platform 3D Display Manager: BRL-CAD uses display managers for visualizing 3D geometry in a window. We want one that uses a cross-platform toolkit such as Qt or OGRE. C/C++ HUGE MEDIUM
New Cross-Platform 2D Framebuffer: BRL-CAD uses framebuffers to display 2D imagery. The merits of having a single interface for most platforms is self-evident. C/C++ HUGE EASY
GUI Integration of Analysis Tools: There are a lot of visualization tools in BRL-CAD, but most aren't integrated with the GUI. Visualizing directly within the GUI would improve usability. C BIG MEDIUM
Level of Detail Wireframes: Geometry is drawn with fixed wireframe detail which can screw performance and make shape recognition difficult. Detail needs to adjust based on view size. C HUGE EASY


Rendering & Analysis Projects

Raytracing is at the heart of all that is BRL-CAD. The 'rt' application was the world's first distributed parallel ray tracer, implemented in early 80's, and it's backed by our powerful LIBRT library. We're the best as full path shotlines (shooting all the way through geometry, not just first hit) and can handle crazy big geometry very efficiently, very fast. It's how we perform geometric analysis, generate images, perform simulations, and more.

Idea Languages Impact Difficulty
Material and Shader Objects: This one is a biggie but easy. Implement new object entities for describing shaders and material properties, use them during ray tracing. C HUGE EASY
High Dynamic Range Support: We don't have displays supporting this yet, but that's never stopped us before. Implement support for images with more than 8-bits per channel. C and/or C++ BIG EASY
Vector output from raytracing: Huge impact here if you can update one or more of our raytracers to provide vector output instead of raster. Perhaps render directly to PDF or SVG. C and/or C++ HUGE EASY
Analysis Library: There are various tools in BRL-CAD for calculating weights, moments of inertia, and more. They're stand-alone applications. Turn them into a library. C HUGE HARD


Web Development Projects

We're generally less interested in web development projects because the developer rarely sticks around after development is over. We're not interested in web code, we're interested in people. If you're willing to work on your project long after this summer is over (i.e., joining the project) and are willing to make a long-term commitment to BRL-CAD, then we welcome your web development project idea.

Idea Languages Impact Difficulty
Materials Database: Create a Materials Database web site for collecting, managing, and providing programmatic interfaces to material properties. Depends BIG MEDIUM
Benchmark Performance Database: BRL-CAD's Benchmark has been used for two decades to compare performance across configurations. Build a database website. Depends BIG MEDIUM


<AN IDEA OF YOUR OWN>

Do you have an idea of your own? Let us know and maybe we'll like it too. We're very open to new ideas, areas of academic research, industry applications, and any other ways that may help get you hooked on BRL-CAD development. Just remember that BRL-CAD is a solid modeling CAD suite so keep that in mind when scoping your project.

Requirements:

  • Passion for the task being suggested
  • Buy-in from one of the existing developers


Mentors

BRL-CAD operates under group mentorship. That means you can contact anyone, not just the person assigned to you, for assistance. The mailing list and IRC channel are the preferred communication methods.

  • Christopher Sean Morrison
    • brlcad on irc.freenode.net
    • Org admin, core dev, open source project lead
  • Erik Greenwald
    • ``Erik on irc.freenode.net
    • Org admin, core dev
  • Cliff Yapp
    • starseeker on irc.freenode.net
    • Mentor, core dev
  • Daniel Rossberg
    • d_rossberg on irc.freenode.net
    • Mentor, core dev, math expert
  • Dave Loman
    • dloman on irc.freenode.net
    • Mentor, expert modeler
  • Ed Davisson
    • brlcad-devel mailing list
    • Part-time mentor, Math guru extraordinaire
  • Keith Bowman
    • brlcad-devel mailing list
    • Part-time mentor, core dev, NURBS guru
  • Tom Browder
    • brlcad-devel maling list
    • Part-time mentor, analysis, code support
  • Richard Weiss
    • brlcad-devel mailing list
    • Part-time mentor, NMG info, code support