GSoC 2013 Project Proposal: Web Interface of BRL-CAD
Personal Information
Name: Harmanpreet Singh
Email Address: harman052@gmail.com
IRC Username: harmanpreet
Brief Background Information:
Final year B.Tech student of Computer Science and Engineering at Guru Nanak Dev
Engineering College, Ludhiana, India.
Google Code-In 2012 mentor with BRL-CAD.
Phone number: +91 998 877 1252
Brief Summary
This project aims at making a web interface of BRL-CAD, means shifting all features and functionality of desktop software to browser. Although, the scope of this project is much wider, but for current GSoC project, we are implementing it as a basic concept, enabling the user to access BRL-CAD from anywhere through the web, perform basic operations and finally obtain the output in desired way (.g, .raytraced image) itself on the browser. This way, we are extending BRL-CAD’s capability to mobile users hence helping to reach more users, without any need for download and installation. Hence, this web interface of BRL-CAD is adding another dimension in the improvement for its ease of use.
Why this project?
BRL-CAD is a desktop application. Everything from desktop is ultimately shifting to the web. Cloud computing is example. It is very popular and has high demand among business houses and normal user nowadays. Main reasons behind this are ease of access and readily available services for 24x7. With the advancement of web and web based applications, it has become a preferred way to do things. For example, Google's almost every service is web based with ease of collaboration and sharing with no overheads involved on the part of user.
There are numerous excellent examples that prove the power of today's web and craze among people doing such things. Here are some mind blowing examples: http://threejs.org/. By looking at current trend, this craze will increase in coming future and browser will dominate over desktop.
Moving in the same direction I am proposing the project of web interface of BRL-CAD that will deliver a more reliable, user friendly experience with very basic, yet important feature of platform independent being readily accessible for 24 hours.
Detailed Summary
Overview
This project has two ends: Front end and Back end. User will make model on the browser (front end) and his inputs will be written as BRL-CAD commands in a shell script. When model completes, on submission, the shell script will be executed and processed by BRL-CAD at back end and return desired file (.g, .pix, .png) to the user back on browser. This is the basic flow of the project. There are certain challenges involved in both these ends.
Detailed Approach
Various tools and technologies to be used to accomplish this project while resolving the following challenges are being discussed further in the proposal.
Challenges involved in front end are: Providing option on browser to make primitives. Positioning. Applying boolean operations on primitives to make complex models.
Above points are the basic tasks needed to be carried out by user while working on BRL-CAD.
Challenges involved in back end are: Processing user inputs. Returning desired file (end product) to user.
Front End Approach
Let's discuss the front end part. For providing options on browser to make primitives, this should be graphical and easy to work with. However, it would be excellent if we make it as drag 'n' drop feature and enabling the user to see such primitives as 3D where he can rotate and move it, but for starting point we can make it as click 'n' click feature having four standard views (top, left, right, bottom (and may be isometric also)). These four different views will be captured in four different display windows.
Wireframe Replacement Here user will not see any wireframe as in BRL-CAD's graphics window. When user will come to web page, he/she will see options to make primitive and four display windows that will be empty at that time. When user click on any option to make primitive, he will see 2D projections of four different views of primitive or object in four display windows.
In other words, this projection will be just an area filled with some default colour say grey, representing the object. Thus, when user creates a primitive he / she will see 2D views of 3D object.
On clicking any primitive option, the user will be asked to enter input values specific for that primitive. This input can be taken in variety of ways, such as, on clicking, a dialog box will ask for values, or just input boxes appear. The later seems to be more reliable and user friendly.
For this GSoC project, we will consider only one or two basic primitives. Other primitives can be added in the same way as these basic ones would be added.
Resizing: In case, the user need to resize the object / primitive, it can be done either by again asking for new values or a slider option can be provided. Slider option will be more easier to use. When slider moves, changes in each view will be visible in the display windows simultaneously. When user satisfies with size, he click on SAVE button to save changes.
All explained above can be implemented using webGL or HTML5 canvas. I am sure it is capable to do much more than this. To be more specific we can use threejs.org and scenejs.org libraries to accomplish above tasks.
Performing Boolean Operations: Second challenge is to perform boolean operations on primitives to make complex models. For this we need two primitives and need to consider their respective position. For better user experience we will make it graphical i.e. user will be able to place objects at any given position with mouse within the display windows. The four standard views will help the user to place objects accurately. Again this can be accomplish using power of open source javascript libraries like kineticjs, threejs, scenejs etc.
Operations like 'Resize', 'Reposition', and 'Boolean Operations' will have pair of 'EDIT' and 'SAVE' buttons. For example, if a user need to resize a primitive, first he selects desired primitive, then clicks on 'EDIT' button, input fields specific to that primitive will appear to accept new values or a slider appears to adjust size. When size finally adjusted, user click on SAVE button to save changes.
Similar is the case of Repositioning and boolean operations.
To implement boolean operations, a function corresponding to each operation (union(), subtraction(), intersection()) will be written in programming. First let's discuss the boolean operations in BRL-CAD.
Union (u): Joins the shapes. Subtract (-): Remove volume of one shape from another. Intersect (+): Keeps only overlapping parts of objects.
Union can combine any number of primitives, so our corresponding union function will accept names of any number of primitives. One of the ways to implement this is using variadic functions that accept number of total arguments and arguments itself.
Subtraction and intersection are performed only on two primitives hence their corresponding functions will accept names of two selected primitives as arguments.
Now the question "How user will perform boolean operations?" is answered in following steps:
1. Select an operation from list of boolean operations. Selecting the operation will enable the primitive selection mode.
2. Since the user entered the primitive selection mode, type of operation selected determines how many objects the user can select. If selected operation is Union, then he / she can select as many objects as possible and if intersection or subtraction is selected, only two objects will be allowed to be selected. In case of subtraction, order of primitive selection will have importance because primitive selected at second number will be completely removed from assembly.
3. Selection of primitives in step 2 will extract their names and will be passed as arguments to corresponding function. This function will use these arguments to write corresponding command in file.
Back End Approach
BRL-CAD will be used as backend CAD engine that forms the foundation of whole project. When user will be satisfied with the model, he click on one of the options for example 'DOWNLOAD DATABASE FILE', 'DOWNLOAD RAW IMAGE FILE', 'DOWNLOAD PNG FILE' to obtain results in desired form i.e. database file (.g), raw image file format (.pix) or Portable Network Graphics format (.png) respectively.
Remember, all inputs in the form of MGED commands are being written in the shell script that forms the connection between frontend and backend (also showed in figure under 'Overview' section). Recall the SGI cube script: http://brlcad.org/wiki/SGI_Cube that take advantage of scripting of MGED. Same is the case of our script.
When user choose any of the option mentioned above (further commands for generating images will be added if option chosen for pix or png) will trigger the script and generate model in BRL-CAD.
Hence, except MGED GUI interface, we are using complete infrastructure of BRL-CAD through scripting.
After model generation, the desired file will be available on browser for download.
Tools intended to be used:
Following are the links to open source javaScript 3D libraries that can be used to implement front end of the project. http://threejs.org/ http://scenejs.org/ http://kineticjs.com/ http://paperjs.org/
Milestones:
17 June - 24 June: 8 days Explore threejs and scenejs and other tools. Try with examples and making small applications in them.
25 June - 4 July: 10 days Implement options to make primitives such as sph and rpp, and display windows.
5 July - 11 July: 1 week Option of resizing the primitives / objects.
12 July - 19 July: 8 days Positioning of objects.
20 July - 24 July: 5 days Add option of union, subtraction and intersection.
24 July - 25 July: 2 days Back up days, for any backlogs or any pending tasks.
26 July - 29 July: 4 days Final go through, and preparation for mid term evaluation.
3 August – 7 August: 5 days Option of downloading the output files.
8 August - 17 August: 15 days Fixing bugs, testing.
18 August - Start documentation (using LaTeX or any other tool recommended) including developers documentation, user’s documentation.
Future scope of project: As mentioned in the beginning, this project has much wider scope and will definitely expand in the near future there are various ideas that can be implemented in the future, so I decided to list them down here:
3D model display and editing. Connectivity with material database if there is any possibility. Different user accounts. Model Sharing.
Time Availability:
I will be available 40 hours / week, if needed can spend more. No restriction of time.
Why BRL-CAD?
BRL-CAD is software of my interest. I started interacting on BRL-CAD developers’ mailing list from last year, where I found many helpful friends and learned a lot, now I want to contribute my part. I want to be involved in the development work during and after GSoC.
Why me?
I am intensely excited to work this year in GSoC with BRL-CAD. Excited to interact and work with BRL-CAD developers on such real world project. I have been mentor in Google Code-In 2012 and enjoyed a lot while contributing and mentoring to GCI students and now I wish to continue working with this community through GSoC this year.