CS594 Computer Graphics

Lab 3

Objectives

• To expand your software with new functions with well thought out designs and high quality implementation.
• To understand the concepts of multi-pass rendering and the functionalities of mutli-texturing units.
• To implement the shadow map algorithm as the regular part of the lab and bump map algorithm as extra credit (if you choose to).
• Be able to use A-buffer for anti-aliasing.

Assignment Description and Provided Resources
Three new sub-directory under class directory:

• ~huangj/CS594F03/OpenGLext/:  you can find some tutorials that we have not covered in class about openGL extension programmings. You should not have to read those to do lab3. But in case you get confused about anything, this directory could serve as an additional source of information.
• ~huangj/CS594F03/shadowmap: the class slides and a demo package, as well as the source code, are provided. Please refer to the README file under that directory for details.
• ~huangj/CS594F03/bumpmap: the class slides, a demo package, as well as the source code, nVidia tutorial on real bump mapping, a copy of the NeHe tutorial on emboss bump mapping, a set of color and bump textures are provided. Please read the README file under that directory for details.

The basic steps you will go through in this lab include:

• shadow map, implement the two pass algorithm

· first render the scene from the view of the camera, but only use ambient light

· second pass, render with shadow turned on

· render the shadow buffer from the view of the light source

· render from the view of camera, for each fragment

· look up the shadow map, and make the comparison of the shadow test with multi-texturing units.

· only render the fragments that are not in shadow at fully-lit colors

• anti-aliasing with A-buffer
• render a single quad as a floor  extending from the feet of the viewer towards the far end of the viewing frustum, and then texture map one of the two images: ~huangj/CS594F03/ ringstar*.jpg onto the quad. without doing anti-aliasing, you will see much visual artifacts.
• try use MIPmap to reduce aliasing.
• try render 2x2, 4x4 and 8x8 supersampling with A-buffer serving as the averaging functional unit, and compare the results with those obtained with MIPmapping.

Extra credit

• bump map
• first, you need to load a bump texture into your code as an alpha channel  texture
• rendering step
• on per vertex basis, compute the light source vector in the corresponding normal space, and obtain Fd, offset_s and offset_t. Feed these parameters, as well as the texture coordinates, (s,t),  into the OpenGL pipeline. Use the multi-texturing units to compute the difference that gives out the m value (as covered in class).
• shade and texture each fragment with color texture (if you are using a color texture) as well.

Try to arrange a scene with the models that you have access to. (We will talk about scene graph in the later half of the semester. But with not too many object, it is doable without using scene graphs.)

· create a scene where a Buddha or a sculpture is standing in front of a mirror in a room with texture mapped, emboss bump mapped floor, with shadows.  30%

· anything else you would like to do. Please run it by me before you do it, so that I have an idea how much extra credit I would be giving out

Detailed Lab Setup Information
Please consult TA's page.

Suggested Strategy
The assignment description is meant as a brief guidance of what steps you need to go through. It is highly desirable for you to have a big picture and seriously spend a couple hours designing your software, and think about the trade off that you would face, BEFORE you actually go implement anything.  You should not only consider performance, readability, maintainability but also possibilities for further development of new functions as key issues in your design.  This would end up saving you a lot of time later on, because of less number of patches you have in your code and a much more efficient debugging process in result.

During your implementation, an incremental way of development is always helpful with debugging too. As usual, it is strongly suggested that you start early on this lab. There is a lot of stuff involved in this lab.

Finally, you are strongly discouraged to 'cut and paste' from the demo source code into your own code. You should thoroughly understand how those advanced OpenGL features are used and program your own software accordingly. The source code similarity utlity is up running in our department. Our course is using that now. Yes, discussion is encouraged, but you must do your own work.

This lab is expected to be done using the Crux lab machines only.

Happy rendering!