CS4102 图形算法

CS4102 – Computer Graphics – Academic Year 2020/21, Semester 2, Practical 2
General information: For your benefit please take note of the following information:
• Assignment title: 3D Rendering.
• Submission deadline: 9pm (2100) on the 23rd April 2021.
• Assignment weight: 60% of the coursework.
• Lateness penalty: Scheme B, 1 mark per 8-hour period, or part thereof.
• Required submission content: a single file report (in PDF format), a standalone application, and
the full source code.
• Mark descriptors: https://info.cs.st-andrews.ac...
feedback.html
• Good academic practice: https://www.st-andrews.ac.uk/...
Aims: The aim of this practical is to help you understand the key principles behind various techniques
frequently used for the rendering of 3D objects, and give you hands-on experience with their implementation
and manipulation.
Basic specification (up to 16 marks): Your task involves the creation of an application which
facilitates interactive modelling of faces in 3D.
All input data you need can be found zipped in the archive CS4102 2021 P2 data.zip. The archive
contains files named sh xxx.csv and tx xxx.csv (where ‘xxx’ stands for a zero-padded integer between
0 and 199), sh EV.csv, tx EV.csv, and mesh.csv. Each line in mesh.csv corresponds to a triangle in a
mesh used to represent a face; specifically, each line contains three integer indices. These index the 3D
coordinates and colours of face mesh vertices in respectively sh xxx.csv and tx xxx.csv.
Your application should start by drawing a triangle in the main window. Each of the corners of this
triangle correspond to a rendered face. The n-th (of 3) of these will have the 3D coordinates of its
vertices computed by adding the coordinates in sh 000.csv (the average face shape) summed with the
coordinate offsets in sh 00n.csv multiplied by the n-th weight in sh EV.csv. The corresponding colours
are similarly computed by adding the colours in tx 000.csv (the average face colour) summed with the
colour offsets in tx 00n.csv multiplied by the n-th weight in tx EV.csv.
The interactive design performed by the user is simple: a click within the screen area communicates
how similar (close) the synthetic face should be from the three reference faces. The synthetic face is
generated by interpolating 3D shape and colour between the three reference faces, and is displayed on
the side (or in a different window if more convenient) using Painter’s algorithm. You can use flat shading
and orthographic projection, and assume that faces are perfectly matte with a unity diffuse coefficient
and that there is a single directional light source aligned with the viewing direction. You are free to use
any library you wish for basic mathematical operations but you must implement all relevant computer
graphics techniques from scratch.
1
Advanced specification (up to 18 marks): For additional marks, there are several extensions which
you may wish to consider (you are also free to come up with your own). For example, you may extend
your application to allow the user to interpolate between possibly all 199 possible reference faces by
replacing one or more of the initial reference faces. Note that the contribution of any replaced reference
faces should not be lost but rather kept unchanged while other contributions are added. Consider allowing
the user to preview the possible reference faces before making the choice.
Highly advanced specification (up to 20 marks): Assuming that you have successfully met the
advanced specification requirements, you may wish to add further interactive elements such as the ability
to rotate faces, change the direction or the number of light sources, the shading model, etc.
Hints: The following suggestions should help you prevent common mistakes, and save time and effort:
• Make sure that your submission is complete i.e. that the application can be executed on different
machines without the need for tinkering by the marker. It is not reasonable to expect the marker
to debug and fix your code (e.g. hard-coded paths).
• In your report, focus on the quality of content. Do not be overly verbose – well formed, succinct
explanations are easier to read and more convincing than convoluted and excessively long verbiage.
Aim for up to two pages of text but feel free to include images or screenshots to complement this
content and illustrate your work better.
• If you are interested in learning more about the model use here, you may find it useful to have a look
at the following paper: https://gravis.dmi.unibas.ch/...
Ognjen Arandjelovi´c (Oggie)