Problem Set 3
CMSC 427
Assigned October 23, 2007, Due November 6, 2007
Introduction: This problem set will focus on writing a program to morph one image into
another. You will implement the approach described in: “Feature-Based Image
Metamorphosis,” by T. Beier and S. Neely, Computer Graphics, 26(2) 25-42 (available
on-line at: http://www.cs.princeton.edu/courses/archive/fall00/cs426/papers/beier92.pdf).
You will be given some basic code, as well as an executable for finished code. Some
things to notice about this code:
• BMPImage.cpp and .h define an image class for bmp type images. This includes
functions to read in an image and to save an image. Note there is no function to
copy images, so if you want to call a function using an image as an argument, you
should pass an address, so that it doesn’t need to be copied.
• texture.cpp and .h contain functions that load an image from a file and turn them
into a texture.
• main.cpp contains code that reads in two images, and provides an interactive tool
that allows you to specify corresponding line segments in the two images. These
will be needed to create correspondences for morphing. To use this interface, try
the following operations:
o Click left on an image to start a line segment. Drag the mouse to specify
the end of the segment.
o ‘u’ undoes the last line segment on the left image. ‘v’ does this for the
right image.
You will modify this code to use these line segments to morph between the images. In
addition to these features, the executable contains a couple of other features:
o ‘m’ causes the left image to morph into the right, using the line segments
that you’ve specified in each image. Note that the order and direction
used to create the line segments matters. The first line in the left image
matches the first one on the right, and the first points of each segment
match.
o ‘f’ causes the left image to fade into the right.
o ‘s’ causes all the images in the morph to be saved in a directory called
‘Results’ (you should create this directory ahead of time.
o ‘r’ replays the morphing using the images saved in ‘Results’. This is
useful if you are morphing larger images, which my program will not
morph in real time. Replaying gives you a better sense of the morph.
1. Fade (15 points): Implement a fade operation. This is begun by typing ‘f’. The
image on the left of the screen should gradually fade out, being replaced by the
image on the right. Let’s call the image on the left I, and the image on the right J.
If the fade lasts n units of time, then after k units of time pixel (x,y) will have
intensities given by (1-k/n)*I(x,y) + (k/n)*J(x,y). You might try, for example,
generating a fade that lasts 4-6 seconds, at 30 frames a second.
