http:^^www.cs.cornell.edu^info^people^amith^project.html

来自「This data set contains WWW-pages collect」· HTML 代码 · 共 93 行

MIME-Version: 1.0
Server: CERN/3.0
Date: Sunday, 01-Dec-96 19:46:18 GMT
Content-Type: text/html
Content-Length: 3971
Last-Modified: Saturday, 10-Aug-96 00:52:08 GMT

<title>Image and Video Transition FX for Rivl</title>
<center><h2>Image and Video Transition FX for Rivl</h2>
<b><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><!WA0><a href="http://www.cs.cornell.edu/Info/People/joseluis/joseluis.html">Jose-Luis Fernandez</a>, <!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><!WA1><a href="http://orac.ee.cornell.edu/unit1/valerie/index.html">Valerie J. Ohm</a> and <!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><!WA2><a href="http://www.cs.cornell.edu/Info/People/amith/yam.html">Amith Yamasani</a></b>
<p>16 October 1995
</center>
<p><hr>

<p><h2>1. Overview</h2>

<!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><!WA3><A HREF="http://www.cs.cornell.edu/Info/Projects/zeno/rivl/rivl.html">Rivl</A> is a language extension to Tcl/Tk for manipulating images and video
sequences. It comes with a set of primitives such as fade, rotate, overlay,
etc. The purpose of this project is to extend the set of primitives and use
them for creating complex transitional effects between two images or video
sequences. 


<p><h2>2. Deliverables</h2>


<ol>
<li>A set of primitives to create transitional effects between two images or on one image
by creating a sequence of intermediate images. Effects would include Flip, Deform, 
Squish, Peel, Fold-up, Melt, Bounce-squish, Morph(?).
<p>
<li>The same transitional effects on video sequences. 
</ol>

<p><h2>3. Milestones</h2>

<ul>
<li>Study the implementation of Rivl and understand the internal representation of 
     images and video sequences. 
<li>Work on some of the simpler transitional effects such as Flip, Melt and Peel.
    Get these effects working on a single image and on transition between two images. 
<li>The algorithms will be modified to work on video sequences.
<li>Implementation of the other effects like Fold-up, Squish and Deform to work on both
    images and video sequences.
<li>Test the functions extensively and optimize them for maximum speed.
<li>If time permits - Create a simple interface to accept points on two images and use 
    the points as references for the morphing of one image to the other. 
<li>Documentation - parts of which could be done concurrently with the coding and testing.
</ul>

<p><h2>4. Technical Overview</h2>

<H3>The effects</H3>
<p>

<B>Flip</B>: This is basically a shearing operation with a fixed line on either of the two axes.
The parameters would be the position and orientation of the fulcrum, the image, and the angle
through which to flip the image. In the case of flipping video, the frames would be processed
one by one with the angle increasing gradually.
<p>
<B>Melt</B>: The image seems to melt into a liquid and flow down out of sight. This would involve
generating a random smooth wave and shearing the image with the wave. The wave would also 
have to change in shape gradually so that it doesn't look like a curtain falling.
<p>
<B>Fold-up</B>: This effect simulates the folding up of a piece of paper over and over again. How 
the image is to be folded can be determined by parameters to the function. The image can be 
folded as an opaque piece of paper or as a transparent one.
<p>
<B>Peel</B>: This is a real cool effect - like peeling off a sheet of paper placed over another one.
Even a video sequence can be "peeled" off to expose another sequence. The direction of peeling
can be specified as a parameter. This effect could be extended to rolling it away like a scroll. 
<p>
<B>Squish</B>: Squishing an image would cause it to get compressed with a bulge in the middle. Bounce-squish is a variant of this where the squishing is a little springy. This involves shearing the image unevenly, applying a sine function to the shear and gradually increasing the amplitude of the sine
wave.

<p>
<B>Deform</B>: Deforming an image would involve squeezing a rectangular image into a nonrectangular 
area which could be a quadrilateral or an oval. First the image would be resized in both dimensions
to fit the bounding rectangle of the destination shape and then it would be sheared to fit the actual
shape. 
<p>
In most of these effects, as one video sequence gets transformed, the next sequence fills the
background. 

<br>